/*
called when a ctdb_call times out
*/
void ctdb_call_timeout(struct event_context *ev, struct timed_event *te,
struct timeval t, void *private_data)
{
struct ctdb_call_state *state = talloc_get_type(private_data, struct ctdb_call_state);
state->state = CTDB_CALL_ERROR;
ctdb_set_error(state->node->ctdb, "ctdb_call %u timed out",
state->c->hdr.reqid);
if (state->async.fn) {
state->async.fn(state);
}
}
int ctdb_ibw_get_address(struct ctdb_context *ctdb,
const char *address, struct in_addr *addr)
{
if (inet_pton(AF_INET, address, addr) <= 0) {
struct hostent *he = gethostbyname(address);
if (he == NULL || he->h_length > sizeof(*addr)) {
ctdb_set_error(ctdb, "invalid nework address '%s'\n",
address);
return -1;
}
memcpy(addr, he->h_addr, he->h_length);
}
return 0;
}
/*
make a remote ctdb call - async recv - called in daemon context
This is called when the program wants to wait for a ctdb_call to complete and get the
results. This call will block unless the call has already completed.
*/
int ctdb_daemon_call_recv(struct ctdb_call_state *state, struct ctdb_call *call)
{
while (state->state < CTDB_CALL_DONE) {
event_loop_once(state->node->ctdb->ev);
}
if (state->state != CTDB_CALL_DONE) {
ctdb_set_error(state->node->ctdb, "%s", state->errmsg);
talloc_free(state);
return -1;
}
if (state->call.reply_data.dsize) {
call->reply_data.dptr = talloc_memdup(state->node->ctdb,
state->call.reply_data.dptr,
state->call.reply_data.dsize);
call->reply_data.dsize = state->call.reply_data.dsize;
} else {
call->reply_data.dptr = NULL;
call->reply_data.dsize = 0;
}
call->status = state->call.status;
talloc_free(state);
return 0;
}
/*
local version of ctdb_call
*/
int ctdb_call_local(struct ctdb_db_context *ctdb_db, struct ctdb_call *call,
struct ctdb_ltdb_header *header, TDB_DATA *data,
uint32_t caller)
{
struct ctdb_call_info *c;
struct ctdb_registered_call *fn;
struct ctdb_context *ctdb = ctdb_db->ctdb;
c = talloc(ctdb, struct ctdb_call_info);
CTDB_NO_MEMORY(ctdb, c);
c->key = call->key;
c->call_data = &call->call_data;
c->record_data.dptr = talloc_memdup(c, data->dptr, data->dsize);
c->record_data.dsize = data->dsize;
CTDB_NO_MEMORY(ctdb, c->record_data.dptr);
c->new_data = NULL;
c->reply_data = NULL;
c->status = 0;
for (fn=ctdb_db->calls;fn;fn=fn->next) {
if (fn->id == call->call_id) break;
}
if (fn == NULL) {
ctdb_set_error(ctdb, "Unknown call id %u\n", call->call_id);
talloc_free(c);
return -1;
}
if (fn->fn(c) != 0) {
ctdb_set_error(ctdb, "ctdb_call %u failed\n", call->call_id);
talloc_free(c);
return -1;
}
if (header->laccessor != caller) {
header->lacount = 0;
}
header->laccessor = caller;
header->lacount++;
/* we need to force the record to be written out if this was a remote access,
so that the lacount is updated */
if (c->new_data == NULL && header->laccessor != ctdb->vnn) {
c->new_data = &c->record_data;
}
if (c->new_data) {
if (ctdb_ltdb_store(ctdb_db, call->key, header, *c->new_data) != 0) {
ctdb_set_error(ctdb, "ctdb_call tdb_store failed\n");
talloc_free(c);
return -1;
}
}
if (c->reply_data) {
call->reply_data = *c->reply_data;
talloc_steal(ctdb, call->reply_data.dptr);
talloc_set_name_const(call->reply_data.dptr, __location__);
} else {
call->reply_data.dptr = NULL;
call->reply_data.dsize = 0;
}
call->status = c->status;
talloc_free(c);
return 0;
}
/*
listen on our own address
*/
int ctdb_tcp_listen(struct ctdb_context *ctdb)
{
struct ctdb_tcp *ctcp = talloc_get_type(ctdb->private_data,
struct ctdb_tcp);
ctdb_sock_addr sock;
int sock_size;
int one = 1;
struct tevent_fd *fde;
/* we can either auto-bind to the first available address, or we can
use a specified address */
if (!ctdb->address) {
return ctdb_tcp_listen_automatic(ctdb);
}
sock = *ctdb->address;
switch (sock.sa.sa_family) {
case AF_INET:
sock_size = sizeof(sock.ip);
break;
case AF_INET6:
sock_size = sizeof(sock.ip6);
break;
default:
DEBUG(DEBUG_ERR, (__location__ " unknown family %u\n",
sock.sa.sa_family));
goto failed;
}
ctcp->listen_fd = socket(sock.sa.sa_family, SOCK_STREAM, IPPROTO_TCP);
if (ctcp->listen_fd == -1) {
ctdb_set_error(ctdb, "socket failed\n");
return -1;
}
set_close_on_exec(ctcp->listen_fd);
if (setsockopt(ctcp->listen_fd,SOL_SOCKET,SO_REUSEADDR,(char *)&one,sizeof(one)) == -1) {
DEBUG(DEBUG_WARNING, ("Failed to set REUSEADDR on fd - %s\n",
strerror(errno)));
}
if (bind(ctcp->listen_fd, (struct sockaddr * )&sock, sock_size) != 0) {
DEBUG(DEBUG_ERR,(__location__ " Failed to bind() to socket. %s(%d)\n", strerror(errno), errno));
goto failed;
}
if (listen(ctcp->listen_fd, 10) == -1) {
goto failed;
}
fde = tevent_add_fd(ctdb->ev, ctcp, ctcp->listen_fd, TEVENT_FD_READ,
ctdb_listen_event, ctdb);
tevent_fd_set_auto_close(fde);
return 0;
failed:
if (ctcp->listen_fd != -1) {
close(ctcp->listen_fd);
}
ctcp->listen_fd = -1;
return -1;
}
/*
automatically find which address to listen on
*/
static int ctdb_tcp_listen_automatic(struct ctdb_context *ctdb)
{
struct ctdb_tcp *ctcp = talloc_get_type(ctdb->private_data,
struct ctdb_tcp);
ctdb_sock_addr sock;
int lock_fd, i;
const char *lock_path = CTDB_RUNDIR "/.socket_lock";
struct flock lock;
int one = 1;
int sock_size;
struct tevent_fd *fde;
/* If there are no nodes, then it won't be possible to find
* the first one. Log a failure and short circuit the whole
* process.
*/
if (ctdb->num_nodes == 0) {
DEBUG(DEBUG_CRIT,("No nodes available to attempt bind to - is the nodes file empty?\n"));
return -1;
}
/* in order to ensure that we don't get two nodes with the
same adddress, we must make the bind() and listen() calls
atomic. The SO_REUSEADDR setsockopt only prevents double
binds if the first socket is in LISTEN state */
lock_fd = open(lock_path, O_RDWR|O_CREAT, 0666);
if (lock_fd == -1) {
DEBUG(DEBUG_CRIT,("Unable to open %s\n", lock_path));
return -1;
}
lock.l_type = F_WRLCK;
lock.l_whence = SEEK_SET;
lock.l_start = 0;
lock.l_len = 1;
lock.l_pid = 0;
if (fcntl(lock_fd, F_SETLKW, &lock) != 0) {
DEBUG(DEBUG_CRIT,("Unable to lock %s\n", lock_path));
close(lock_fd);
return -1;
}
for (i=0; i < ctdb->num_nodes; i++) {
if (ctdb->nodes[i]->flags & NODE_FLAGS_DELETED) {
continue;
}
sock = ctdb->nodes[i]->address;
switch (sock.sa.sa_family) {
case AF_INET:
sock_size = sizeof(sock.ip);
break;
case AF_INET6:
sock_size = sizeof(sock.ip6);
break;
default:
DEBUG(DEBUG_ERR, (__location__ " unknown family %u\n",
sock.sa.sa_family));
continue;
}
ctcp->listen_fd = socket(sock.sa.sa_family, SOCK_STREAM, IPPROTO_TCP);
if (ctcp->listen_fd == -1) {
ctdb_set_error(ctdb, "socket failed\n");
continue;
}
set_close_on_exec(ctcp->listen_fd);
if (setsockopt(ctcp->listen_fd,SOL_SOCKET,SO_REUSEADDR,
(char *)&one,sizeof(one)) == -1) {
DEBUG(DEBUG_WARNING, ("Failed to set REUSEADDR on fd - %s\n",
strerror(errno)));
}
if (bind(ctcp->listen_fd, (struct sockaddr * )&sock, sock_size) == 0) {
break;
}
if (errno == EADDRNOTAVAIL) {
DEBUG(DEBUG_DEBUG,(__location__ " Failed to bind() to socket. %s(%d)\n",
strerror(errno), errno));
} else {
DEBUG(DEBUG_ERR,(__location__ " Failed to bind() to socket. %s(%d)\n",
strerror(errno), errno));
}
close(ctcp->listen_fd);
ctcp->listen_fd = -1;
}
if (i == ctdb->num_nodes) {
DEBUG(DEBUG_CRIT,("Unable to bind to any of the node addresses - giving up\n"));
goto failed;
}
ctdb->address = talloc_memdup(ctdb,
&ctdb->nodes[i]->address,
sizeof(ctdb_sock_addr));
if (ctdb->address == NULL) {
ctdb_set_error(ctdb, "Out of memory at %s:%d",
__FILE__, __LINE__);
goto failed;
}
ctdb->name = talloc_asprintf(ctdb, "%s:%u",
ctdb_addr_to_str(ctdb->address),
ctdb_addr_to_port(ctdb->address));
if (ctdb->name == NULL) {
ctdb_set_error(ctdb, "Out of memory at %s:%d",
__FILE__, __LINE__);
goto failed;
}
DEBUG(DEBUG_INFO,("ctdb chose network address %s\n", ctdb->name));
if (listen(ctcp->listen_fd, 10) == -1) {
goto failed;
}
fde = tevent_add_fd(ctdb->ev, ctcp, ctcp->listen_fd, TEVENT_FD_READ,
ctdb_listen_event, ctdb);
tevent_fd_set_auto_close(fde);
close(lock_fd);
return 0;
failed:
close(lock_fd);
if (ctcp->listen_fd != -1) {
close(ctcp->listen_fd);
ctcp->listen_fd = -1;
}
return -1;
}