int main(int argc, char *argv[])
{
int argi;
int showeval = 1;
int cleanexpr = 0;
setup_signalhandler(argv[0]);
for (argi = 1; (argi < argc); argi++) {
if ((strcmp(argv[argi], "--help") == 0)) {
printf("%s version %s\n\n", argv[0], VERSION);
printf("Usage:\n%s [--quiet] [--clean] [--debug] [--no-update]\n", argv[0]);
exit(0);
}
else if ((strcmp(argv[argi], "--version") == 0)) {
printf("%s version %s\n", argv[0], VERSION);
exit(0);
}
else if ((strcmp(argv[argi], "--debug") == 0)) {
debug = 1;
}
else if ((strcmp(argv[argi], "--no-update") == 0)) {
dontsendmessages = 1;
}
else if ((strcmp(argv[argi], "--quiet") == 0)) {
showeval = 0;
}
else if ((strcmp(argv[argi], "--clean") == 0)) {
cleanexpr = 1;
}
else if ((strncmp(argv[argi], "--error-colors=", 15) == 0)) {
char *tok;
int newerrorcolors = 0;
tok = strtok(strchr(argv[argi], '=')+1, ",");
while (tok) {
int col = parse_color(tok);
if ((col >= 0) && (col <= COL_RED)) newerrorcolors |= (1 << parse_color(tok));
tok = strtok(NULL, ",");
}
if (newerrorcolors) errorcolors = newerrorcolors;
}
}
return update_combotests(showeval, cleanexpr);
}
int main(int argc, char *argv[])
{
int daemonize = 1;
int listenq = 10;
char *pidfile = "msgcache.pid";
int lsocket;
struct sockaddr_in laddr;
struct sigaction sa;
int opt;
/* Dont save the output from errprintf() */
save_errbuf = 0;
memset(&laddr, 0, sizeof(laddr));
inet_aton("0.0.0.0", (struct in_addr *) &laddr.sin_addr.s_addr);
laddr.sin_port = htons(1984);
laddr.sin_family = AF_INET;
for (opt=1; (opt < argc); opt++) {
if (argnmatch(argv[opt], "--listen=")) {
char *locaddr, *p;
int locport;
locaddr = strchr(argv[opt], '=')+1;
p = strchr(locaddr, ':');
if (p) { locport = atoi(p+1); *p = '\0'; } else locport = 1984;
memset(&laddr, 0, sizeof(laddr));
laddr.sin_port = htons(locport);
laddr.sin_family = AF_INET;
if (inet_aton(locaddr, (struct in_addr *) &laddr.sin_addr.s_addr) == 0) {
errprintf("Invalid listen address %s\n", locaddr);
return 1;
}
}
else if (argnmatch(argv[opt], "--server=")) {
/* Who is allowed to fetch cached messages */
char *p = strchr(argv[opt], '=');
serverlist = getsenderlist(p+1);
}
else if (argnmatch(argv[opt], "--max-age=")) {
char *p = strchr(argv[opt], '=');
maxage = atoi(p+1);
}
else if (argnmatch(argv[opt], "--lqueue=")) {
char *p = strchr(argv[opt], '=');
listenq = atoi(p+1);
}
else if (strcmp(argv[opt], "--daemon") == 0) {
daemonize = 1;
}
else if (strcmp(argv[opt], "--no-daemon") == 0) {
daemonize = 0;
}
else if (argnmatch(argv[opt], "--pidfile=")) {
char *p = strchr(argv[opt], '=');
pidfile = strdup(p+1);
}
else if (argnmatch(argv[opt], "--logfile=")) {
char *p = strchr(argv[opt], '=');
logfile = strdup(p+1);
}
else if (strcmp(argv[opt], "--debug") == 0) {
debug = 1;
}
else if (strcmp(argv[opt], "--version") == 0) {
printf("xymonproxy version %s\n", VERSION);
return 0;
}
}
/* Set up a socket to listen for new connections */
lsocket = socket(AF_INET, SOCK_STREAM, 0);
if (lsocket == -1) {
errprintf("Cannot create listen socket (%s)\n", strerror(errno));
return 1;
}
opt = 1;
setsockopt(lsocket, SOL_SOCKET, SO_REUSEADDR, &opt, sizeof(opt));
fcntl(lsocket, F_SETFL, O_NONBLOCK);
if (bind(lsocket, (struct sockaddr *)&laddr, sizeof(laddr)) == -1) {
errprintf("Cannot bind to listen socket (%s)\n", strerror(errno));
return 1;
}
if (listen(lsocket, listenq) == -1) {
errprintf("Cannot listen (%s)\n", strerror(errno));
return 1;
}
/* Redirect logging to the logfile, if requested */
if (logfile) {
reopen_file(logfile, "a", stdout);
reopen_file(logfile, "a", stderr);
}
errprintf("Xymon msgcache version %s starting\n", VERSION);
errprintf("Listening on %s:%d\n", inet_ntoa(laddr.sin_addr), ntohs(laddr.sin_port));
if (daemonize) {
pid_t childpid;
reopen_file("/dev/null", "r", stdin);
/* Become a daemon */
childpid = fork();
if (childpid < 0) {
/* Fork failed */
errprintf("Could not fork\n");
exit(1);
}
else if (childpid > 0) {
/* Parent - save PID and exit */
FILE *fd = fopen(pidfile, "w");
if (fd) {
fprintf(fd, "%d\n", (int)childpid);
fclose(fd);
}
exit(0);
}
/* Child (daemon) continues here */
setsid();
}
setup_signalhandler("msgcache");
memset(&sa, 0, sizeof(sa));
sa.sa_handler = sigmisc_handler;
sigaction(SIGHUP, &sa, NULL);
sigaction(SIGTERM, &sa, NULL);
do {
fd_set fdread, fdwrite;
int maxfd;
int n;
conn_t *cwalk, *cprev;
msgqueue_t *qwalk, *qprev;
time_t mintstamp;
/* Remove any finished connections */
cwalk = chead; cprev = NULL;
while (cwalk) {
conn_t *zombie;
if (cwalk->action != C_DONE) {
cprev = cwalk;
cwalk = cwalk->next;
continue;
}
/* Close the socket */
close(cwalk->sockfd);
zombie = cwalk;
if (cprev == NULL) {
chead = zombie->next;
cwalk = chead;
cprev = NULL;
}
else {
cprev->next = zombie->next;
cwalk = zombie->next;
}
freestrbuffer(zombie->msgbuf);
xfree(zombie);
}
ctail = chead;
if (ctail) { while (ctail->next) ctail = ctail->next; }
/* Remove expired messages */
qwalk = qhead; qprev = NULL;
mintstamp = getcurrenttime(NULL) - maxage;
while (qwalk) {
msgqueue_t *zombie;
if (qwalk->tstamp > mintstamp) {
/* Hasn't expired yet */
qprev = qwalk;
qwalk = qwalk->next;
continue;
}
zombie = qwalk;
if (qprev == NULL) {
qhead = zombie->next;
qwalk = qhead;
qprev = NULL;
}
else {
qprev->next = zombie->next;
qwalk = zombie->next;
}
freestrbuffer(zombie->msgbuf);
xfree(zombie);
}
qtail = qhead;
if (qtail) { while (qtail->next) qtail = qtail->next; }
/* Now we're ready to handle some data */
FD_ZERO(&fdread);
FD_ZERO(&fdwrite);
/* Add the listen socket */
FD_SET(lsocket, &fdread);
maxfd = lsocket;
for (cwalk = chead; (cwalk); cwalk = cwalk->next) {
switch (cwalk->action) {
case C_READING:
FD_SET(cwalk->sockfd, &fdread);
if (cwalk->sockfd > maxfd) maxfd = cwalk->sockfd;
break;
case C_WRITING:
FD_SET(cwalk->sockfd, &fdwrite);
if (cwalk->sockfd > maxfd) maxfd = cwalk->sockfd;
break;
case C_DONE:
break;
}
}
n = select(maxfd+1, &fdread, &fdwrite, NULL, NULL);
if (n < 0) {
if (errno == EINTR) continue;
errprintf("select failed: %s\n", strerror(errno));
return 0;
}
if (n == 0) continue; /* Timeout */
for (cwalk = chead; (cwalk); cwalk = cwalk->next) {
switch (cwalk->action) {
case C_READING:
if (FD_ISSET(cwalk->sockfd, &fdread)) grabdata(cwalk);
break;
case C_WRITING:
if (FD_ISSET(cwalk->sockfd, &fdwrite)) senddata(cwalk);
break;
case C_DONE:
break;
}
}
if (FD_ISSET(lsocket, &fdread)) {
/* New incoming connection */
conn_t *newconn;
int caddrsize;
dbgprintf("New connection\n");
newconn = calloc(1, sizeof(conn_t));
caddrsize = sizeof(newconn->caddr);
newconn->sockfd = accept(lsocket, (struct sockaddr *)&newconn->caddr, &caddrsize);
if (newconn->sockfd == -1) {
/* accept() failure. Yes, it does happen! */
dbgprintf("accept failure, ignoring connection (%s)\n", strerror(errno));
xfree(newconn);
newconn = NULL;
}
else {
fcntl(newconn->sockfd, F_SETFL, O_NONBLOCK);
newconn->action = C_READING;
newconn->msgbuf = newstrbuffer(0);
newconn->tstamp = getcurrenttime(NULL);
}
if (newconn) {
if (ctail) { ctail->next = newconn; ctail = newconn; }
else chead = ctail = newconn;
}
}
} while (keeprunning);
if (pidfile) unlink(pidfile);
return 0;
}
int main(int argc, char *argv[])
{
int daemonize = 0;
int lsocket;
struct sockaddr_in laddr;
struct sigaction sa;
int argi, opt;
libxymon_init(argv[0]);
/* Dont save the output from errprintf() */
save_errbuf = 0;
memset(&laddr, 0, sizeof(laddr));
laddr.sin_addr.s_addr = htonl(INADDR_ANY);
laddr.sin_port = htons(1984);
laddr.sin_family = AF_INET;
for (argi=1; (argi < argc); argi++) {
if (argnmatch(argv[argi], "--listen=")) {
char *locaddr, *p;
int locport;
locaddr = strchr(argv[argi], '=')+1;
p = strchr(locaddr, ':');
if (p) { locport = atoi(p+1); *p = '\0'; } else locport = 1984;
memset(&laddr, 0, sizeof(laddr));
laddr.sin_port = htons(locport);
laddr.sin_family = AF_INET;
if (inet_aton(locaddr, (struct in_addr *) &laddr.sin_addr.s_addr) == 0) {
errprintf("Invalid listen address %s\n", locaddr);
return 1;
}
}
else if (strcmp(argv[argi], "--daemon") == 0) {
daemonize = 1;
}
else if (strcmp(argv[argi], "--no-daemon") == 0) {
daemonize = 0;
}
else if (standardoption(argv[argi])) {
if (showhelp) return 0;
}
}
/* Set up a socket to listen for new connections */
lsocket = socket(AF_INET, SOCK_DGRAM, IPPROTO_UDP);
if (lsocket == -1) {
errprintf("Cannot create listen socket (%s)\n", strerror(errno));
return 1;
}
opt = 1;
setsockopt(lsocket, SOL_SOCKET, SO_REUSEADDR, &opt, sizeof(opt));
if (bind(lsocket, (struct sockaddr *)&laddr, sizeof(laddr)) == -1) {
errprintf("Cannot bind to listener address (%s)\n", strerror(errno));
return 1;
}
/* Make it non-blocking */
fcntl(lsocket, F_SETFL, O_NONBLOCK);
/* Redirect logging to the logfn, if requested */
if (logfn) {
reopen_file(logfn, "a", stdout);
reopen_file(logfn, "a", stderr);
}
errprintf("Xymon locator version %s starting\n", VERSION);
errprintf("Listening on %s:%d\n", inet_ntoa(laddr.sin_addr), ntohs(laddr.sin_port));
if (daemonize) {
pid_t childpid;
reopen_file("/dev/null", "r", stdin);
/* Become a daemon */
childpid = fork();
if (childpid < 0) {
/* Fork failed */
errprintf("Could not fork\n");
exit(1);
}
else if (childpid > 0) {
/* Parent - exit */
exit(0);
}
/* Child (daemon) continues here */
setsid();
}
setup_signalhandler("xymond_locator");
memset(&sa, 0, sizeof(sa));
sa.sa_handler = sigmisc_handler;
sigaction(SIGHUP, &sa, NULL);
sigaction(SIGTERM, &sa, NULL);
tree_init();
load_state();
do {
ssize_t n;
struct sockaddr_in remaddr;
socklen_t remaddrsz;
char buf[32768];
fd_set fdread;
/* Wait for a message */
FD_ZERO(&fdread);
FD_SET(lsocket, &fdread);
n = select(lsocket+1, &fdread, NULL, NULL, NULL);
if (n == -1) {
/* Select error */
errprintf("select error, aborting: %s\n", strerror(errno));
keeprunning = 0;
continue;
}
/* We know there is some data */
remaddrsz = sizeof(remaddr);
n = recvfrom(lsocket, buf, sizeof(buf), 0, (struct sockaddr *)&remaddr, &remaddrsz);
if (n == -1) {
/* We may get EAGAIN if there is not a full message yet */
if (errno != EAGAIN) {
errprintf("Recv error: %s\n", strerror(errno));
}
continue;
}
else if (n == 0) {
continue;
}
buf[n] = '\0';
dbgprintf("Got message from %s:%d : '%s'\n",
inet_ntoa(remaddr.sin_addr), ntohs(remaddr.sin_port), buf);
handle_request(buf);
n = sendto(lsocket, buf, strlen(buf)+1, 0, (struct sockaddr *)&remaddr, remaddrsz);
if (n == -1) {
if (errno == EAGAIN) {
errprintf("Out-queue full to %s, dropping response\n", inet_ntoa(remaddr.sin_addr));
}
else {
errprintf("Send failure %s while sending to %s\n",
strerror(errno), inet_ntoa(remaddr.sin_addr));
}
}
} while (keeprunning);
save_state();
return 0;
}
int main(int argc, char *argv[])
{
char *msg;
int argi;
struct sigaction sa;
char *exthandler = NULL;
char *extids = NULL;
char *processor = NULL;
struct sockaddr_un ctlsockaddr;
int ctlsocket;
/* Handle program options. */
for (argi = 1; (argi < argc); argi++) {
if (strcmp(argv[argi], "--debug") == 0) {
debug = 1;
}
else if (argnmatch(argv[argi], "--rrddir=")) {
char *p = strchr(argv[argi], '=');
rrddir = strdup(p+1);
}
else if (argnmatch(argv[argi], "--extra-script=")) {
char *p = strchr(argv[argi], '=');
exthandler = strdup(p+1);
}
else if (argnmatch(argv[argi], "--extra-tests=")) {
char *p = strchr(argv[argi], '=');
extids = strdup(p+1);
}
else if (argnmatch(argv[argi], "--processor=")) {
char *p = strchr(argv[argi], '=');
processor = strdup(p+1);
}
else if (strcmp(argv[argi], "--no-cache") == 0) {
use_rrd_cache = 0;
}
else if (net_worker_option(argv[argi])) {
/* Handled in the subroutine */
}
}
save_errbuf = 0;
if ((rrddir == NULL) && xgetenv("XYMONRRDS")) {
rrddir = strdup(xgetenv("XYMONRRDS"));
}
if (exthandler && extids) setup_exthandler(exthandler, extids);
/* Do the network stuff if needed */
net_worker_run(ST_RRD, LOC_STICKY, update_locator_hostdata);
setup_signalhandler("xymond_rrd");
memset(&sa, 0, sizeof(sa));
sa.sa_handler = sig_handler;
sigaction(SIGHUP, &sa, NULL);
sigaction(SIGCHLD, &sa, NULL);
sigaction(SIGTERM, &sa, NULL);
sigaction(SIGINT, &sa, NULL);
signal(SIGPIPE, SIG_DFL);
/* Setup the control socket that receives cache-flush commands */
memset(&ctlsockaddr, 0, sizeof(ctlsockaddr));
sprintf(ctlsockaddr.sun_path, "%s/rrdctl.%d", xgetenv("XYMONTMP"), getpid());
unlink(ctlsockaddr.sun_path); /* In case it was accidentally left behind */
ctlsockaddr.sun_family = AF_UNIX;
ctlsocket = socket(AF_UNIX, SOCK_DGRAM, 0);
if (ctlsocket == -1) {
errprintf("Cannot create cache-control socket (%s)\n", strerror(errno));
return 1;
}
fcntl(ctlsocket, F_SETFL, O_NONBLOCK);
if (bind(ctlsocket, (struct sockaddr *)&ctlsockaddr, sizeof(ctlsockaddr)) == -1) {
errprintf("Cannot bind to cache-control socket (%s)\n", strerror(errno));
return 1;
}
/* Linux obeys filesystem permissions on the socket file, so make it world-accessible */
if (chmod(ctlsockaddr.sun_path, S_IRUSR|S_IWUSR|S_IRGRP|S_IWGRP|S_IROTH|S_IWOTH) == -1) {
errprintf("Setting permissions on cache-control socket failed: %s\n", strerror(errno));
}
/* Load the RRD definitions */
load_rrddefs();
/* If we are passing data to an external processor, create the pipe to it */
setup_extprocessor(processor);
running = 1;
while (running) {
char *eoln, *restofmsg = NULL;
char *metadata[MAX_META+1];
int metacount;
char *p;
char *hostname = NULL, *testname = NULL, *sender = NULL, *classname = NULL, *pagepaths = NULL;
xymonrrd_t *ldef = NULL;
time_t tstamp;
int childstat;
ssize_t n;
char ctlbuf[PATH_MAX];
int gotcachectlmessage;
time_t now;
/* See if we have any cache-control messages pending */
do {
n = recv(ctlsocket, ctlbuf, sizeof(ctlbuf), 0);
gotcachectlmessage = (n > 0);
if (gotcachectlmessage) {
/* We have a control message */
char *bol, *eol;
ctlbuf[n] = '\0';
bol = ctlbuf;
do {
eol = strchr(bol, '\n'); if (eol) *eol = '\0';
rrdcacheflushhost(bol);
if (eol) { bol = eol+1; } else bol = NULL;
} while (bol && *bol);
}
} while (gotcachectlmessage);
/* Get next message */
msg = get_xymond_message(C_LAST, argv[0], &seq, NULL);
if (msg == NULL) {
running = 0;
continue;
}
now = gettimer();
if (reloadtime < now) {
/* Reload configuration files */
load_hostnames(xgetenv("HOSTSCFG"), NULL, get_fqdn());
load_client_config(NULL);
reloadtime = now + 600;
}
/* Split the message in the first line (with meta-data), and the rest */
eoln = strchr(msg, '\n');
if (eoln) {
*eoln = '\0';
restofmsg = eoln+1;
}
/* Parse the meta-data */
metacount = 0;
memset(&metadata, 0, sizeof(metadata));
p = gettok(msg, "|");
while (p && (metacount < MAX_META)) {
metadata[metacount++] = p;
p = gettok(NULL, "|");
}
metadata[metacount] = NULL;
if ((metacount >= 14) && (strncmp(metadata[0], "@@status", 8) == 0)) {
/*
* @@status|timestamp|sender|origin|hostname|testname|expiretime|color|testflags|\
* prevcolor|changetime|ackexpiretime|ackmessage|disableexpiretime|disablemessage|\
* clienttstamp|flapping|classname|pagepaths
*/
int color = parse_color(metadata[7]);
switch (color) {
case COL_GREEN:
case COL_YELLOW:
case COL_RED:
case COL_BLUE: /* Blue is OK, because it only arrives here when an update is sent */
tstamp = atoi(metadata[1]);
sender = metadata[2];
hostname = metadata[4];
testname = metadata[5];
classname = (metadata[17] ? metadata[17] : "");
pagepaths = (metadata[18] ? metadata[18] : "");
ldef = find_xymon_rrd(testname, metadata[8]);
update_rrd(hostname, testname, restofmsg, tstamp, sender, ldef, classname, pagepaths);
break;
default:
/* Ignore reports with purple, blue or clear - they have no data we want. */
break;
}
}
else if ((metacount > 5) && (strncmp(metadata[0], "@@data", 6) == 0)) {
/* @@data|timestamp|sender|origin|hostname|testname|classname|pagepaths */
tstamp = atoi(metadata[1]);
sender = metadata[2];
hostname = metadata[4];
testname = metadata[5];
classname = (metadata[6] ? metadata[6] : "");
pagepaths = (metadata[7] ? metadata[7] : "");
ldef = find_xymon_rrd(testname, "");
update_rrd(hostname, testname, restofmsg, tstamp, sender, ldef, classname, pagepaths);
}
else if (strncmp(metadata[0], "@@shutdown", 10) == 0) {
running = 0;
continue;
}
else if (strncmp(metadata[0], "@@idle", 6) == 0) {
/* Ignored */
continue;
}
else if (strncmp(metadata[0], "@@logrotate", 11) == 0) {
char *fn = xgetenv("XYMONCHANNEL_LOGFILENAME");
if (fn && strlen(fn)) {
freopen(fn, "a", stdout);
freopen(fn, "a", stderr);
}
continue;
}
else if (strncmp(metadata[0], "@@reload", 8) == 0) {
reloadtime = 0;
}
else if ((metacount > 3) && (strncmp(metadata[0], "@@drophost", 10) == 0)) {
char hostdir[PATH_MAX];
hostname = metadata[3];
MEMDEFINE(hostdir);
sprintf(hostdir, "%s/%s", rrddir, hostname);
dropdirectory(hostdir, 1);
MEMUNDEFINE(hostdir);
}
else if ((metacount > 4) && (strncmp(metadata[0], "@@droptest", 10) == 0)) {
/*
* Not implemented. Mappings of testnames -> rrd files is
* too complex, so on the rare occasion that a single test
* is deleted, they will have to delete the rrd files themselves.
*/
}
else if ((metacount > 4) && (strncmp(metadata[0], "@@renamehost", 12) == 0)) {
char oldhostdir[PATH_MAX];
char newhostdir[PATH_MAX];
char *newhostname;
MEMDEFINE(oldhostdir);
MEMDEFINE(newhostdir);
hostname = metadata[3];
newhostname = metadata[4];
sprintf(oldhostdir, "%s/%s", rrddir, hostname);
sprintf(newhostdir, "%s/%s", rrddir, newhostname);
rename(oldhostdir, newhostdir);
if (net_worker_locatorbased()) locator_rename_host(hostname, newhostname, ST_RRD);
MEMUNDEFINE(newhostdir);
MEMUNDEFINE(oldhostdir);
}
else if ((metacount > 5) && (strncmp(metadata[0], "@@renametest", 12) == 0)) {
/* Not implemented. See "droptest". */
}
/*
* We fork a subprocess when processing drophost requests.
* Pickup any finished child processes to avoid zombies
*/
while (wait3(&childstat, WNOHANG, NULL) > 0) ;
}
/* Flush all cached updates to disk */
errprintf("Shutting down, flushing cached updates to disk\n");
rrdcacheflushall();
errprintf("Cache flush completed\n");
/* Close the external processor */
shutdown_extprocessor();
/* Close the control socket */
close(ctlsocket);
unlink(ctlsockaddr.sun_path);
return 0;
}
int main(int argc, char *argv[])
{
char *pagedir;
bbgen_page_t *p;
dispsummary_t *s;
int i;
char *pageset = NULL;
char *nssidebarfilename = NULL;
char *egocolumn = NULL;
char *csvfile = NULL;
char csvdelim = ',';
int embedded = 0;
int hobbitddump = 0;
char *envarea = NULL;
int do_normal_pages = 1;
/* Setup standard header+footer (might be modified by option pageset) */
select_headers_and_footers("bb");
bb_color = bb2_color = bbnk_color = -1;
pagedir = NULL;
init_timestamp();
fqdn = get_fqdn();
/* Setup values from env. vars that may be overridden via commandline options */
if (xgetenv("MKBB2COLREPEAT")) {
int i = atoi(xgetenv("MKBB2COLREPEAT"));
if (i > 0) maxrowsbeforeheading = i;
}
for (i = 1; (i < argc); i++) {
if ( (strcmp(argv[i], "--hobbitd") == 0) ||
(argnmatch(argv[i], "--purplelifetime=")) ||
(strcmp(argv[i], "--nopurple") == 0) ) {
/* Deprecated */
}
else if (argnmatch(argv[i], "--env=")) {
char *lp = strchr(argv[i], '=');
loadenv(lp+1, envarea);
}
else if (argnmatch(argv[i], "--area=")) {
char *lp = strchr(argv[i], '=');
envarea = strdup(lp+1);
}
else if (argnmatch(argv[i], "--hobbitddump")) {
hobbitddump = 1;
}
else if (argnmatch(argv[i], "--ignorecolumns=")) {
char *lp = strchr(argv[i], '=');
ignorecolumns = (char *) malloc(strlen(lp)+2);
sprintf(ignorecolumns, ",%s,", (lp+1));
}
else if (argnmatch(argv[i], "--nk-reds-only")) {
nkonlyreds = 1;
}
else if (argnmatch(argv[i], "--bb2-ignorecolumns=")) {
char *lp = strchr(argv[i], '=');
bb2ignorecolumns = (char *) malloc(strlen(lp)+2);
sprintf(bb2ignorecolumns, ",%s,", (lp+1));
}
else if (argnmatch(argv[i], "--bb2-colors=")) {
char *lp = strchr(argv[i], '=') + 1;
bb2colors = colorset(lp, (1 << COL_GREEN));
}
else if (argnmatch(argv[i], "--bb2-ignorepurples")) {
bb2colors = (bb2colors & ~(1 << COL_PURPLE));
}
else if (argnmatch(argv[i], "--bb2-ignoredialups")) {
bb2nodialups = 1;
}
else if (argnmatch(argv[i], "--includecolumns=")) {
char *lp = strchr(argv[i], '=');
includecolumns = (char *) malloc(strlen(lp)+2);
sprintf(includecolumns, ",%s,", (lp+1));
}
else if (argnmatch(argv[i], "--eventignore=")) {
char *lp = strchr(argv[i], '=');
eventignorecolumns = (char *) malloc(strlen(lp)+2);
sprintf(eventignorecolumns, ",%s,", (lp+1));
}
else if (argnmatch(argv[i], "--doccgi=")) {
char *lp = strchr(argv[i], '=');
char *url = (char *)malloc(strlen(xgetenv("CGIBINURL"))+strlen(lp+1)+2);
sprintf(url, "%s/%s", xgetenv("CGIBINURL"), lp+1);
setdocurl(url);
xfree(url);
}
else if (argnmatch(argv[i], "--docurl=")) {
char *lp = strchr(argv[i], '=');
setdocurl(lp+1);
}
else if (argnmatch(argv[i], "--no-doc-window")) {
/* This is a no-op now */
}
else if (argnmatch(argv[i], "--doc-window")) {
setdocurl("TARGET=\"_blank\"");
}
else if (argnmatch(argv[i], "--htmlextension=")) {
char *lp = strchr(argv[i], '=');
htmlextension = strdup(lp+1);
}
else if (argnmatch(argv[i], "--htaccess")) {
char *lp = strchr(argv[i], '=');
if (lp) htaccess = strdup(lp+1);
else htaccess = ".htaccess";
}
else if ((strcmp(argv[i], "--wml") == 0) || argnmatch(argv[i], "--wml=")) {
char *lp = strchr(argv[i], '=');
if (lp) {
wapcolumns = (char *) malloc(strlen(lp)+2);
sprintf(wapcolumns, ",%s,", (lp+1));
}
enable_wmlgen = 1;
}
else if (argnmatch(argv[i], "--nstab=")) {
char *lp = strchr(argv[i], '=');
if (strlen(lp+1) > 0) {
nssidebarfilename = strdup(lp+1);
}
else errprintf("--nstab requires a filename\n");
}
else if (argnmatch(argv[i], "--nslimit=")) {
char *lp = strchr(argv[i], '=');
nssidebarcolorlimit = parse_color(lp+1);
}
else if (argnmatch(argv[i], "--rssversion=")) {
char *lp = strchr(argv[i], '=');
rssversion = strdup(lp+1);
}
else if (argnmatch(argv[i], "--rsslimit=")) {
char *lp = strchr(argv[i], '=');
rsscolorlimit = parse_color(lp+1);
}
else if (argnmatch(argv[i], "--rss")) {
wantrss = 1;
}
else if (argnmatch(argv[i], "--rssextension=")) {
char *lp = strchr(argv[i], '=');
rssextension = strdup(lp+1);
}
else if (argnmatch(argv[i], "--reportopts=")) {
char style[MAX_LINE_LEN];
unsigned int rstart, rend;
int count = sscanf(argv[i], "--reportopts=%u:%u:%d:%s",
&rstart, &rend, &dynamicreport, style);
reportstart = rstart; reportend = rend;
if (count < 2) {
errprintf("Invalid --reportopts option: Must have start- and end-times\n");
return 1;
}
if (count < 3) dynamicreport = 1;
if (count == 4) {
if (strcmp(style, stylenames[STYLE_CRIT]) == 0) reportstyle = STYLE_CRIT;
else if (strcmp(style, stylenames[STYLE_NONGR]) == 0) reportstyle = STYLE_NONGR;
else reportstyle = STYLE_OTHER;
}
if (reportstart < 788918400) reportstart = 788918400;
if (reportend > time(NULL)) reportend = time(NULL);
if (xgetenv("BBREPWARN")) reportwarnlevel = atof(xgetenv("BBREPWARN"));
if (xgetenv("BBREPGREEN")) reportgreenlevel = atof(xgetenv("BBREPGREEN"));
if ((reportwarnlevel < 0.0) || (reportwarnlevel > 100.0)) reportwarnlevel = 97.0;
if ((reportgreenlevel < 0.0) || (reportgreenlevel > 100.0)) reportgreenlevel = 99.995;
select_headers_and_footers("bbrep");
sethostenv_report(reportstart, reportend, reportwarnlevel, reportgreenlevel);
}
else if (argnmatch(argv[i], "--csv=")) {
char *lp = strchr(argv[i], '=');
csvfile = strdup(lp+1);
}
else if (argnmatch(argv[i], "--csvdelim=")) {
char *lp = strchr(argv[i], '=');
csvdelim = *(lp+1);
}
else if (argnmatch(argv[i], "--snapshot=")) {
char *lp = strchr(argv[i], '=');
snapshot = atol(lp+1);
select_headers_and_footers("bbsnap");
sethostenv_snapshot(snapshot);
}
else if (strcmp(argv[i], "--pages-first") == 0) {
hostsbeforepages = 0;
}
else if (strcmp(argv[i], "--pages-last") == 0) {
hostsbeforepages = 1;
}
else if (argnmatch(argv[i], "--subpagecolumns=")) {
char *lp = strchr(argv[i], '=');
subpagecolumns = atoi(lp+1);
if (subpagecolumns < 1) subpagecolumns=1;
}
else if (argnmatch(argv[i], "--maxrows=")) {
char *lp = strchr(argv[i], '=');
maxrowsbeforeheading = atoi(lp+1);
if (maxrowsbeforeheading < 0) maxrowsbeforeheading=0;
}
else if (strcmp(argv[i], "--recentgifs") == 0) {
use_recentgifs = 1;
}
else if (argnmatch(argv[i], "--recentgifs=")) {
char *lp = strchr(argv[i], '=');
use_recentgifs = 1;
recentgif_limit = 60*durationvalue(lp+1);
}
else if (strcmp(argv[i], "--sort-group-only-items") == 0) {
sort_grouponly_items = 1;
}
else if (argnmatch(argv[i], "--page-title=")) {
char *lp = strchr(argv[i], '=');
defaultpagetitle = strdup(lp+1);
}
else if (argnmatch(argv[i], "--dialupskin=")) {
char *lp = strchr(argv[i], '=');
dialupskin = strdup(lp+1);
}
else if (argnmatch(argv[i], "--reverseskin=")) {
char *lp = strchr(argv[i], '=');
reverseskin = strdup(lp+1);
}
else if (strcmp(argv[i], "--pagetitle-links") == 0) {
pagetitlelinks = 1;
}
else if (strcmp(argv[i], "--pagetext-headings") == 0) {
pagetextheadings = 1;
}
else if (strcmp(argv[i], "--underline-headings") == 0) {
underlineheadings = 1;
}
else if (strcmp(argv[i], "--no-underline-headings") == 0) {
underlineheadings = 0;
}
else if (strcmp(argv[i], "--no-eventlog") == 0) {
bb2eventlog = 0;
}
else if (argnmatch(argv[i], "--max-eventcount=")) {
char *lp = strchr(argv[i], '=');
bb2eventlogmaxcount = atoi(lp+1);
}
else if (argnmatch(argv[i], "--max-eventtime=")) {
char *lp = strchr(argv[i], '=');
bb2eventlogmaxtime = atoi(lp+1);
}
else if (argnmatch(argv[i], "--max-ackcount=")) {
char *lp = strchr(argv[i], '=');
bb2acklogmaxcount = atoi(lp+1);
}
else if (argnmatch(argv[i], "--max-acktime=")) {
char *lp = strchr(argv[i], '=');
bb2acklogmaxtime = atoi(lp+1);
}
else if (strcmp(argv[i], "--no-acklog") == 0) {
bb2acklog = 0;
}
else if (strcmp(argv[i], "--no-pages") == 0) {
do_normal_pages = 0;
}
else if (argnmatch(argv[i], "--noprop=")) {
char *lp = strchr(argv[i], '=');
nopropyellowdefault = (char *) malloc(strlen(lp)+2);
sprintf(nopropyellowdefault, ",%s,", (lp+1));
errprintf("--noprop is deprecated - use --nopropyellow instead\n");
}
else if (argnmatch(argv[i], "--nopropyellow=")) {
char *lp = strchr(argv[i], '=');
nopropyellowdefault = (char *) malloc(strlen(lp)+2);
sprintf(nopropyellowdefault, ",%s,", (lp+1));
}
else if (argnmatch(argv[i], "--nopropred=")) {
char *lp = strchr(argv[i], '=');
nopropreddefault = (char *) malloc(strlen(lp)+2);
sprintf(nopropreddefault, ",%s,", (lp+1));
}
else if (argnmatch(argv[i], "--noproppurple=")) {
char *lp = strchr(argv[i], '=');
noproppurpledefault = (char *) malloc(strlen(lp)+2);
sprintf(noproppurpledefault, ",%s,", (lp+1));
}
else if (argnmatch(argv[i], "--nopropack=")) {
char *lp = strchr(argv[i], '=');
nopropackdefault = (char *) malloc(strlen(lp)+2);
sprintf(nopropackdefault, ",%s,", (lp+1));
}
else if (strcmp(argv[i], "--bbpageONLY") == 0) {
/* Deprecated */
errprintf("--bbpageONLY is deprecated - use --pageset=NAME to generate pagesets\n");
}
else if (strcmp(argv[i], "--embedded") == 0) {
embedded = 1;
}
else if (argnmatch(argv[i], "--pageset=")) {
char *lp = strchr(argv[i], '=');
pageset = strdup(lp+1);
}
else if (argnmatch(argv[i], "--template=")) {
char *lp = strchr(argv[i], '=');
lp++;
select_headers_and_footers(lp);
}
else if (argnmatch(argv[i], "--purplelog=")) {
char *lp = strchr(argv[i], '=');
if (*(lp+1) == '/') purplelogfn = strdup(lp+1);
else {
purplelogfn = (char *) malloc(strlen(xgetenv("BBHOME"))+1+strlen(lp+1)+1);
sprintf(purplelogfn, "%s/%s", xgetenv("BBHOME"), (lp+1));
}
}
else if (argnmatch(argv[i], "--report=") || (strcmp(argv[i], "--report") == 0)) {
char *lp = strchr(argv[i], '=');
if (lp) {
egocolumn = strdup(lp+1);
}
else egocolumn = "bbgen";
timing = 1;
}
else if (argnmatch(argv[i], "--nklog=") || (strcmp(argv[i], "--nklog") == 0)) {
char *lp = strchr(argv[i], '=');
if (lp) {
lognkstatus = strdup(lp+1);
}
else lognkstatus = "nk";
}
else if (strcmp(argv[i], "--timing") == 0) {
timing = 1;
}
else if (strcmp(argv[i], "--debug") == 0) {
debug = 1;
}
else if (strcmp(argv[i], "--no-update") == 0) {
dontsendmessages = 1;
}
else if (strcmp(argv[i], "--version") == 0) {
printf("bbgen version %s\n", VERSION);
printf("\n");
exit(0);
}
else if ((strcmp(argv[i], "--help") == 0) || (strcmp(argv[i], "-?") == 0)) {
printf("bbgen for Hobbit version %s\n\n", VERSION);
printf("Usage: %s [options] [WebpageDirectory]\n", argv[0]);
printf("Options:\n");
printf(" --ignorecolumns=test[,test] : Completely ignore these columns\n");
printf(" --nk-reds-only : Only show red statuses on the NK page\n");
printf(" --bb2-ignorecolumns=test[,test]: Ignore these columns for the BB2 page\n");
printf(" --bb2-ignorepurples : Ignore all-purple hosts on BB2 page\n");
printf(" --includecolumns=test[,test]: Always include these columns on BB2 page\n");
printf(" --max-eventcount=N : Max number of events to include in eventlog\n");
printf(" --max-eventtime=N : Show events that occurred within the last N minutes\n");
printf(" --eventignore=test[,test] : Columns to ignore in bb2 event-log display\n");
printf(" --no-eventlog : Do not generate the bb2 eventlog display\n");
printf(" --no-acklog : Do not generate the bb2 ack-log display\n");
printf(" --no-pages : Generate only the bb2 and bbnk pages\n");
printf(" --docurl=documentation-URL : Hostnames link to a general (dynamic) web page for docs\n");
printf(" --no-doc-window : Open doc-links in same window\n");
printf(" --htmlextension=.EXT : Sets filename extension for generated file (default: .html\n");
printf(" --report[=COLUMNNAME] : Send a status report about the running of bbgen\n");
printf(" --reportopts=ST:END:DYN:STL : Run in Hobbit Reporting mode\n");
printf(" --csv=FILENAME : For Hobbit Reporting, output CSV file\n");
printf(" --csvdelim=CHARACTER : Delimiter in CSV file output (default: comma)\n");
printf(" --snapshot=TIME : Snapshot mode\n");
printf("\nPage layout options:\n");
printf(" --pages-first : Put page- and subpage-links before hosts (default)\n");
printf(" --pages-last : Put page- and subpage-links after hosts\n");
printf(" --subpagecolumns=N : Number of columns for links to pages and subpages\n");
printf(" --maxrows=N : Repeat column headings for every N hosts shown\n");
printf(" --recentgifs : Use xxx-recent.gif icons for newly changed tests\n");
printf(" --sort-group-only-items : Display group-only items in alphabetical order\n");
printf(" --page-title=TITLE : Set a default page title for all pages\n");
printf(" --dialupskin=URL : Use a different icon skin for dialup tests\n");
printf(" --reverseskin=URL : Use a different icon skin for reverse tests\n");
printf(" --pagetitle-links : Make page- and subpage-titles act as links\n");
printf(" --pagetext-headings : Use page texts as headings\n");
printf(" --no-underline-headings : Do not underline the page headings\n");
printf("\nStatus propagation control options:\n");
printf(" --noprop=test[,test] : Disable upwards status propagation when YELLOW\n");
printf(" --nopropred=test[,test] : Disable upwards status propagation when RED or YELLOW\n");
printf(" --noproppurple=test[,test] : Disable upwards status propagation when PURPLE\n");
printf("\nAlternate pageset generation support:\n");
printf(" --pageset=SETNAME : Generate non-standard pageset with tag SETNAME\n");
printf(" --template=TEMPLATE : template for header and footer files\n");
printf("\nAlternate output formats:\n");
printf(" --wml[=test1,test2,...] : Generate a small (bb2-style) WML page\n");
printf(" --nstab=FILENAME : Generate a Netscape Sidebar feed\n");
printf(" --nslimit=COLOR : Minimum color to include on Netscape sidebar\n");
printf(" --rss : Generate a RSS/RDF feed of alerts\n");
printf(" --rssextension=.EXT : Sets filename extension for RSS files (default: .rss\n");
printf(" --rssversion={0.91|0.92|1.0|2.0} : Specify RSS/RDF version (default: 0.91)\n");
printf(" --rsslimit=COLOR : Minimum color to include on RSS feed\n");
printf("\nDebugging/troubleshooting options:\n");
printf(" --timing : Collect timing information\n");
printf(" --debug : Debugging information\n");
printf(" --version : Show version information\n");
printf(" --purplelog=FILENAME : Create a log of purple hosts and tests\n");
exit(0);
}
else if (argnmatch(argv[i], "-")) {
errprintf("Unknown option : %s\n", argv[i]);
}
else {
/* Last argument is pagedir */
pagedir = strdup(argv[i]);
}
}
/* In case they changed the name of our column ... */
if (egocolumn) setup_signalhandler(egocolumn);
if (debug) {
int i;
printf("Command: bbgen");
for (i=1; (i<argc); i++) printf(" '%s'", argv[i]);
printf("\n");
printf("Environment BBHOSTS='%s'\n", textornull(xgetenv("BBHOSTS")));
printf("\n");
}
add_timestamp("Startup");
/* Check that all needed environment vars are defined */
envcheck(reqenv);
/* Catch a SEGV fault */
setup_signalhandler("bbgen");
/* Set umask to 0022 so that the generated HTML pages have world-read access */
umask(0022);
if (pagedir == NULL) {
if (xgetenv("BBWWW")) {
pagedir = strdup(xgetenv("BBWWW"));
}
else {
pagedir = (char *) malloc(strlen(xgetenv("BBHOME"))+5);
sprintf(pagedir, "%s/www", xgetenv("BBHOME"));
}
}
if (xgetenv("BBHTACCESS")) bbhtaccess = strdup(xgetenv("BBHTACCESS"));
if (xgetenv("BBPAGEHTACCESS")) bbpagehtaccess = strdup(xgetenv("BBPAGEHTACCESS"));
if (xgetenv("BBSUBPAGEHTACCESS")) bbsubpagehtaccess = strdup(xgetenv("BBSUBPAGEHTACCESS"));
/*
* When doing alternate pagesets, disable some stuff:
* No WML or RSS pages.
*/
if (pageset || embedded || snapshot) enable_wmlgen = wantrss = 0;
if (embedded) {
egocolumn = htaccess = NULL;
/*
* Need to have default SIGPIPE handling when doing embedded stuff.
* We are probably run from a CGI script or something similar.
*/
signal(SIGPIPE, SIG_DFL);
}
/* Load all data from the various files */
load_all_links();
add_timestamp("Load links done");
pagehead = load_bbhosts(pageset);
add_timestamp("Load bbhosts done");
if (!embedded) {
/* Remove old acknowledgements */
delete_old_acks();
add_timestamp("ACK removal done");
}
statehead = load_state(&dispsums);
if (embedded || snapshot) dispsums = NULL;
add_timestamp("Load STATE done");
if (hobbitddump) {
dump_hobbitdchk();
return 0;
}
/* Calculate colors of hosts and pages */
calc_hostcolors(bb2ignorecolumns);
calc_pagecolors(pagehead);
/* Topmost page (background color for bb.html) */
for (p=pagehead; (p); p = p->next) {
if (p->color > pagehead->color) pagehead->color = p->color;
}
bb_color = pagehead->color;
if (xgetenv("SUMMARY_SET_BKG") && (strcmp(xgetenv("SUMMARY_SET_BKG"), "TRUE") == 0)) {
/*
* Displayed summaries affect the BB page only,
* but should not go into the color we report to
* others.
*/
for (s=dispsums; (s); s = s->next) {
if (s->color > pagehead->color) pagehead->color = s->color;
}
}
add_timestamp("Color calculation done");
if (debug) dumpall(pagehead);
/* Generate pages */
if (chdir(pagedir) != 0) {
errprintf("Cannot change to webpage directory %s\n", pagedir);
exit(1);
}
if (embedded) {
/* Just generate that one page */
do_one_page(pagehead, NULL, 1);
return 0;
}
/* The main page - bb.html and pages/subpages thereunder */
add_timestamp("Hobbit pagegen start");
if (reportstart && csvfile) {
csv_availability(csvfile, csvdelim);
}
else if (do_normal_pages) {
do_page_with_subs(pagehead, dispsums);
}
add_timestamp("Hobbit pagegen done");
if (reportstart) {
/* Reports end here */
return 0;
}
/* The full summary page - bb2.html */
bb2_color = do_bb2_page(nssidebarfilename, PAGE_BB2);
add_timestamp("BB2 generation done");
/* Reduced summary (alerts) page - bbnk.html */
bbnk_color = do_bb2_page(NULL, PAGE_NK);
add_timestamp("BBNK generation done");
if (snapshot) {
/* Snapshots end here */
return 0;
}
/* Send summary notices - only once, so not on pagesets */
if (pageset == NULL) {
send_summaries(sumhead);
add_timestamp("Summary transmission done");
}
/* Generate WML cards */
if (enable_wmlgen) {
do_wml_cards(pagedir);
add_timestamp("WML generation done");
}
/* Need to do this before sending in our report */
add_timestamp("Run completed");
/* Tell about us */
if (egocolumn) {
char msgline[4096];
char *timestamps;
long bbsleep = (xgetenv("BBSLEEP") ? atol(xgetenv("BBSLEEP")) : 300);
int color;
/* Go yellow if it runs for too long */
if (total_runtime() > bbsleep) {
errprintf("WARNING: Runtime %ld longer than BBSLEEP (%ld)\n", total_runtime(), bbsleep);
}
color = (errbuf ? COL_YELLOW : COL_GREEN);
combo_start();
init_status(color);
sprintf(msgline, "status %s.%s %s %s\n\n", xgetenv("MACHINE"), egocolumn, colorname(color), timestamp);
addtostatus(msgline);
sprintf(msgline, "bbgen for Hobbit version %s\n", VERSION);
addtostatus(msgline);
sprintf(msgline, "\nStatistics:\n Hosts : %5d\n Status messages : %5d\n Purple messages : %5d\n Pages : %5d\n",
hostcount, statuscount, purplecount, pagecount);
addtostatus(msgline);
if (errbuf) {
addtostatus("\n\nError output:\n");
addtostatus(errbuf);
}
show_timestamps(×tamps);
addtostatus(timestamps);
finish_status();
combo_end();
}
else show_timestamps(NULL);
return 0;
}
int main(int argc, char *argv[])
{
char *msg;
int seq;
int argi;
int alertcolors, alertinterval;
char *configfn = NULL;
char *checkfn = NULL;
int checkpointinterval = 900;
char acklogfn[PATH_MAX];
FILE *acklogfd = NULL;
char notiflogfn[PATH_MAX];
FILE *notiflogfd = NULL;
char *tracefn = NULL;
struct sigaction sa;
int configchanged;
time_t lastxmit = 0;
MEMDEFINE(acklogfn);
MEMDEFINE(notiflogfn);
libxymon_init(argv[0]);
/* Dont save the error buffer */
save_errbuf = 0;
/* Load alert config */
alertcolors = colorset(xgetenv("ALERTCOLORS"), ((1 << COL_GREEN) | (1 << COL_BLUE)));
alertinterval = 60*atoi(xgetenv("ALERTREPEAT"));
/* Create our loookup-trees */
hostnames = xtreeNew(strcasecmp);
testnames = xtreeNew(strcasecmp);
locations = xtreeNew(strcasecmp);
for (argi=1; (argi < argc); argi++) {
if (argnmatch(argv[argi], "--config=")) {
configfn = strdup(strchr(argv[argi], '=')+1);
}
else if (argnmatch(argv[argi], "--checkpoint-file=")) {
checkfn = strdup(strchr(argv[argi], '=')+1);
}
else if (argnmatch(argv[argi], "--checkpoint-interval=")) {
char *p = strchr(argv[argi], '=') + 1;
checkpointinterval = atoi(p);
}
else if (argnmatch(argv[argi], "--dump-config")) {
load_alertconfig(configfn, alertcolors, alertinterval);
dump_alertconfig(1);
return 0;
}
else if (argnmatch(argv[argi], "--cfid")) {
include_configid = 1;
}
else if (argnmatch(argv[argi], "--test")) {
char *testhost = NULL, *testservice = NULL, *testpage = NULL,
*testcolor = "red", *testgroups = NULL;
void *hinfo;
int testdur = 0;
FILE *logfd = NULL;
activealerts_t *awalk = NULL;
int paramno = 0;
argi++; if (argi < argc) testhost = argv[argi];
argi++; if (argi < argc) testservice = argv[argi];
argi++;
while (argi < argc) {
if (strncasecmp(argv[argi], "--duration=", 11) == 0) {
testdur = durationvalue(strchr(argv[argi], '=')+1);
}
else if (strncasecmp(argv[argi], "--color=", 8) == 0) {
testcolor = strchr(argv[argi], '=')+1;
}
else if (strncasecmp(argv[argi], "--group=", 8) == 0) {
testgroups = strchr(argv[argi], '=')+1;
}
else if (strncasecmp(argv[argi], "--time=", 7) == 0) {
fakestarttime = (time_t)atoi(strchr(argv[argi], '=')+1);
}
else {
paramno++;
if (paramno == 1) testdur = atoi(argv[argi]);
else if (paramno == 2) testcolor = argv[argi];
else if (paramno == 3) fakestarttime = (time_t) atoi(argv[argi]);
}
argi++;
}
if ((testhost == NULL) || (testservice == NULL)) {
printf("Usage: xymond_alert --test HOST SERVICE [options]\n");
printf("Possible options:\n\t[--duration=MINUTES]\n\t[--color=COLOR]\n\t[--group=GROUPNAME]\n\t[--time=TIMESPEC]\n");
return 1;
}
load_hostnames(xgetenv("HOSTSCFG"), NULL, get_fqdn());
hinfo = hostinfo(testhost);
if (hinfo) {
testpage = strdup(xmh_item(hinfo, XMH_ALLPAGEPATHS));
}
else {
errprintf("Host not found in hosts.cfg - assuming it is on the top page\n");
testpage = "";
}
awalk = (activealerts_t *)calloc(1, sizeof(activealerts_t));
awalk->hostname = find_name(hostnames, testhost);
awalk->testname = find_name(testnames, testservice);
awalk->location = find_name(locations, testpage);
awalk->ip = strdup("127.0.0.1");
awalk->color = awalk->maxcolor = parse_color(testcolor);
awalk->pagemessage = "Test of the alert configuration";
awalk->eventstart = getcurrenttime(NULL) - testdur*60;
awalk->groups = (testgroups ? strdup(testgroups) : NULL);
awalk->state = A_PAGING;
awalk->cookie = 12345;
awalk->next = NULL;
logfd = fopen("/dev/null", "w");
starttrace(NULL);
testonly = 1;
load_alertconfig(configfn, alertcolors, alertinterval);
load_holidays(0);
send_alert(awalk, logfd);
return 0;
}
else if (argnmatch(argv[argi], "--trace=")) {
tracefn = strdup(strchr(argv[argi], '=')+1);
starttrace(tracefn);
}
else if (net_worker_option(argv[argi])) {
/* Handled in the subroutine */
}
else if (standardoption(argv[argi])) {
if (showhelp) return 0;
}
else {
errprintf("Unknown option '%s'\n", argv[argi]);
}
}
/* Do the network stuff if needed */
net_worker_run(ST_ALERT, LOC_SINGLESERVER, NULL);
if (checkfn) {
load_checkpoint(checkfn);
nextcheckpoint = gettimer() + checkpointinterval;
dbgprintf("Next checkpoint at %d, interval %d\n", (int) nextcheckpoint, checkpointinterval);
}
setup_signalhandler("xymond_alert");
/* Need to handle these ourselves, so we can shutdown and save state-info */
memset(&sa, 0, sizeof(sa));
sa.sa_handler = sig_handler;
sigaction(SIGPIPE, &sa, NULL);
sigaction(SIGTERM, &sa, NULL);
sigaction(SIGINT, &sa, NULL);
sigaction(SIGCHLD, &sa, NULL);
sigaction(SIGUSR1, &sa, NULL);
sigaction(SIGHUP, &sa, NULL);
if (xgetenv("XYMONSERVERLOGS")) {
sprintf(acklogfn, "%s/acknowledge.log", xgetenv("XYMONSERVERLOGS"));
acklogfd = fopen(acklogfn, "a");
sprintf(notiflogfn, "%s/notifications.log", xgetenv("XYMONSERVERLOGS"));
notiflogfd = fopen(notiflogfn, "a");
}
/*
* The general idea here is that this loop handles receiving of alert-
* and ack-messages from the master daemon, and maintains a list of
* host+test combinations that may have alerts going out.
*
* This module does not deal with any specific alert-configuration,
* it just picks up the alert messages, maintains the list of
* known tests that are in some sort of critical condition, and
* periodically pushes alerts to the do_alert.c module for handling.
*
* The only modification of alerts that happen here is the handling
* of when the next alert is due. It calls into the next_alert()
* routine to learn when an alert should be repeated, and also
* deals with Acknowledgments that stop alerts from going out for
* a period of time.
*/
while (running) {
char *eoln, *restofmsg;
char *metadata[20];
char *p;
int metacount;
char *hostname = NULL, *testname = NULL;
struct timespec timeout;
time_t now, nowtimer;
int anytogo;
activealerts_t *awalk;
int childstat;
nowtimer = gettimer();
if (checkfn && (nowtimer > nextcheckpoint)) {
dbgprintf("Saving checkpoint\n");
nextcheckpoint = nowtimer + checkpointinterval;
save_checkpoint(checkfn);
if (acklogfd) acklogfd = freopen(acklogfn, "a", acklogfd);
if (notiflogfd) notiflogfd = freopen(notiflogfn, "a", notiflogfd);
}
timeout.tv_sec = 60; timeout.tv_nsec = 0;
msg = get_xymond_message(C_PAGE, "xymond_alert", &seq, &timeout);
if (msg == NULL) {
running = 0;
continue;
}
/* See what time it is - must happen AFTER the timeout */
now = getcurrenttime(NULL);
/* Split the message in the first line (with meta-data), and the rest */
eoln = strchr(msg, '\n');
if (eoln) {
*eoln = '\0';
restofmsg = eoln+1;
}
else {
restofmsg = "";
}
/*
* Now parse the meta-data into elements.
* We use our own "gettok()" routine which works
* like strtok(), but can handle empty elements.
*/
metacount = 0;
memset(&metadata, 0, sizeof(metadata));
p = gettok(msg, "|");
while (p && (metacount < 19)) {
metadata[metacount] = p;
metacount++;
p = gettok(NULL, "|");
}
metadata[metacount] = NULL;
if (metacount > 3) hostname = metadata[3];
if (metacount > 4) testname = metadata[4];
if ((metacount > 10) && (strncmp(metadata[0], "@@page", 6) == 0)) {
/* @@page|timestamp|sender|hostname|testname|hostip|expiretime|color|prevcolor|changetime|location|cookie|osname|classname|grouplist|modifiers */
int newcolor, newalertstatus, oldalertstatus;
dbgprintf("Got page message from %s:%s\n", hostname, testname);
traceprintf("@@page %s:%s:%s=%s\n", hostname, testname, metadata[10], metadata[7]);
awalk = find_active(hostname, testname);
if (awalk == NULL) {
char *hwalk = find_name(hostnames, hostname);
char *twalk = find_name(testnames, testname);
char *pwalk = find_name(locations, metadata[10]);
awalk = (activealerts_t *)calloc(1, sizeof(activealerts_t));
awalk->hostname = hwalk;
awalk->testname = twalk;
awalk->location = pwalk;
awalk->cookie = -1;
awalk->state = A_DEAD;
/*
* Use changetime here, if we restart the alert module then
* this gets the duration values more right than using "now".
* Also, define this only when a new alert arrives - we should
* NOT clear this when a status goes yellow->red, or if it
* flaps between yellow and red.
*/
awalk->eventstart = atoi(metadata[9]);
add_active(awalk->hostname, awalk);
traceprintf("New record\n");
}
newcolor = parse_color(metadata[7]);
oldalertstatus = ((alertcolors & (1 << awalk->color)) != 0);
newalertstatus = ((alertcolors & (1 << newcolor)) != 0);
traceprintf("state %d->%d\n", oldalertstatus, newalertstatus);
if (newalertstatus) {
/* It's in an alert state. */
awalk->color = newcolor;
awalk->state = A_PAGING;
if (newcolor > awalk->maxcolor) {
if (awalk->maxcolor != 0) {
/*
* Severity has increased (yellow -> red).
* Clear the repeat-interval, and set maxcolor to
* the new color. If it drops to yellow again,
* maxcolor stays at red, so a test that flaps
* between yellow and red will only alert on red
* the first time, and then follow the repeat
* interval.
*/
dbgprintf("Severity increased, cleared repeat interval: %s/%s %s->%s\n",
awalk->hostname, awalk->testname,
colorname(awalk->maxcolor), colorname(newcolor));
clear_interval(awalk);
}
awalk->maxcolor = newcolor;
}
}
else {
/*
* Send one "recovered" message out now, then go to A_DEAD.
* Dont update the color here - we want recoveries to go out
* only if the alert color triggered an alert
*/
awalk->state = (newcolor == COL_BLUE) ? A_DISABLED : A_RECOVERED;
}
if (oldalertstatus != newalertstatus) {
dbgprintf("Alert status changed from %d to %d\n", oldalertstatus, newalertstatus);
clear_interval(awalk);
}
if (awalk->ip) xfree(awalk->ip);
awalk->ip = strdup(metadata[5]);
awalk->cookie = atoi(metadata[11]);
if (awalk->osname) xfree(awalk->osname);
awalk->osname = (metadata[12] ? strdup(metadata[12]) : NULL);
if (awalk->classname) xfree(awalk->classname);
awalk->classname = (metadata[13] ? strdup(metadata[13]) : NULL);
if (awalk->groups) xfree(awalk->groups);
awalk->groups = (metadata[14] ? strdup(metadata[14]) : NULL);
if (awalk->pagemessage) xfree(awalk->pagemessage);
if (metadata[15]) {
/* Modifiers are more interesting than the message itself */
awalk->pagemessage = (char *)malloc(strlen(awalk->hostname) + strlen(awalk->testname) + strlen(colorname(awalk->color)) + strlen(metadata[15]) + strlen(restofmsg) + 10);
sprintf(awalk->pagemessage, "%s:%s %s\n%s\n%s",
awalk->hostname, awalk->testname, colorname(awalk->color), metadata[15], restofmsg);
}
else {
awalk->pagemessage = strdup(restofmsg);
}
}
else if ((metacount > 5) && (strncmp(metadata[0], "@@ack", 5) == 0)) {
/* @@ack|timestamp|sender|hostname|testname|hostip|expiretime */
/*
* An ack is handled simply by setting the next
* alert-time to when the ack expires.
*/
time_t nextalert = atoi(metadata[6]);
dbgprintf("Got ack message from %s:%s\n", hostname, testname);
traceprintf("@@ack: %s:%s now=%d, ackeduntil %d\n",
hostname, testname, (int)now, (int)nextalert);
awalk = find_active(hostname, testname);
if (acklogfd) {
int cookie = (awalk ? awalk->cookie : -1);
int color = (awalk ? awalk->color : 0);
fprintf(acklogfd, "%d\t%d\t%d\t%d\t%s\t%s.%s\t%s\t%s\n",
(int)now, cookie,
(int)((nextalert - now) / 60), cookie,
"np_filename_not_used",
hostname, testname,
colorname(color),
nlencode(restofmsg));
fflush(acklogfd);
}
if (awalk && (awalk->state == A_PAGING)) {
traceprintf("Record updated\n");
awalk->state = A_ACKED;
awalk->nextalerttime = nextalert;
if (awalk->ackmessage) xfree(awalk->ackmessage);
awalk->ackmessage = strdup(restofmsg);
}
else {
traceprintf("No record\n");
}
}
else if ((metacount > 4) && (strncmp(metadata[0], "@@notify", 5) == 0)) {
/* @@notify|timestamp|sender|hostname|testname|pagepath */
char *hwalk = find_name(hostnames, hostname);
char *twalk = find_name(testnames, testname);
char *pwalk = find_name(locations, (metadata[5] ? metadata[5] : ""));
awalk = (activealerts_t *)calloc(1, sizeof(activealerts_t));
awalk->hostname = hwalk;
awalk->testname = twalk;
awalk->location = pwalk;
awalk->cookie = -1;
awalk->pagemessage = strdup(restofmsg);
awalk->eventstart = getcurrenttime(NULL);
awalk->state = A_NOTIFY;
add_active(awalk->hostname, awalk);
}
else if ((metacount > 3) &&
((strncmp(metadata[0], "@@drophost", 10) == 0) || (strncmp(metadata[0], "@@dropstate", 11) == 0))) {
/* @@drophost|timestamp|sender|hostname */
/* @@dropstate|timestamp|sender|hostname */
xtreePos_t handle;
handle = xtreeFind(hostnames, hostname);
if (handle != xtreeEnd(hostnames)) {
alertanchor_t *anchor = (alertanchor_t *)xtreeData(hostnames, handle);
for (awalk = anchor->head; (awalk); awalk = awalk->next) awalk->state = A_DEAD;
}
}
else if ((metacount > 4) && (strncmp(metadata[0], "@@droptest", 10) == 0)) {
/* @@droptest|timestamp|sender|hostname|testname */
awalk = find_active(hostname, testname);
if (awalk) awalk->state = A_DEAD;
}
else if ((metacount > 4) && (strncmp(metadata[0], "@@renamehost", 12) == 0)) {
/* @@renamehost|timestamp|sender|hostname|newhostname */
/*
* We handle rename's simply by dropping the alert. If there is still an
* active alert for the host, it will have to be dealt with when the next
* status update arrives.
*/
xtreePos_t handle;
handle = xtreeFind(hostnames, hostname);
if (handle != xtreeEnd(hostnames)) {
alertanchor_t *anchor = (alertanchor_t *)xtreeData(hostnames, handle);
for (awalk = anchor->head; (awalk); awalk = awalk->next) awalk->state = A_DEAD;
}
}
else if ((metacount > 5) && (strncmp(metadata[0], "@@renametest", 12) == 0)) {
/* @@renametest|timestamp|sender|hostname|oldtestname|newtestname */
/*
* We handle rename's simply by dropping the alert. If there is still an
* active alert for the host, it will have to be dealt with when the next
* status update arrives.
*/
awalk = find_active(hostname, testname);
if (awalk) awalk->state = A_DEAD;
}
else if (strncmp(metadata[0], "@@shutdown", 10) == 0) {
running = 0;
errprintf("Got a shutdown message\n");
continue;
}
else if (strncmp(metadata[0], "@@logrotate", 11) == 0) {
char *fn = xgetenv("XYMONCHANNEL_LOGFILENAME");
if (fn && strlen(fn)) {
reopen_file(fn, "a", stdout);
reopen_file(fn, "a", stderr);
if (tracefn) {
stoptrace();
starttrace(tracefn);
}
}
continue;
}
else if (strncmp(metadata[0], "@@reload", 8) == 0) {
/* Nothing ... right now */
}
else if (strncmp(metadata[0], "@@idle", 6) == 0) {
/* Timeout */
}
/*
* When a burst of alerts happen, we get lots of alert messages
* coming in quickly. So lets handle them in bunches and only
* do the full alert handling once every 10 secs - that lets us
* combine a bunch of alerts into one transmission process.
*/
if (nowtimer < (lastxmit+10)) continue;
lastxmit = nowtimer;
/*
* Loop through the activealerts list and see if anything is pending.
* This is an optimization, we could just as well just fork off the
* notification child and let it handle all of it. But there is no
* reason to fork a child process unless it is going to do something.
*/
configchanged = load_alertconfig(configfn, alertcolors, alertinterval);
configchanged += load_holidays(0);
anytogo = 0;
for (awalk = alistBegin(); (awalk); awalk = alistNext()) {
int anymatch = 0;
switch (awalk->state) {
case A_NORECIP:
if (!configchanged) break;
/* The configuration has changed - switch NORECIP -> PAGING */
awalk->state = A_PAGING;
clear_interval(awalk);
/* Fall through */
case A_PAGING:
if (have_recipient(awalk, &anymatch)) {
if (awalk->nextalerttime <= now) anytogo++;
}
else {
if (!anymatch) {
awalk->state = A_NORECIP;
cleanup_alert(awalk);
}
}
break;
case A_ACKED:
if (awalk->nextalerttime <= now) {
/* An ack has expired, so drop the ack message and switch to A_PAGING */
anytogo++;
if (awalk->ackmessage) xfree(awalk->ackmessage);
awalk->state = A_PAGING;
}
break;
case A_RECOVERED:
case A_DISABLED:
case A_NOTIFY:
anytogo++;
break;
case A_DEAD:
break;
}
}
dbgprintf("%d alerts to go\n", anytogo);
if (anytogo) {
pid_t childpid;
childpid = fork();
if (childpid == 0) {
/* The child */
start_alerts();
for (awalk = alistBegin(); (awalk); awalk = alistNext()) {
switch (awalk->state) {
case A_PAGING:
if (awalk->nextalerttime <= now) {
send_alert(awalk, notiflogfd);
}
break;
case A_ACKED:
/* Cannot be A_ACKED unless the ack is still valid, so no alert. */
break;
case A_RECOVERED:
case A_DISABLED:
case A_NOTIFY:
send_alert(awalk, notiflogfd);
break;
case A_NORECIP:
case A_DEAD:
break;
}
}
finish_alerts();
/* Child does not continue */
exit(0);
}
else if (childpid < 0) {
errprintf("Fork failed, cannot send alerts: %s\n", strerror(errno));
}
}
/* Update the state flag and the next-alert timestamp */
for (awalk = alistBegin(); (awalk); awalk = alistNext()) {
switch (awalk->state) {
case A_PAGING:
if (awalk->nextalerttime <= now) awalk->nextalerttime = next_alert(awalk);
break;
case A_NORECIP:
break;
case A_ACKED:
/* Still cannot get here except if ack is still valid */
break;
case A_RECOVERED:
case A_DISABLED:
case A_NOTIFY:
awalk->state = A_DEAD;
/* Fall through */
case A_DEAD:
cleanup_alert(awalk);
break;
}
}
clean_all_active();
/* Pickup any finished child processes to avoid zombies */
while (wait3(&childstat, WNOHANG, NULL) > 0) ;
}
if (checkfn) save_checkpoint(checkfn);
if (acklogfd) fclose(acklogfd);
if (notiflogfd) fclose(notiflogfd);
stoptrace();
MEMUNDEFINE(notiflogfn);
MEMUNDEFINE(acklogfn);
if (termsig >= 0) {
errprintf("Terminated by signal %d\n", termsig);
}
return 0;
}
static int net_worker_listener(char *ipport)
{
/*
* Setup a listener socket on IP+port. When a connection arrives,
* pick it up, fork() and let the rest of the input go via the
* network socket.
*/
char *listenip, *p;
int listenport = 0;
int lsocket = -1;
struct sockaddr_in laddr;
struct sigaction sa;
int opt;
listenip = ipport;
p = strchr(listenip, ':');
if (p) {
*p = '\0';
listenport = atoi(p+1);
}
if (listenport == 0) {
errprintf("Must include PORT number in --listen=IP:PORT option\n");
return -1;
}
/* Set up a socket to listen for new connections */
errprintf("Setting up network listener on %s:%d\n", listenip, listenport);
memset(&laddr, 0, sizeof(laddr));
if ((strlen(listenip) == 0) || (strcmp(listenip, "0.0.0.0") == 0)) {
listenip = "0.0.0.0";
laddr.sin_addr.s_addr = INADDR_ANY;
}
else if (inet_aton(listenip, (struct in_addr *) &laddr.sin_addr.s_addr) == 0) {
/* Not an IP */
errprintf("Listener IP must be an IP-address, not hostname\n");
return -1;
}
laddr.sin_port = htons(listenport);
laddr.sin_family = AF_INET;
lsocket = socket(AF_INET, SOCK_STREAM, 0);
if (lsocket == -1) {
errprintf("Cannot create listen socket (%s)\n", strerror(errno));
return -1;
}
opt = 1;
setsockopt(lsocket, SOL_SOCKET, SO_REUSEADDR, &opt, sizeof(opt));
if (bind(lsocket, (struct sockaddr *)&laddr, sizeof(laddr)) == -1) {
errprintf("Cannot bind to listen socket (%s)\n", strerror(errno));
return -1;
}
if (listen(lsocket, 5) == -1) {
errprintf("Cannot listen (%s)\n", strerror(errno));
return -1;
}
/* Make listener socket non-blocking, so we can send keep-alive's while waiting for connections */
fcntl(lsocket, F_SETFL, O_NONBLOCK);
/* Catch some signals */
setup_signalhandler("xymond_listener");
memset(&sa, 0, sizeof(sa));
sa.sa_handler = netinp_sighandler;
sigaction(SIGCHLD, &sa, NULL);
sigaction(SIGTERM, &sa, NULL);
sigaction(SIGINT, &sa, NULL);
while (running) {
struct sockaddr_in netaddr;
int addrsz, sock;
int childstat;
pid_t childpid;
fd_set fdread;
struct timeval tmo;
int n;
FD_ZERO(&fdread);
FD_SET(lsocket, &fdread);
tmo.tv_sec = 60; tmo.tv_usec = 0;
n = select(lsocket+1, &fdread, NULL, NULL, &tmo);
if (n == -1) {
if (errno != EINTR) {
errprintf("select() failed while waiting for connection : %s\n", strerror(errno));
running = 0;
}
}
else if (n == 0) {
/* Timeout */
net_worker_heartbeat();
}
else {
/* We have a connection ready */
addrsz = sizeof(netaddr);
sock = accept(lsocket, (struct sockaddr *)&netaddr, &addrsz);
if (sock >= 0) {
/* Got a new connection */
childpid = fork();
if (childpid == 0) {
/* Child takes input from the new socket, and starts working */
close(lsocket); /* Close the listener socket */
inputfd = sock;
return 0;
}
else if (childpid > 0) {
/* Parent closes the new socket (child has it) */
close(sock);
continue;
}
else {
errprintf("Error forking worker for new connection: %s\n", strerror(errno));
running = 0;
continue;
}
}
else {
/* Error while waiting for accept() to complete */
if (errno != EINTR) {
errprintf("accept() failed: %s\n", strerror(errno));
running = 0;
}
}
}
/* Pickup failed children */
while ((childpid = wait3(&childstat, WNOHANG, NULL)) > 0);
}
/* Close the listener socket */
close(lsocket);
/* Kill any children that are still around */
kill(0, SIGTERM);
return 1;
}
int main(int argc, char *argv[])
{
char *msg;
int running;
int argi, seq;
int recentperiod = 3600;
int maxrecentcount = 5;
int logdirfull = 0;
int minlogspace = 5;
struct sigaction sa;
/* Handle program options. */
for (argi = 1; (argi < argc); argi++) {
if (argnmatch(argv[argi], "--logdir=")) {
clientlogdir = strchr(argv[argi], '=')+1;
}
else if (argnmatch(argv[argi], "--recent-period=")) {
char *p = strchr(argv[argi], '=');
recentperiod = 60*atoi(p+1);
}
else if (argnmatch(argv[argi], "--recent-count=")) {
char *p = strchr(argv[argi], '=');
maxrecentcount = atoi(p+1);
}
else if (argnmatch(argv[argi], "--minimum-free=")) {
minlogspace = atoi(strchr(argv[argi], '=')+1);
}
else if (strcmp(argv[argi], "--debug") == 0) {
/*
* A global "debug" variable is available. If
* it is set, then "dbgprintf()" outputs debug messages.
*/
debug = 1;
}
else if (net_worker_option(argv[argi])) {
/* Handled in the subroutine */
}
}
if (clientlogdir == NULL) clientlogdir = xgetenv("CLIENTLOGS");
if (clientlogdir == NULL) {
clientlogdir = (char *)malloc(strlen(xgetenv("XYMONVAR")) + 10);
sprintf(clientlogdir, "%s/hostdata", xgetenv("XYMONVAR"));
}
save_errbuf = 0;
/* Do the network stuff if needed */
net_worker_run(ST_HOSTDATA, LOC_STICKY, update_locator_hostdata);
setup_signalhandler("xymond_hostdata");
memset(&sa, 0, sizeof(sa));
sa.sa_handler = sig_handler;
signal(SIGCHLD, SIG_IGN);
sigaction(SIGHUP, &sa, NULL);
signal(SIGPIPE, SIG_DFL);
savetimes = xtreeNew(strcasecmp);
running = 1;
while (running) {
char *eoln, *restofmsg, *p;
char *metadata[MAX_META+1];
int metacount;
msg = get_xymond_message(C_CLICHG, "xymond_hostdata", &seq, NULL);
if (msg == NULL) {
/*
* get_xymond_message will return NULL if xymond_channel closes
* the input pipe. We should shutdown when that happens.
*/
running = 0;
continue;
}
if (nextfscheck < gettimer()) {
logdirfull = (chkfreespace(clientlogdir, minlogspace, minlogspace) != 0);
if (logdirfull) errprintf("Hostdata directory %s has less than %d%% free space - disabling save of data for 5 minutes\n", clientlogdir, minlogspace);
nextfscheck = gettimer() + 300;
}
/* Split the message in the first line (with meta-data), and the rest */
eoln = strchr(msg, '\n');
if (eoln) {
*eoln = '\0';
restofmsg = eoln+1;
}
else {
restofmsg = "";
}
metacount = 0;
memset(&metadata, 0, sizeof(metadata));
p = gettok(msg, "|");
while (p && (metacount < MAX_META)) {
metadata[metacount++] = p;
p = gettok(NULL, "|");
}
metadata[metacount] = NULL;
if (strncmp(metadata[0], "@@clichg", 8) == 0) {
xtreePos_t handle;
savetimes_t *itm;
int i, recentcount;
time_t now = gettimer();
char hostdir[PATH_MAX];
char fn[PATH_MAX];
FILE *fd;
/* metadata[3] is the hostname */
handle = xtreeFind(savetimes, metadata[3]);
if (handle != xtreeEnd(savetimes)) {
itm = (savetimes_t *)xtreeData(savetimes, handle);
}
else {
itm = (savetimes_t *)calloc(1, sizeof(savetimes_t));
itm->hostname = strdup(metadata[3]);
xtreeAdd(savetimes, itm->hostname, itm);
}
/* See how many times we've saved the hostdata recently (within the past 'recentperiod' seconds) */
for (i=0, recentcount=0; ((i < 12) && (itm->tstamp[i] > (now - recentperiod))); i++) recentcount++;
/* If it's been saved less than 'maxrecentcount' times, then save it. Otherwise just drop it */
if (!logdirfull && (recentcount < maxrecentcount)) {
int written, closestatus, ok = 1;
for (i = 10; (i > 0); i--) itm->tstamp[i+1] = itm->tstamp[i];
itm->tstamp[0] = now;
sprintf(hostdir, "%s/%s", clientlogdir, metadata[3]);
mkdir(hostdir, S_IRWXU|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH);
sprintf(fn, "%s/%s", hostdir, metadata[4]);
fd = fopen(fn, "w");
if (fd == NULL) {
errprintf("Cannot create file %s: %s\n", fn, strerror(errno));
continue;
}
written = fwrite(restofmsg, 1, strlen(restofmsg), fd);
if (written != strlen(restofmsg)) {
errprintf("Cannot write hostdata file %s: %s\n", fn, strerror(errno));
closestatus = fclose(fd); /* Ignore any close errors */
ok = 0;
}
else {
closestatus = fclose(fd);
if (closestatus != 0) {
errprintf("Cannot write hostdata file %s: %s\n", fn, strerror(errno));
ok = 0;
}
}
if (!ok) remove(fn);
}
}
/*
* A "shutdown" message is sent when the master daemon
* terminates. The child workers should shutdown also.
*/
else if (strncmp(metadata[0], "@@shutdown", 10) == 0) {
running = 0;
continue;
}
else if (strncmp(metadata[0], "@@idle", 6) == 0) {
/* Ignored */
continue;
}
/*
* A "logrotate" message is sent when the Xymon logs are
* rotated. The child workers must re-open their logfiles,
* typically stdin and stderr - the filename is always
* provided in the XYMONCHANNEL_LOGFILENAME environment.
*/
else if (strncmp(metadata[0], "@@logrotate", 11) == 0) {
char *fn = xgetenv("XYMONCHANNEL_LOGFILENAME");
if (fn && strlen(fn)) {
reopen_file(fn, "a", stdout);
reopen_file(fn, "a", stderr);
}
continue;
}
else if ((metacount > 3) && (strncmp(metadata[0], "@@drophost", 10) == 0)) {
/* @@drophost|timestamp|sender|hostname */
char hostdir[PATH_MAX];
snprintf(hostdir, sizeof(hostdir), "%s/%s", clientlogdir, basename(metadata[3]));
dropdirectory(hostdir, 1);
}
else if ((metacount > 4) && (strncmp(metadata[0], "@@renamehost", 12) == 0)) {
/* @@renamehost|timestamp|sender|hostname|newhostname */
char oldhostdir[PATH_MAX], newhostdir[PATH_MAX];
snprintf(oldhostdir, sizeof(oldhostdir), "%s/%s", clientlogdir, basename(metadata[3]));
snprintf(newhostdir, sizeof(newhostdir), "%s/%s", clientlogdir, basename(metadata[4]));
rename(oldhostdir, newhostdir);
if (net_worker_locatorbased()) locator_rename_host(metadata[3], metadata[4], ST_HOSTDATA);
}
else if (strncmp(metadata[0], "@@reload", 8) == 0) {
/* Do nothing */
}
}
return 0;
}
int main(int argc, char *argv[])
{
int argi;
struct sigaction sa;
namelist_t *hostwalk;
time_t nexttimeout;
for (argi=1; (argi < argc); argi++) {
if (argnmatch(argv[argi], "--server=")) {
char *p = strchr(argv[argi], '=');
serverip = strdup(p+1);
}
else if (argnmatch(argv[argi], "--interval=")) {
char *p = strchr(argv[argi], '=');
pollinterval = atoi(p+1);
}
else if (argnmatch(argv[argi], "--log-interval=")) {
char *p = strchr(argv[argi], '=');
errorloginterval = atoi(p+1);
}
else if (argnmatch(argv[argi], "--id=")) {
char *p = strchr(argv[argi], '=');
serverid = atoi(p+1);
}
else if (strcmp(argv[argi], "--debug") == 0) {
debug = 1;
}
}
setup_signalhandler("hobbitfetch");
memset(&sa, 0, sizeof(sa));
sa.sa_handler = sigmisc_handler;
sigaction(SIGHUP, &sa, NULL);
sigaction(SIGTERM, &sa, NULL);
sigaction(SIGUSR1, &sa, NULL); /* SIGUSR1 triggers logging of active requests */
clients = rbtNew(name_compare);
nexttimeout = time(NULL) + 60;
{
/* Seed the random number generator */
struct timeval tv;
struct timezone tz;
gettimeofday(&tv, &tz);
srandom(tv.tv_usec);
}
do {
RbtIterator handle;
conn_t *connwalk, *cprev;
fd_set fdread, fdwrite;
int n, maxfd;
struct timeval tmo;
time_t now;
now = time(NULL);
if (now > reloadtime) {
/* Time to reload the bb-hosts file */
reloadtime = now + 600;
load_hostnames(xgetenv("BBHOSTS"), NULL, get_fqdn());
for (hostwalk = first_host(); (hostwalk); hostwalk = hostwalk->next) {
char *hname;
clients_t *newclient;
if (!bbh_item(hostwalk, BBH_FLAG_PULLDATA)) continue;
hname = bbh_item(hostwalk, BBH_HOSTNAME);
handle = rbtFind(clients, hname);
if (handle == rbtEnd(clients)) {
newclient = (clients_t *)calloc(1, sizeof(clients_t));
newclient->hostname = strdup(hname);
rbtInsert(clients, newclient->hostname, newclient);
whentoqueue = now;
}
}
}
now = time(NULL);
if (now > nexttimeout) {
/* Check for connections that have timed out */
nexttimeout = now + 60;
for (connwalk = chead; (connwalk); connwalk = connwalk->next) {
if ((connwalk->tstamp + 60) < now) {
if (debug || (connwalk->client->nexterrortxt < now)) {
errprintf("Timeout while talking to %s (req %lu): Aborting session\n",
addrstring(&connwalk->caddr), connwalk->seq);
connwalk->client->nexterrortxt = now + errorloginterval;
}
flag_cleanup(connwalk);
}
}
}
if (needcleanup) {
/* Remove any finished requests */
needcleanup = 0;
connwalk = chead; cprev = NULL;
dbgprintf("Doing cleanup\n");
while (connwalk) {
conn_t *zombie;
if ((connwalk->action == C_READING) || (connwalk->action == C_WRITING)) {
/* Active connection - skip to the next conn_t record */
cprev = connwalk;
connwalk = connwalk->next;
continue;
}
if (connwalk->action == C_CLEANUP) {
if (connwalk->ctype == C_CLIENT) {
/*
* Finished getting data from a client,
* flag idle and set next poll time.
*/
connwalk->client->busy = 0;
set_polltime(connwalk->client);
}
else if (connwalk->ctype == C_SERVER) {
/* Nothing needed for server cleanups */
}
}
/* Unlink the request from the list of active connections */
zombie = connwalk;
if (cprev == NULL) {
chead = zombie->next;
connwalk = chead;
cprev = NULL;
}
else {
cprev->next = zombie->next;
connwalk = zombie->next;
}
/* Purge the zombie */
dbgprintf("Request completed: req %lu, peer %s, action was %d, type was %d\n",
zombie->seq, addrstring(&zombie->caddr),
zombie->action, zombie->ctype);
close(zombie->sockfd);
freestrbuffer(zombie->msgbuf);
xfree(zombie);
}
/* Set the tail pointer correctly */
ctail = chead;
if (ctail) { while (ctail->next) ctail = ctail->next; }
}
if (dumpsessions) {
/* Set by SIGUSR1 - dump the list of active requests */
dumpsessions = 0;
for (connwalk = chead; (connwalk); connwalk = connwalk->next) {
char *ctypestr, *actionstr;
char timestr[30];
switch (connwalk->ctype) {
case C_CLIENT: ctypestr = "client"; break;
case C_SERVER: ctypestr = "server"; break;
}
switch (connwalk->action) {
case C_READING: actionstr = "reading"; break;
case C_WRITING: actionstr = "writing"; break;
case C_CLEANUP: actionstr = "cleanup"; break;
}
strftime(timestr, sizeof(timestr), "%Y-%m-%d %H:%M:%S",
localtime(&connwalk->tstamp));
errprintf("Request %lu: state %s/%s, peer %s, started %s (%lu secs ago)\n",
connwalk->seq, ctypestr, actionstr, addrstring(&connwalk->caddr),
timestr, (now - connwalk->tstamp));
}
}
now = time(NULL);
if (now >= whentoqueue) {
/* Scan host-tree for clients we need to contact */
for (handle = rbtBegin(clients); (handle != rbtEnd(clients)); handle = rbtNext(clients, handle)) {
clients_t *clientwalk;
char msgline[100];
strbuffer_t *request;
char *pullstr, *ip;
int port;
clientwalk = (clients_t *)gettreeitem(clients, handle);
if (clientwalk->busy) continue;
if (clientwalk->nextpoll > now) continue;
/* Deleted hosts stay in our tree - but should disappear from the known hosts */
hostwalk = hostinfo(clientwalk->hostname); if (!hostwalk) continue;
pullstr = bbh_item(hostwalk, BBH_FLAG_PULLDATA); if (!pullstr) continue;
ip = strchr(pullstr, '=');
port = atoi(xgetenv("BBPORT"));
if (!ip) {
ip = strdup(bbh_item(hostwalk, BBH_IP));
}
else {
/* There is an explicit IP setting in the pulldata tag */
char *p;
ip++; /* Skip the '=' */
ip = strdup(ip);
p = strchr(ip, ':');
if (p) { *p = '\0'; port = atoi(p+1); }
if (*ip == '\0') {
/* No IP given, just a port number */
xfree(ip);
ip = strdup(bbh_item(hostwalk, BBH_IP));
}
}
if (strcmp(ip, "0.0.0.0") == 0) {
struct hostent *hent;
xfree(ip); ip = NULL;
hent = gethostbyname(clientwalk->hostname);
if (hent) {
struct in_addr addr;
memcpy(&addr, *(hent->h_addr_list), sizeof(addr));
ip = strdup(inet_ntoa(addr));
}
}
if (!ip) continue;
/*
* Build the "pullclient" request, which includes the latest
* clientdata config we got from the server. Keep the clientdata
* here - we send "pullclient" requests more often that we actually
* contact the server, but we should provide the config data always.
*/
request = newstrbuffer(0);
sprintf(msgline, "pullclient %d\n", serverid);
addtobuffer(request, msgline);
if (clientwalk->clientdata) addtobuffer(request, clientwalk->clientdata);
/* Put the request on the connection queue */
addrequest(C_CLIENT, ip, port, request, clientwalk);
clientwalk->busy = 1;
xfree(ip);
}
}
/* Handle request queue */
FD_ZERO(&fdread);
FD_ZERO(&fdwrite);
maxfd = -1;
for (connwalk = chead; (connwalk); connwalk = connwalk->next) {
switch (connwalk->action) {
case C_READING:
FD_SET(connwalk->sockfd, &fdread);
if (connwalk->sockfd > maxfd) maxfd = connwalk->sockfd;
break;
case C_WRITING:
FD_SET(connwalk->sockfd, &fdwrite);
if (connwalk->sockfd > maxfd) maxfd = connwalk->sockfd;
break;
case C_CLEANUP:
break;
}
}
/* Do select with a 1 second timeout */
tmo.tv_sec = 1;
tmo.tv_usec = 0;
n = select(maxfd+1, &fdread, &fdwrite, NULL, &tmo);
if (n == -1) {
if (errno == EINTR) continue; /* Interrupted, e.g. a SIGHUP */
/* This is a "cannot-happen" failure. Bail out */
errprintf("select failure: %s\n", strerror(errno));
return 0;
}
if (n == 0) continue; /* Timeout */
for (connwalk = chead; (connwalk); connwalk = connwalk->next) {
switch (connwalk->action) {
case C_READING:
if (FD_ISSET(connwalk->sockfd, &fdread)) grabdata(connwalk);
break;
case C_WRITING:
if (FD_ISSET(connwalk->sockfd, &fdwrite)) senddata(connwalk);
break;
case C_CLEANUP:
break;
}
}
} while (running);
return 0;
}
int main(int argc, char *argv[])
{
tasklist_t *twalk, *dwalk;
grouplist_t *gwalk;
int argi;
int daemonize = 1;
int verbose = 0;
char *config = "/etc/tasks.cfg";
char *logfn = NULL;
char *pidfn = NULL;
pid_t cpid;
int status;
struct sigaction sa;
char *envarea = NULL;
for (argi=1; (argi < argc); argi++) {
if (strcmp(argv[argi], "--debug") == 0) {
debug = 1;
}
else if (strcmp(argv[argi], "--no-daemon") == 0) {
daemonize = 0;
}
else if (strcmp(argv[argi], "--verbose") == 0) {
verbose = 1;
}
else if (argnmatch(argv[argi], "--config=")) {
char *p = strchr(argv[argi], '=');
config = strdup(expand_env(p+1));
}
else if (argnmatch(argv[argi], "--log=")) {
char *p = strchr(argv[argi], '=');
logfn = strdup(expand_env(p+1));
}
else if (argnmatch(argv[argi], "--area=")) {
char *p = strchr(argv[argi], '=');
envarea = strdup(p+1);
}
else if (argnmatch(argv[argi], "--env=")) {
char *p = strchr(argv[argi], '=');
loadenv(p+1, envarea);
}
else if (argnmatch(argv[argi], "--pidfile=")) {
char *p = strchr(argv[argi], '=');
pidfn = strdup(expand_env(p+1));
}
else if (strcmp(argv[argi], "--dump") == 0) {
/* Dump configuration */
forcereload=1;
load_config(config);
forcereload=0;
for (gwalk = grouphead; (gwalk); gwalk = gwalk->next) {
if (gwalk->maxuse > 1) printf("GROUP %s %d\n", gwalk->groupname, gwalk->maxuse);
}
printf("\n");
for (twalk = taskhead; (twalk); twalk = twalk->next) {
printf("[%s]\n", twalk->key);
printf("\tCMD %s\n", twalk->cmd);
if (twalk->disabled) printf("\tDISABLED\n");
if (twalk->group) printf("\tGROUP %s\n", twalk->group->groupname);
if (twalk->depends) printf("\tNEEDS %s\n", twalk->depends->key);
if (twalk->interval > 0) printf("\tINTERVAL %d\n", twalk->interval);
if (twalk->cronstr) printf("\tCRONDATE %s\n", twalk->cronstr);
if (twalk->maxruntime) printf("\tMAXTIME %d\n", twalk->maxruntime);
if (twalk->logfile) printf("\tLOGFILE %s\n", twalk->logfile);
if (twalk->envfile) printf("\tENVFILE %s\n", twalk->envfile);
if (twalk->envarea) printf("\tENVAREA %s\n", twalk->envarea);
if (twalk->onhostptn) printf("\tONHOST %s\n", twalk->onhostptn);
printf("\n");
}
fflush(stdout);
return 0;
}
else {
fprintf(stderr,"%s: Unsupported argument: %s\n",argv[0],argv[argi]);
fflush(stderr);
return 1;
}
}
/* Go daemon */
if (daemonize) {
pid_t childpid;
/* Become a daemon */
childpid = fork();
if (childpid < 0) {
/* Fork failed */
errprintf("Could not fork child\n");
exit(1);
}
else if (childpid > 0) {
/* Parent exits */
if (pidfn) {
FILE *pidfd = fopen(pidfn, "w");
if (pidfd) {
fprintf(pidfd, "%d\n", (int)childpid);
fclose(pidfd);
}
}
exit(0);
}
/* Child (daemon) continues here */
setsid();
}
/* If using a logfile, switch stdout and stderr to go there */
if (logfn) {
/* Should we close stdin here ? No ... */
reopen_file("/dev/null", "r", stdin);
reopen_file(logfn, "a", stdout);
reopen_file(logfn, "a", stderr);
}
save_errbuf = 0;
setup_signalhandler("xymonlaunch");
memset(&sa, 0, sizeof(sa));
sa.sa_handler = sig_handler;
sigaction(SIGHUP, &sa, NULL);
sigaction(SIGTERM, &sa, NULL);
sigaction(SIGCHLD, &sa, NULL);
errprintf("xymonlaunch starting\n");
while (running) {
time_t now = gettimer();
if (now >= nextcfgload) {
load_config(config);
nextcfgload = (now + 30);
}
if (logfn && dologswitch) {
reopen_file(logfn, "a", stdout);
reopen_file(logfn, "a", stderr);
dologswitch = 0;
}
/* Pick up children that have terminated */
while ((cpid = wait3(&status, WNOHANG, NULL)) > 0) {
for (twalk = taskhead; (twalk && (twalk->pid != cpid)); twalk = twalk->next);
if (twalk) {
twalk->pid = 0;
twalk->beingkilled = 0;
if (WIFEXITED(status)) {
twalk->exitcode = WEXITSTATUS(status);
if (twalk->exitcode) {
errprintf("Task %s terminated, status %d\n", twalk->key, twalk->exitcode);
twalk->failcount++;
}
else {
twalk->failcount = 0;
}
}
else if (WIFSIGNALED(status)) {
twalk->exitcode = -WTERMSIG(status);
twalk->failcount++;
errprintf("Task %s terminated by signal %d\n", twalk->key, abs(twalk->exitcode));
}
if (twalk->group) twalk->group->currentuse--;
/* Tasks that depend on this task should be killed ... */
for (dwalk = taskhead; (dwalk); dwalk = dwalk->next) {
if ((dwalk->depends == twalk) && (dwalk->pid > 0)) {
dwalk->beingkilled = 1;
kill(dwalk->pid, SIGTERM);
}
}
}
}
/* See what new tasks need to get going */
dbgprintf("\n");
dbgprintf("Starting tasklist scan\n");
crongettime();
for (twalk = taskhead; (twalk); twalk = twalk->next) {
if ( (twalk->pid == 0) && !twalk->disabled &&
( ((twalk->interval >= 0) && (now >= (twalk->laststart + twalk->interval))) || /* xymon interval condition */
(twalk->crondate && ((twalk->laststart + 55) < now) && cronmatch(twalk->crondate)) /* cron date */
)
) {
if (twalk->depends && ((twalk->depends->pid == 0) || (twalk->depends->laststart > (now - 5)))) {
dbgprintf("Postponing start of %s due to %s not yet running\n",
twalk->key, twalk->depends->key);
continue;
}
if (twalk->group && (twalk->group->currentuse >= twalk->group->maxuse)) {
dbgprintf("Postponing start of %s due to group %s being busy\n",
twalk->key, twalk->group->groupname);
continue;
}
if ((twalk->failcount > MAX_FAILS) && ((twalk->laststart + 600) < now)) {
dbgprintf("Releasing %s from failure hold\n", twalk->key);
twalk->failcount = 0;
}
if (twalk->failcount > MAX_FAILS) {
dbgprintf("Postponing start of %s due to multiple failures\n", twalk->key);
continue;
}
if (twalk->laststart > (now - 5)) {
dbgprintf("Postponing start of %s, will not try more than once in 5 seconds\n", twalk->key);
continue;
}
dbgprintf("About to start task %s\n", twalk->key);
twalk->laststart = now;
twalk->pid = fork();
if (twalk->pid == 0) {
/* Exec the task */
char *cmd;
char **cmdargs = NULL;
static char tasksleepenv[20],bbsleepenv[20];
/* Setup environment */
if (twalk->envfile) {
dbgprintf("%s -> Loading environment from %s area %s\n",
twalk->key, expand_env(twalk->envfile),
(twalk->envarea ? twalk->envarea : ""));
loadenv(expand_env(twalk->envfile), twalk->envarea);
}
/* Setup TASKSLEEP to match the interval */
sprintf(tasksleepenv, "TASKSLEEP=%d", twalk->interval);
sprintf(bbsleepenv, "BBSLEEP=%d", twalk->interval); /* For compatibility */
putenv(tasksleepenv); putenv(bbsleepenv);
/* Setup command line and arguments */
cmdargs = setup_commandargs(twalk->cmd, &cmd);
/* Point stdout/stderr to a logfile, if requested */
if (twalk->logfile) {
char *logfn = expand_env(twalk->logfile);
dbgprintf("%s -> Assigning stdout/stderr to log '%s'\n", twalk->key, logfn);
reopen_file(logfn, "a", stdout);
reopen_file(logfn, "a", stderr);
}
/* Go! */
dbgprintf("%s -> Running '%s', XYMONHOME=%s\n", twalk->key, cmd, xgetenv("XYMONHOME"));
execvp(cmd, cmdargs);
/* Should never go here */
errprintf("Could not start task %s using command '%s': %s\n",
twalk->key, cmd, strerror(errno));
exit(0);
}
else if (twalk->pid == -1) {
/* Fork failed */
errprintf("Fork failed!\n");
twalk->pid = 0;
}
else {
if (twalk->group) twalk->group->currentuse++;
if (verbose) errprintf("Task %s started with PID %d\n", twalk->key, (int)twalk->pid);
}
}
else if (twalk->pid > 0) {
dbgprintf("Task %s active with PID %d\n", twalk->key, (int)twalk->pid);
if (twalk->maxruntime && ((now - twalk->laststart) > twalk->maxruntime)) {
errprintf("Killing hung task %s (PID %d) after %d seconds\n",
twalk->key, (int)twalk->pid,
(now - twalk->laststart));
kill(twalk->pid, (twalk->beingkilled ? SIGKILL : SIGTERM));
twalk->beingkilled = 1; /* Next time it's a real kill */
}
}
}
sleep(5);
}
/* Shutdown running tasks */
for (twalk = taskhead; (twalk); twalk = twalk->next) {
if (twalk->pid) kill(twalk->pid, SIGTERM);
}
if (pidfn) unlink(pidfn);
return 0;
}
int main(int argc, char *argv[])
{
time_t starttime = time(NULL);
char *histdir = NULL;
char *histlogdir = NULL;
char *msg;
int argi, seq;
int save_allevents = 1;
int save_hostevents = 1;
int save_statusevents = 1;
int save_histlogs = 1;
FILE *alleventsfd = NULL;
int running = 1;
struct sigaction sa;
char newcol2[3];
char oldcol2[3];
MEMDEFINE(newcol2);
MEMDEFINE(oldcol2);
/* Dont save the error buffer */
save_errbuf = 0;
if (xgetenv("BBALLHISTLOG")) save_allevents = (strcmp(xgetenv("BBALLHISTLOG"), "TRUE") == 0);
if (xgetenv("BBHOSTHISTLOG")) save_hostevents = (strcmp(xgetenv("BBHOSTHISTLOG"), "TRUE") == 0);
if (xgetenv("SAVESTATUSLOG")) save_histlogs = (strcmp(xgetenv("SAVESTATUSLOG"), "TRUE") == 0);
for (argi = 1; (argi < argc); argi++) {
if (argnmatch(argv[argi], "--histdir=")) {
histdir = strchr(argv[argi], '=')+1;
}
else if (argnmatch(argv[argi], "--histlogdir=")) {
histlogdir = strchr(argv[argi], '=')+1;
}
else if (argnmatch(argv[argi], "--debug")) {
debug = 1;
}
}
if (xgetenv("BBHIST") && (histdir == NULL)) {
histdir = strdup(xgetenv("BBHIST"));
}
if (histdir == NULL) {
errprintf("No history directory given, aborting\n");
return 1;
}
if (save_histlogs && (histlogdir == NULL) && xgetenv("BBHISTLOGS")) {
histlogdir = strdup(xgetenv("BBHISTLOGS"));
}
if (save_histlogs && (histlogdir == NULL)) {
errprintf("No history-log directory given, aborting\n");
return 1;
}
if (save_allevents) {
char alleventsfn[PATH_MAX];
MEMDEFINE(alleventsfn);
sprintf(alleventsfn, "%s/allevents", histdir);
alleventsfd = fopen(alleventsfn, "a");
if (alleventsfd == NULL) {
errprintf("Cannot open the all-events file '%s'\n", alleventsfn);
}
setvbuf(alleventsfd, (char *)NULL, _IOLBF, 0);
MEMUNDEFINE(alleventsfn);
}
/* For picking up lost children */
setup_signalhandler("hobbitd_history");
memset(&sa, 0, sizeof(sa));
sa.sa_handler = sig_handler;
sigaction(SIGCHLD, &sa, NULL);
signal(SIGPIPE, SIG_DFL);
while (running) {
char *metadata[20] = { NULL, };
int metacount;
char *p;
char *statusdata = "";
char *hostname, *hostnamecommas, *testname, *dismsg;
time_t tstamp, lastchg, disabletime, clienttstamp;
int tstamp_i, lastchg_i;
int newcolor, oldcolor;
int downtimeactive;
struct tm tstamptm;
int trend;
int childstat;
/* Pickup any finished child processes to avoid zombies */
while (wait3(&childstat, WNOHANG, NULL) > 0) ;
msg = get_hobbitd_message(C_STACHG, "hobbitd_history", &seq, NULL);
if (msg == NULL) {
running = 0;
continue;
}
p = strchr(msg, '\n');
if (p) {
*p = '\0';
statusdata = msg_data(p+1);
}
metacount = 0;
memset(&metadata, 0, sizeof(metadata));
p = gettok(msg, "|");
while (p && (metacount < 20)) {
metadata[metacount++] = p;
p = gettok(NULL, "|");
}
if ((metacount > 9) && (strncmp(metadata[0], "@@stachg", 8) == 0)) {
/* @@stachg#seq|timestamp|sender|origin|hostname|testname|expiretime|color|prevcolor|changetime|disabletime|disablemsg|downtimeactive|clienttstamp */
sscanf(metadata[1], "%d.%*d", &tstamp_i); tstamp = tstamp_i;
hostname = metadata[4];
testname = metadata[5];
newcolor = parse_color(metadata[7]);
oldcolor = parse_color(metadata[8]);
lastchg = atoi(metadata[9]);
disabletime = atoi(metadata[10]);
dismsg = metadata[11];
downtimeactive = (atoi(metadata[12]) > 0);
clienttstamp = atoi(metadata[13]);
if (newcolor == -1) {
errprintf("Bad message: newcolor is unknown '%s'\n", metadata[7]);
continue;
}
p = hostnamecommas = strdup(hostname); while ((p = strchr(p, '.')) != NULL) *p = ',';
if (save_statusevents) {
char statuslogfn[PATH_MAX];
int logexists;
FILE *statuslogfd;
char oldcol[100];
char timestamp[40];
struct stat st;
MEMDEFINE(statuslogfn);
MEMDEFINE(oldcol);
MEMDEFINE(timestamp);
sprintf(statuslogfn, "%s/%s.%s", histdir, hostnamecommas, testname);
stat(statuslogfn, &st);
statuslogfd = fopen(statuslogfn, "r+");
logexists = (statuslogfd != NULL);
*oldcol = '\0';
if (logexists) {
/*
* There is a fair chance hobbitd has not been
* running all the time while this system was monitored.
* So get the time of the latest status change from the file,
* instead of relying on the "lastchange" value we get
* from hobbitd. This is also needed when migrating from
* standard bbd to hobbitd.
*/
off_t pos = -1;
char l[1024];
int gotit;
MEMDEFINE(l);
fseeko(statuslogfd, 0, SEEK_END);
if (ftello(statuslogfd) > 512) {
/* Go back 512 from EOF, and skip to start of a line */
fseeko(statuslogfd, -512, SEEK_END);
gotit = (fgets(l, sizeof(l)-1, statuslogfd) == NULL);
}
else {
/* Read from beginning of file */
fseeko(statuslogfd, 0, SEEK_SET);
gotit = 0;
}
while (!gotit) {
off_t tmppos = ftello(statuslogfd);
time_t dur;
int dur_i;
if (fgets(l, sizeof(l)-1, statuslogfd)) {
/* Sun Oct 10 06:49:42 2004 red 1097383782 602 */
if ((strlen(l) > 24) &&
(sscanf(l+24, " %s %d %d", oldcol, &lastchg_i, &dur_i) == 2) &&
(parse_color(oldcol) != -1)) {
/*
* Record the start location of the line
*/
pos = tmppos;
lastchg = lastchg_i;
dur = dur_i;
}
}
else {
gotit = 1;
}
}
if (pos == -1) {
/*
* Couldnt find anything in the log.
* Take lastchg from the timestamp of the logfile,
* and just append the data.
*/
lastchg = st.st_mtime;
fseeko(statuslogfd, 0, SEEK_END);
}
else {
/*
* lastchg was updated above.
* Seek to where the last line starts.
*/
fseeko(statuslogfd, pos, SEEK_SET);
}
MEMUNDEFINE(l);
}
else {
/*
* Logfile does not exist.
*/
lastchg = tstamp;
statuslogfd = fopen(statuslogfn, "a");
if (statuslogfd == NULL) {
errprintf("Cannot open status historyfile '%s' : %s\n",
statuslogfn, strerror(errno));
}
}
if (strcmp(oldcol, colorname(newcolor)) == 0) {
/* We wont update history unless the color did change. */
if ((time(NULL) - starttime) > 300) {
errprintf("Will not update %s - color unchanged (%s)\n",
statuslogfn, oldcol);
}
if (hostnamecommas) xfree(hostnamecommas);
if (statuslogfd) fclose(statuslogfd);
MEMUNDEFINE(statuslogfn);
MEMUNDEFINE(oldcol);
MEMUNDEFINE(timestamp);
continue;
}
if (statuslogfd) {
if (logexists) {
struct tm oldtm;
/* Re-print the old record, now with the final duration */
memcpy(&oldtm, localtime(&lastchg), sizeof(oldtm));
strftime(timestamp, sizeof(timestamp), "%a %b %e %H:%M:%S %Y", &oldtm);
fprintf(statuslogfd, "%s %s %d %d\n",
timestamp, oldcol, (int)lastchg, (int)(tstamp - lastchg));
}
/* And the new record. */
memcpy(&tstamptm, localtime(&tstamp), sizeof(tstamptm));
strftime(timestamp, sizeof(timestamp), "%a %b %e %H:%M:%S %Y", &tstamptm);
fprintf(statuslogfd, "%s %s %d", timestamp, colorname(newcolor), (int)tstamp);
fclose(statuslogfd);
}
MEMUNDEFINE(statuslogfn);
MEMUNDEFINE(oldcol);
MEMUNDEFINE(timestamp);
}
if (save_histlogs) {
char *hostdash;
char fname[PATH_MAX];
FILE *histlogfd;
MEMDEFINE(fname);
p = hostdash = strdup(hostname); while ((p = strchr(p, '.')) != NULL) *p = '_';
sprintf(fname, "%s/%s", histlogdir, hostdash);
mkdir(fname, S_IRWXU|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH);
p = fname + sprintf(fname, "%s/%s/%s", histlogdir, hostdash, testname);
mkdir(fname, S_IRWXU|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH);
p += sprintf(p, "/%s", histlogtime(tstamp));
histlogfd = fopen(fname, "w");
if (histlogfd) {
/*
* When a host gets disabled or goes purple, the status
* message data is not changed - so it will include a
* wrong color as the first word of the message.
* Therefore we need to fixup this so it matches the
* newcolor value.
*/
int txtcolor = parse_color(statusdata);
char *origstatus = statusdata;
if (txtcolor != -1) {
fprintf(histlogfd, "%s", colorname(newcolor));
statusdata += strlen(colorname(txtcolor));
}
if (dismsg && *dismsg) nldecode(dismsg);
if (disabletime > 0) {
fprintf(histlogfd, " Disabled until %s\n%s\n\n",
ctime(&disabletime), (dismsg ? dismsg : ""));
fprintf(histlogfd, "Status message when disabled follows:\n\n");
statusdata = origstatus;
}
else if (dismsg && *dismsg) {
fprintf(histlogfd, " Planned downtime: %s\n\n", dismsg);
fprintf(histlogfd, "Original status message follows:\n\n");
statusdata = origstatus;
}
fwrite(statusdata, strlen(statusdata), 1, histlogfd);
fprintf(histlogfd, "Status unchanged in 0.00 minutes\n");
fprintf(histlogfd, "Message received from %s\n", metadata[2]);
if (clienttstamp) fprintf(histlogfd, "Client data ID %d\n", (int) clienttstamp);
fclose(histlogfd);
}
else {
errprintf("Cannot create histlog file '%s' : %s\n", fname, strerror(errno));
}
xfree(hostdash);
MEMUNDEFINE(fname);
}
strncpy(oldcol2, ((oldcolor >= 0) ? colorname(oldcolor) : "-"), 2);
strncpy(newcol2, colorname(newcolor), 2);
newcol2[2] = oldcol2[2] = '\0';
if (oldcolor == -1) trend = -1; /* we dont know how bad it was */
else if (newcolor > oldcolor) trend = 2; /* It's getting worse */
else if (newcolor < oldcolor) trend = 1; /* It's getting better */
else trend = 0; /* Shouldn't happen ... */
if (save_hostevents) {
char hostlogfn[PATH_MAX];
FILE *hostlogfd;
MEMDEFINE(hostlogfn);
sprintf(hostlogfn, "%s/%s", histdir, hostname);
hostlogfd = fopen(hostlogfn, "a");
if (hostlogfd) {
fprintf(hostlogfd, "%s %d %d %d %s %s %d\n",
testname, (int)tstamp, (int)lastchg, (int)(tstamp - lastchg),
newcol2, oldcol2, trend);
fclose(hostlogfd);
}
else {
errprintf("Cannot open host logfile '%s' : %s\n", hostlogfn, strerror(errno));
}
MEMUNDEFINE(hostlogfn);
}
if (save_allevents) {
fprintf(alleventsfd, "%s %s %d %d %d %s %s %d\n",
hostname, testname, (int)tstamp, (int)lastchg, (int)(tstamp - lastchg),
newcol2, oldcol2, trend);
fflush(alleventsfd);
}
xfree(hostnamecommas);
}
else if ((metacount > 3) && ((strncmp(metadata[0], "@@drophost", 10) == 0))) {
/* @@drophost|timestamp|sender|hostname */
hostname = metadata[3];
if (save_histlogs) {
char *hostdash;
char testdir[PATH_MAX];
MEMDEFINE(testdir);
/* Remove all directories below the host-specific histlog dir */
p = hostdash = strdup(hostname); while ((p = strchr(p, '.')) != NULL) *p = '_';
sprintf(testdir, "%s/%s", histlogdir, hostdash);
dropdirectory(testdir, 1);
xfree(hostdash);
MEMUNDEFINE(testdir);
}
if (save_hostevents) {
char hostlogfn[PATH_MAX];
struct stat st;
MEMDEFINE(hostlogfn);
sprintf(hostlogfn, "%s/%s", histdir, hostname);
if ((stat(hostlogfn, &st) == 0) && S_ISREG(st.st_mode)) {
unlink(hostlogfn);
}
MEMUNDEFINE(hostlogfn);
}
if (save_statusevents) {
DIR *dirfd;
struct dirent *de;
char *hostlead;
char statuslogfn[PATH_MAX];
struct stat st;
MEMDEFINE(statuslogfn);
/* Remove bbvar/hist/host,name.* */
p = hostnamecommas = strdup(hostname); while ((p = strchr(p, '.')) != NULL) *p = ',';
hostlead = malloc(strlen(hostname) + 2);
strcpy(hostlead, hostnamecommas); strcat(hostlead, ".");
dirfd = opendir(histdir);
if (dirfd) {
while ((de = readdir(dirfd)) != NULL) {
if (strncmp(de->d_name, hostlead, strlen(hostlead)) == 0) {
sprintf(statuslogfn, "%s/%s", histdir, de->d_name);
if ((stat(statuslogfn, &st) == 0) && S_ISREG(st.st_mode)) {
unlink(statuslogfn);
}
}
}
closedir(dirfd);
}
xfree(hostlead);
xfree(hostnamecommas);
MEMUNDEFINE(statuslogfn);
}
}
else if ((metacount > 4) && ((strncmp(metadata[0], "@@droptest", 10) == 0))) {
/* @@droptest|timestamp|sender|hostname|testname */
hostname = metadata[3];
testname = metadata[4];
if (save_histlogs) {
char *hostdash;
char testdir[PATH_MAX];
MEMDEFINE(testdir);
p = hostdash = strdup(hostname); while ((p = strchr(p, '.')) != NULL) *p = '_';
sprintf(testdir, "%s/%s/%s", histlogdir, hostdash, testname);
dropdirectory(testdir, 1);
xfree(hostdash);
MEMUNDEFINE(testdir);
}
if (save_statusevents) {
char *hostnamecommas;
char statuslogfn[PATH_MAX];
struct stat st;
MEMDEFINE(statuslogfn);
p = hostnamecommas = strdup(hostname); while ((p = strchr(p, '.')) != NULL) *p = ',';
sprintf(statuslogfn, "%s/%s.%s", histdir, hostnamecommas, testname);
if ((stat(statuslogfn, &st) == 0) && S_ISREG(st.st_mode)) unlink(statuslogfn);
xfree(hostnamecommas);
MEMUNDEFINE(statuslogfn);
}
}
else if ((metacount > 4) && ((strncmp(metadata[0], "@@renamehost", 12) == 0))) {
/* @@renamehost|timestamp|sender|hostname|newhostname */
char *newhostname;
hostname = metadata[3];
newhostname = metadata[4];
if (save_histlogs) {
char *hostdash;
char *newhostdash;
char olddir[PATH_MAX];
char newdir[PATH_MAX];
MEMDEFINE(olddir); MEMDEFINE(newdir);
p = hostdash = strdup(hostname); while ((p = strchr(p, '.')) != NULL) *p = '_';
p = newhostdash = strdup(newhostname); while ((p = strchr(p, '.')) != NULL) *p = '_';
sprintf(olddir, "%s/%s", histlogdir, hostdash);
sprintf(newdir, "%s/%s", histlogdir, newhostdash);
rename(olddir, newdir);
xfree(newhostdash);
xfree(hostdash);
MEMUNDEFINE(newdir); MEMUNDEFINE(olddir);
}
if (save_hostevents) {
char hostlogfn[PATH_MAX];
char newhostlogfn[PATH_MAX];
MEMDEFINE(hostlogfn); MEMDEFINE(newhostlogfn);
sprintf(hostlogfn, "%s/%s", histdir, hostname);
sprintf(newhostlogfn, "%s/%s", histdir, newhostname);
rename(hostlogfn, newhostlogfn);
MEMUNDEFINE(hostlogfn); MEMUNDEFINE(newhostlogfn);
}
if (save_statusevents) {
DIR *dirfd;
struct dirent *de;
char *hostlead;
char *newhostnamecommas;
char statuslogfn[PATH_MAX];
char newlogfn[PATH_MAX];
MEMDEFINE(statuslogfn); MEMDEFINE(newlogfn);
p = hostnamecommas = strdup(hostname); while ((p = strchr(p, '.')) != NULL) *p = ',';
hostlead = malloc(strlen(hostname) + 2);
strcpy(hostlead, hostnamecommas); strcat(hostlead, ".");
p = newhostnamecommas = strdup(newhostname); while ((p = strchr(p, '.')) != NULL) *p = ',';
dirfd = opendir(histdir);
if (dirfd) {
while ((de = readdir(dirfd)) != NULL) {
if (strncmp(de->d_name, hostlead, strlen(hostlead)) == 0) {
char *testname = strchr(de->d_name, '.');
sprintf(statuslogfn, "%s/%s", histdir, de->d_name);
sprintf(newlogfn, "%s/%s%s", histdir, newhostnamecommas, testname);
rename(statuslogfn, newlogfn);
}
}
closedir(dirfd);
}
xfree(newhostnamecommas);
xfree(hostlead);
xfree(hostnamecommas);
MEMUNDEFINE(statuslogfn); MEMUNDEFINE(newlogfn);
}
}
else if ((metacount > 5) && (strncmp(metadata[0], "@@renametest", 12) == 0)) {
/* @@renametest|timestamp|sender|hostname|oldtestname|newtestname */
char *newtestname;
hostname = metadata[3];
testname = metadata[4];
newtestname = metadata[5];
if (save_histlogs) {
char *hostdash;
char olddir[PATH_MAX];
char newdir[PATH_MAX];
MEMDEFINE(olddir); MEMDEFINE(newdir);
p = hostdash = strdup(hostname); while ((p = strchr(p, '.')) != NULL) *p = '_';
sprintf(olddir, "%s/%s/%s", histlogdir, hostdash, testname);
sprintf(newdir, "%s/%s/%s", histlogdir, hostdash, newtestname);
rename(olddir, newdir);
xfree(hostdash);
MEMUNDEFINE(newdir); MEMUNDEFINE(olddir);
}
if (save_statusevents) {
char *hostnamecommas;
char statuslogfn[PATH_MAX];
char newstatuslogfn[PATH_MAX];
MEMDEFINE(statuslogfn); MEMDEFINE(newstatuslogfn);
p = hostnamecommas = strdup(hostname); while ((p = strchr(p, '.')) != NULL) *p = ',';
sprintf(statuslogfn, "%s/%s.%s", histdir, hostnamecommas, testname);
sprintf(newstatuslogfn, "%s/%s.%s", histdir, hostnamecommas, newtestname);
rename(statuslogfn, newstatuslogfn);
xfree(hostnamecommas);
MEMUNDEFINE(newstatuslogfn); MEMUNDEFINE(statuslogfn);
}
}
else if (strncmp(metadata[0], "@@shutdown", 10) == 0) {
running = 0;
}
else if (strncmp(metadata[0], "@@logrotate", 11) == 0) {
char *fn = xgetenv("HOBBITCHANNEL_LOGFILENAME");
if (fn && strlen(fn)) {
freopen(fn, "a", stdout);
freopen(fn, "a", stderr);
}
continue;
}
}
MEMUNDEFINE(newcol2);
MEMUNDEFINE(oldcol2);
fclose(alleventsfd);
return 0;
}
int main(int argc, char *argv[])
{
int daemonize = 0;
char *pidfile = NULL;
char *envarea = NULL;
int cnid = -1;
pcre *msgfilter = NULL;
pcre *stdfilter = NULL;
int argi;
struct sigaction sa;
RbtIterator handle;
/* Dont save the error buffer */
save_errbuf = 0;
/* Create the peer container */
peers = rbtNew(name_compare);
for (argi=1; (argi < argc); argi++) {
if (argnmatch(argv[argi], "--debug")) {
debug = 1;
}
else if (argnmatch(argv[argi], "--channel=")) {
char *cn = strchr(argv[argi], '=') + 1;
for (cnid = C_STATUS; (channelnames[cnid] && strcmp(channelnames[cnid], cn)); cnid++) ;
if (channelnames[cnid] == NULL) cnid = -1;
}
else if (argnmatch(argv[argi], "--daemon")) {
daemonize = 1;
}
else if (argnmatch(argv[argi], "--no-daemon")) {
daemonize = 0;
}
else if (argnmatch(argv[argi], "--pidfile=")) {
char *p = strchr(argv[argi], '=');
pidfile = strdup(p+1);
}
else if (argnmatch(argv[argi], "--log=")) {
char *p = strchr(argv[argi], '=');
logfn = strdup(p+1);
}
else if (argnmatch(argv[argi], "--env=")) {
char *p = strchr(argv[argi], '=');
loadenv(p+1, envarea);
}
else if (argnmatch(argv[argi], "--area=")) {
char *p = strchr(argv[argi], '=');
envarea = strdup(p+1);
}
else if (argnmatch(argv[argi], "--locator=")) {
char *p = strchr(argv[argi], '=');
locator_init(p+1);
locatorbased = 1;
}
else if (argnmatch(argv[argi], "--service=")) {
char *p = strchr(argv[argi], '=');
locatorservice = get_servicetype(p+1);
}
else if (argnmatch(argv[argi], "--filter=")) {
char *p = strchr(argv[argi], '=');
msgfilter = compileregex(p+1);
if (!msgfilter) {
errprintf("Invalid filter (bad expression): %s\n", p+1);
}
else {
stdfilter = compileregex("^@@(logrotate|shutdown|drophost|droptest|renamehost|renametest)");
}
}
else {
char *childcmd;
char **childargs;
int i = 0;
childcmd = argv[argi];
childargs = (char **) calloc((1 + argc - argi), sizeof(char *));
while (argi < argc) { childargs[i++] = argv[argi++]; }
addlocalpeer(childcmd, childargs);
}
}
/* Sanity checks */
if (cnid == -1) {
errprintf("No channel/unknown channel specified\n");
return 1;
}
if (locatorbased && (locatorservice == ST_MAX)) {
errprintf("Must specify --service when using locator\n");
return 1;
}
if (!locatorbased && (rbtBegin(peers) == rbtEnd(peers))) {
errprintf("Must specify command for local worker\n");
return 1;
}
/* Do cache responses to avoid doing too many lookups */
if (locatorbased) locator_prepcache(locatorservice, 0);
/* Go daemon */
if (daemonize) {
/* Become a daemon */
pid_t daemonpid = fork();
if (daemonpid < 0) {
/* Fork failed */
errprintf("Could not fork child\n");
exit(1);
}
else if (daemonpid > 0) {
/* Parent creates PID file and exits */
FILE *fd = NULL;
if (pidfile) fd = fopen(pidfile, "w");
if (fd) {
fprintf(fd, "%d\n", (int)daemonpid);
fclose(fd);
}
exit(0);
}
/* Child (daemon) continues here */
setsid();
}
/* Catch signals */
setup_signalhandler("xymond_channel");
memset(&sa, 0, sizeof(sa));
sa.sa_handler = sig_handler;
sigaction(SIGINT, &sa, NULL);
sigaction(SIGTERM, &sa, NULL);
sigaction(SIGCHLD, &sa, NULL);
signal(SIGALRM, SIG_IGN);
/* Switch stdout/stderr to the logfile, if one was specified */
freopen("/dev/null", "r", stdin); /* xymond_channel's stdin is not used */
if (logfn) {
freopen(logfn, "a", stdout);
freopen(logfn, "a", stderr);
}
/* Attach to the channel */
channel = setup_channel(cnid, CHAN_CLIENT);
if (channel == NULL) {
errprintf("Channel not available\n");
running = 0;
}
while (running) {
/*
* Wait for GOCLIENT to go up.
*
* Note that we use IPC_NOWAIT if there are messages in the
* queue, because then we just want to pick up a message if
* there is one, and if not we want to continue pushing the
* queued data to the worker.
*/
struct sembuf s;
int n;
s.sem_num = GOCLIENT; s.sem_op = -1; s.sem_flg = ((pendingcount > 0) ? IPC_NOWAIT : 0);
n = semop(channel->semid, &s, 1);
if (n == 0) {
/*
* GOCLIENT went high, and so we got alerted about a new
* message arriving. Copy the message to our own buffer queue.
*/
char *inbuf = NULL;
if (!msgfilter || matchregex(channel->channelbuf, msgfilter) || matchregex(channel->channelbuf, stdfilter)) {
inbuf = strdup(channel->channelbuf);
}
/*
* Now we have safely stored the new message in our buffer.
* Wait until any other clients on the same channel have picked up
* this message (GOCLIENT reaches 0).
*
* We wrap this into an alarm handler, because it can occasionally
* fail, causing the whole system to lock up. We dont want that....
* We'll set the alarm to trigger after 1 second. Experience shows
* that we'll either succeed in a few milliseconds, or fail completely
* and wait the full alarm-timer duration.
*/
gotalarm = 0; signal(SIGALRM, sig_handler); alarm(2);
do {
s.sem_num = GOCLIENT; s.sem_op = 0; s.sem_flg = 0;
n = semop(channel->semid, &s, 1);
} while ((n == -1) && (errno == EAGAIN) && running && (!gotalarm));
signal(SIGALRM, SIG_IGN);
if (gotalarm) {
errprintf("Gave up waiting for GOCLIENT to go low.\n");
}
/*
* Let master know we got it by downing BOARDBUSY.
* This should not block, since BOARDBUSY is upped
* by the master just before he ups GOCLIENT.
*/
do {
s.sem_num = BOARDBUSY; s.sem_op = -1; s.sem_flg = IPC_NOWAIT;
n = semop(channel->semid, &s, 1);
} while ((n == -1) && (errno == EINTR));
if (n == -1) {
errprintf("Tried to down BOARDBUSY: %s\n", strerror(errno));
}
if (inbuf) {
/*
* See if they want us to rotate logs. We pass this on to
* the worker module as well, but must handle our own logfile.
*/
if (strncmp(inbuf, "@@logrotate", 11) == 0) {
freopen(logfn, "a", stdout);
freopen(logfn, "a", stderr);
}
/*
* Put the new message on our outbound queue.
*/
if (addmessage(inbuf) != 0) {
/* Failed to queue message, free the buffer */
xfree(inbuf);
}
}
}
else {
if (errno != EAGAIN) {
dbgprintf("Semaphore wait aborted: %s\n", strerror(errno));
continue;
}
}
/*
* We've picked up messages from the master. Now we
* must push them to the worker process. Since there
* is no way to hang off both a semaphore and select(),
* we'll just push as much data as possible into the
* pipe. If we get to a point where we would block,
* then wait a teeny bit of time and restart the
* whole loop with checking for new messages from the
* master etc.
*
* In theory, this could become an almost busy-wait loop.
* In practice, however, the queue will be empty most
* of the time because we'll just shove the data to the
* worker child.
*/
for (handle = rbtBegin(peers); (handle != rbtEnd(peers)); handle = rbtNext(peers, handle)) {
int canwrite = 1, hasfailed = 0;
xymon_peer_t *pwalk;
time_t msgtimeout = gettimer() - MSGTIMEOUT;
int flushcount = 0;
pwalk = (xymon_peer_t *) gettreeitem(peers, handle);
if (pwalk->msghead == NULL) continue; /* Ignore peers with nothing queued */
switch (pwalk->peerstatus) {
case P_UP:
canwrite = 1;
break;
case P_DOWN:
openconnection(pwalk);
canwrite = (pwalk->peerstatus == P_UP);
break;
case P_FAILED:
canwrite = 0;
break;
}
/* See if we have stale messages queued */
while (pwalk->msghead && (pwalk->msghead->tstamp < msgtimeout)) {
flushmessage(pwalk);
flushcount++;
}
if (flushcount) {
errprintf("Flushed %d stale messages for %s:%d\n",
flushcount,
inet_ntoa(pwalk->peeraddr.sin_addr),
ntohs(pwalk->peeraddr.sin_port));
}
while (pwalk->msghead && canwrite) {
fd_set fdwrite;
struct timeval tmo;
/* Check that this peer is ready for writing. */
FD_ZERO(&fdwrite); FD_SET(pwalk->peersocket, &fdwrite);
tmo.tv_sec = 0; tmo.tv_usec = 2000;
n = select(pwalk->peersocket+1, NULL, &fdwrite, NULL, &tmo);
if (n == -1) {
errprintf("select() failed: %s\n", strerror(errno));
canwrite = 0;
hasfailed = 1;
continue;
}
else if ((n == 0) || (!FD_ISSET(pwalk->peersocket, &fdwrite))) {
canwrite = 0;
continue;
}
n = write(pwalk->peersocket, pwalk->msghead->bufp, pwalk->msghead->buflen);
if (n >= 0) {
pwalk->msghead->bufp += n;
pwalk->msghead->buflen -= n;
if (pwalk->msghead->buflen == 0) flushmessage(pwalk);
}
else if (errno == EAGAIN) {
/*
* Write would block ... stop for now.
*/
canwrite = 0;
}
else {
hasfailed = 1;
}
if (hasfailed) {
/* Write failed, or message grew stale */
errprintf("Peer at %s:%d failed: %s\n",
inet_ntoa(pwalk->peeraddr.sin_addr), ntohs(pwalk->peeraddr.sin_port),
strerror(errno));
canwrite = 0;
shutdownconnection(pwalk);
if (pwalk->peertype == P_NET) locator_serverdown(pwalk->peername, locatorservice);
pwalk->peerstatus = P_FAILED;
}
}
}
}
/* Detach from channels */
close_channel(channel, CHAN_CLIENT);
/* Close peer connections */
for (handle = rbtBegin(peers); (handle != rbtEnd(peers)); handle = rbtNext(peers, handle)) {
xymon_peer_t *pwalk = (xymon_peer_t *) gettreeitem(peers, handle);
shutdownconnection(pwalk);
}
/* Remove the PID file */
if (pidfile) unlink(pidfile);
return 0;
}
int main(int argc, char *argv[])
{
char *filedir = NULL;
char *htmldir = NULL;
char *htmlextension = "html";
char *onlytests = NULL;
char *msg;
enum role_t role = ROLE_STATUS;
enum msgchannels_t chnid = C_STATUS;
int argi;
int seq;
int running = 1;
/* Dont save the error buffer */
save_errbuf = 0;
for (argi = 1; (argi < argc); argi++) {
if (strcmp(argv[argi], "--status") == 0) {
role = ROLE_STATUS;
chnid = C_STATUS;
if (!filedir) filedir = xgetenv("XYMONRAWSTATUSDIR");
}
else if (strcmp(argv[argi], "--html") == 0) {
role = ROLE_STATUS;
chnid = C_STATUS;
if (!htmldir) htmldir = xgetenv("XYMONHTMLSTATUSDIR");
}
else if (strcmp(argv[argi], "--data") == 0) {
role = ROLE_DATA;
chnid = C_DATA;
if (!filedir) filedir = xgetenv("XYMONDATADIR");
}
else if (strcmp(argv[argi], "--notes") == 0) {
role = ROLE_NOTES;
chnid = C_NOTES;
if (!filedir) filedir = xgetenv("XYMONNOTESDIR");
}
else if (strcmp(argv[argi], "--enadis") == 0) {
role = ROLE_ENADIS;
chnid = C_ENADIS;
if (!filedir) filedir = xgetenv("XYMONDISABLEDDIR");
}
else if (strcmp(argv[argi], "--debug") == 0) {
debug = 1;
}
else if (argnmatch(argv[argi], "--dir=")) {
filedir = strchr(argv[argi], '=')+1;
}
else if (argnmatch(argv[argi], "--htmldir=")) {
htmldir = strchr(argv[argi], '=')+1;
}
else if (argnmatch(argv[argi], "--htmlext=")) {
htmlextension = strchr(argv[argi], '=')+1;
}
else if (argnmatch(argv[argi], "--only=")) {
char *p = strchr(argv[argi], '=') + 1;
onlytests = (char *)malloc(3 + strlen(p));
sprintf(onlytests, ",%s,", p);
}
else if (argnmatch(argv[argi], "--multigraphs=")) {
char *p = strchr(argv[argi], '=');
multigraphs = (char *)malloc(strlen(p+1) + 3);
sprintf(multigraphs, ",%s,", p+1);
}
else if (argnmatch(argv[argi], "--locator=")) {
char *p = strchr(argv[argi], '=');
locator_init(p+1);
locatorbased = 1;
}
}
if (filedir == NULL) {
errprintf("No directory given, aborting\n");
return 1;
}
/* For picking up lost children */
setup_signalhandler("xymond_filestore");
signal(SIGPIPE, SIG_DFL);
if (onlytests) dbgprintf("Storing tests '%s' only\n", onlytests);
else dbgprintf("Storing all tests\n");
while (running) {
char *metadata[20] = { NULL, };
char *statusdata = "";
char *p;
int metacount;
char *hostname, *testname;
time_t expiretime = 0;
char logfn[PATH_MAX];
MEMDEFINE(logfn);
msg = get_xymond_message(chnid, "filestore", &seq, NULL);
if (msg == NULL) {
running = 0;
MEMUNDEFINE(logfn);
continue;
}
p = strchr(msg, '\n');
if (p) {
*p = '\0';
statusdata = p+1;
}
metacount = 0;
memset(&metadata, 0, sizeof(metadata));
p = gettok(msg, "|");
while (p && (metacount < 20)) {
metadata[metacount++] = p;
p = gettok(NULL, "|");
}
if ((role == ROLE_STATUS) && (metacount >= 14) && (strncmp(metadata[0], "@@status", 8) == 0)) {
/* @@status|timestamp|sender|origin|hostname|testname|expiretime|color|testflags|prevcolor|changetime|ackexpiretime|ackmessage|disableexpiretime|disablemessage|clientmsgtstamp|flapping */
int ltime, flapping = 0;
time_t logtime = 0, timesincechange = 0, acktime = 0, disabletime = 0;
hostname = metadata[4];
testname = metadata[5];
if (!wantedtest(onlytests, testname)) {
dbgprintf("Status dropped - not wanted\n");
MEMUNDEFINE(logfn);
continue;
}
sprintf(logfn, "%s/%s.%s", filedir, commafy(hostname), testname);
expiretime = atoi(metadata[6]);
statusdata = msg_data(statusdata);
sscanf(metadata[1], "%d.%*d", <ime); logtime = ltime;
timesincechange = logtime - atoi(metadata[10]);
update_file(logfn, "w", statusdata, expiretime, metadata[2], timesincechange, seq);
if (htmldir) {
char *ackmsg = NULL;
char *dismsg = NULL;
char htmllogfn[PATH_MAX];
MEMDEFINE(htmllogfn);
if (metadata[11]) acktime = atoi(metadata[11]);
if (metadata[12] && strlen(metadata[12])) ackmsg = metadata[12];
if (ackmsg) nldecode(ackmsg);
if (metadata[13]) disabletime = atoi(metadata[13]);
if (metadata[14] && strlen(metadata[14]) && (disabletime > 0)) dismsg = metadata[14];
if (dismsg) nldecode(dismsg);
flapping = (metadata[16] ? (*metadata[16] == '1') : 0);
sprintf(htmllogfn, "%s/%s.%s.%s", htmldir, hostname, testname, htmlextension);
update_htmlfile(htmllogfn, statusdata, hostname, testname, parse_color(metadata[7]), flapping,
metadata[2], metadata[8], logtime, timesincechange,
acktime, ackmsg,
disabletime, dismsg);
MEMUNDEFINE(htmllogfn);
}
}
else if ((role == ROLE_DATA) && (metacount > 5) && (strncmp(metadata[0], "@@data", 6) == 0)) {
/* @@data|timestamp|sender|hostname|testname */
p = hostname = metadata[4]; while ((p = strchr(p, '.')) != NULL) *p = ',';
testname = metadata[5];
if (!wantedtest(onlytests, testname)) {
dbgprintf("data dropped - not wanted\n");
MEMUNDEFINE(logfn);
continue;
}
statusdata = msg_data(statusdata); if (*statusdata == '\n') statusdata++;
sprintf(logfn, "%s/%s.%s", filedir, hostname, testname);
expiretime = 0;
update_file(logfn, "a", statusdata, expiretime, NULL, -1, seq);
}
else if ((role == ROLE_NOTES) && (metacount > 3) && (strncmp(metadata[0], "@@notes", 7) == 0)) {
/* @@notes|timestamp|sender|hostname */
hostname = metadata[3];
statusdata = msg_data(statusdata); if (*statusdata == '\n') statusdata++;
sprintf(logfn, "%s/%s", basename(filedir), hostname);
expiretime = 0;
update_file(logfn, "w", statusdata, expiretime, NULL, -1, seq);
}
else if ((role == ROLE_ENADIS) && (metacount > 5) && (strncmp(metadata[0], "@@enadis", 8) == 0)) {
/* @@enadis|timestamp|sender|hostname|testname|expiretime */
p = hostname = metadata[3]; while ((p = strchr(p, '.')) != NULL) *p = ',';
testname = metadata[4];
expiretime = atoi(metadata[5]);
sprintf(logfn, "%s/%s.%s", filedir, hostname, testname);
update_enable(logfn, expiretime);
}
else if (((role == ROLE_STATUS) || (role == ROLE_DATA) || (role == ROLE_ENADIS)) && (metacount > 3) && (strncmp(metadata[0], "@@drophost", 10) == 0)) {
/* @@drophost|timestamp|sender|hostname */
DIR *dirfd;
struct dirent *de;
char *hostlead;
p = hostname = metadata[3]; while ((p = strchr(p, '.')) != NULL) *p = ',';
hostlead = malloc(strlen(hostname) + 2);
strcpy(hostlead, hostname); strcat(hostlead, ".");
dirfd = opendir(filedir);
if (dirfd) {
while ( (de = readdir(dirfd)) != NULL) {
if (strncmp(de->d_name, hostlead, strlen(hostlead)) == 0) {
sprintf(logfn, "%s/%s", filedir, de->d_name);
unlink(logfn);
}
}
closedir(dirfd);
}
xfree(hostlead);
}
else if (((role == ROLE_STATUS) || (role == ROLE_DATA) || (role == ROLE_ENADIS)) && (metacount > 4) && (strncmp(metadata[0], "@@droptest", 10) == 0)) {
/* @@droptest|timestamp|sender|hostname|testname */
p = hostname = metadata[3]; while ((p = strchr(p, '.')) != NULL) *p = ',';
testname = metadata[4];
sprintf(logfn, "%s/%s.%s", filedir, hostname, testname);
unlink(logfn);
}
else if (((role == ROLE_STATUS) || (role == ROLE_DATA) || (role == ROLE_ENADIS)) && (metacount > 4) && (strncmp(metadata[0], "@@renamehost", 12) == 0)) {
/* @@renamehost|timestamp|sender|hostname|newhostname */
DIR *dirfd;
struct dirent *de;
char *hostlead;
char *newhostname;
char newlogfn[PATH_MAX];
MEMDEFINE(newlogfn);
p = hostname = metadata[3]; while ((p = strchr(p, '.')) != NULL) *p = ',';
hostlead = malloc(strlen(hostname) + 2);
strcpy(hostlead, hostname); strcat(hostlead, ".");
p = newhostname = metadata[4]; while ((p = strchr(p, '.')) != NULL) *p = ',';
dirfd = opendir(filedir);
if (dirfd) {
while ( (de = readdir(dirfd)) != NULL) {
if (strncmp(de->d_name, hostlead, strlen(hostlead)) == 0) {
char *testname = strchr(de->d_name, '.');
sprintf(logfn, "%s/%s", filedir, de->d_name);
sprintf(newlogfn, "%s/%s%s", filedir, newhostname, testname);
rename(logfn, newlogfn);
}
}
closedir(dirfd);
}
xfree(hostlead);
MEMUNDEFINE(newlogfn);
}
else if (((role == ROLE_STATUS) || (role == ROLE_DATA) || (role == ROLE_ENADIS)) && (metacount > 5) && (strncmp(metadata[0], "@@renametest", 12) == 0)) {
/* @@renametest|timestamp|sender|hostname|oldtestname|newtestname */
char *newtestname;
char newfn[PATH_MAX];
MEMDEFINE(newfn);
p = hostname = metadata[3]; while ((p = strchr(p, '.')) != NULL) *p = ',';
testname = metadata[4];
newtestname = metadata[5];
sprintf(logfn, "%s/%s.%s", filedir, hostname, testname);
sprintf(newfn, "%s/%s.%s", filedir, hostname, newtestname);
rename(logfn, newfn);
MEMUNDEFINE(newfn);
}
else if (strncmp(metadata[0], "@@shutdown", 10) == 0) {
running = 0;
}
else if (strncmp(metadata[0], "@@logrotate", 11) == 0) {
char *fn = xgetenv("XYMONCHANNEL_LOGFILENAME");
if (fn && strlen(fn)) {
freopen(fn, "a", stdout);
freopen(fn, "a", stderr);
}
continue;
}
else if (strncmp(metadata[0], "@@idle", 6) == 0) {
/* Ignored */
}
else {
errprintf("Dropping message type %s, metacount=%d\n", metadata[0], metacount);
}
MEMUNDEFINE(logfn);
}
return 0;
}
/**
* Main program
* @param argc argument counter
* @param argv argument vector
* @param appdata application data
* @return application return code to shell
*/
int
oonf_main(int argc, char **argv, const struct oonf_appdata *appdata) {
int return_code;
int result;
/* early initialization */
return_code = 1;
_schema_name = NULL;
_display_schema = false;
_debug_early = false;
_ignore_unknown = false;
/* setup signal handler */
_end_oonf_signal = false;
setup_signalhandler();
/* parse "early" command line arguments */
parse_early_commandline(argc, argv);
/* initialize core */
os_core_init(appdata->app_name);
/* initialize logger */
if (oonf_log_init(appdata, _debug_early ? LOG_SEVERITY_DEBUG : LOG_SEVERITY_WARN)) {
goto oonf_cleanup;
}
/* prepare plugin initialization */
oonf_subsystem_init();
/* initialize configuration system */
if (oonf_cfg_init(argc, argv, appdata->default_cfg_handler)) {
goto oonf_cleanup;
}
/* add custom configuration definitions */
oonf_logcfg_init();
/* parse command line and read configuration files */
return_code = parse_commandline(argc, argv, appdata, false);
if (return_code != -1) {
/* end OONFd now */
goto oonf_cleanup;
}
/* prepare for an error during initialization */
return_code = 1;
/* read global section early */
if ((result = oonf_cfg_update_globalcfg(true))) {
OONF_WARN(LOG_MAIN, "Cannot read global configuration section (%d)", result);
goto oonf_cleanup;
}
/* configure logger */
if (oonf_logcfg_apply(oonf_cfg_get_rawdb())) {
goto oonf_cleanup;
}
/* load plugins */
if (oonf_cfg_load_subsystems()) {
goto oonf_cleanup;
}
/* show schema if necessary */
if (_display_schema) {
return_code = display_schema();
goto oonf_cleanup;
}
/* check if we are root, otherwise stop */
if (appdata->need_root) {
if (geteuid() != 0) {
OONF_WARN(LOG_MAIN, "You must be root(uid = 0) to run %s!\n",
appdata->app_name);
goto oonf_cleanup;
}
}
if (appdata->need_lock
&& config_global.lockfile != NULL && *config_global.lockfile != 0) {
/* create application lock */
if (os_core_create_lockfile(config_global.lockfile)) {
OONF_WARN(LOG_MAIN, "Could not acquire application lock '%s'",
config_global.lockfile);
goto oonf_cleanup;
}
}
/* call initialization callbacks of dynamic plugins */
oonf_cfg_initplugins();
/* apply configuration */
if (oonf_cfg_apply()) {
goto oonf_cleanup;
}
if (!oonf_cfg_is_running()) {
/*
* mayor error during late initialization
* or maybe the user decided otherwise and pressed CTRL-C
*/
return_code = _end_oonf_signal ? 0 : 1;
goto oonf_cleanup;
}
if (!_handle_scheduling) {
OONF_WARN(LOG_MAIN, "No event scheduler present");
return_code = 1;
goto oonf_cleanup;
}
/* see if we need to fork */
if (config_global.fork && !_display_schema) {
/* tell main process that we are finished with initialization */
if (daemon(0,0) < 0) {
OONF_WARN(LOG_MAIN, "Could not fork into background: %s (%d)",
strerror(errno), errno);
goto oonf_cleanup;
}
if (config_global.pidfile && *config_global.pidfile != 0) {
if (_write_pidfile(config_global.pidfile)) {
goto oonf_cleanup;
}
}
}
/* activate mainloop */
return_code = mainloop(argc, argv, appdata);
/* tell plugins shutdown is in progress */
oonf_subsystem_initiate_shutdown();
/* wait for 500 ms and process socket events */
while(!_handle_scheduling());
oonf_cleanup:
/* free plugins */
oonf_cfg_unconfigure_subsystems();
oonf_subsystem_cleanup();
/* free logging/config bridge resources */
oonf_logcfg_cleanup();
/* free configuration resources */
oonf_cfg_cleanup();
/* free logger resources */
oonf_log_cleanup();
/* free core resources */
os_core_cleanup();
return return_code;
}
