void Quorum::start(boost::shared_ptr<sys::Poller> p) {
poller = p;
QPID_LOG(debug, "Connecting to quorum service.");
cman = cman_init(0);
if (cman == 0) throw ErrnoException("Can't connect to cman service");
if (!cman_is_quorate(cman)) {
QPID_LOG(notice, "Waiting for cluster quorum.");
while(!cman_is_quorate(cman)) sys::sleep(5);
}
int err = cman_start_notification(cman, cmanCallbackFn);
if (err != 0) throw ErrnoException("Can't register for cman notifications");
watch(getFd());
}
static void delay_quorum(void)
{
int rv, tries = 0;
while (1) {
rv = cman_is_quorate(ch);
if (rv)
break;
rv = cman_is_active(ch);
if (!rv) {
cman_finish(ch);
die("lost cman connection");
}
tries++;
if (opt_delay_quorum < 0)
goto retry_quorum;
if (!opt_delay_quorum || tries >= opt_delay_quorum) {
fprintf(stderr, "%s: continuing without quorum\n", prog_name);
break;
}
retry_quorum:
if (!(tries % 10))
fprintf(stderr, "%s: delaying for quorum\n", prog_name);
sleep(1);
}
return;
}
static void statechange(void)
{
cman_cluster_t info;
cman_node_t *old;
int i, j, rv;
struct cman_node_address addrs[MAX_NODE_ADDRESSES];
int num_addrs;
struct cman_node_address *addrptr = addrs;
rv = cman_get_cluster(ch, &info);
if (rv < 0) {
log_error("cman_get_cluster error %d %d", rv, errno);
/* keep going, this is just informational */
memset(&info, 0, sizeof(info));
}
cluster_ringid_seq = info.ci_generation;
cluster_quorate = cman_is_quorate(ch);
old_node_count = cman_node_count;
memcpy(&old_nodes, &cman_nodes, sizeof(old_nodes));
cman_node_count = 0;
memset(&cman_nodes, 0, sizeof(cman_nodes));
rv = cman_get_nodes(ch, MAX_NODES, &cman_node_count, cman_nodes);
if (rv < 0) {
log_debug("cman_get_nodes error %d %d", rv, errno);
return;
}
/* Never allow node ID 0 to be considered a member #315711 */
for (i = 0; i < cman_node_count; i++) {
if (cman_nodes[i].cn_nodeid == 0) {
cman_nodes[i].cn_member = 0;
break;
}
}
for (i = 0; i < old_node_count; i++) {
if (old_nodes[i].cn_member &&
!is_cluster_member(old_nodes[i].cn_nodeid)) {
log_debug("cluster node %d removed seq %u",
old_nodes[i].cn_nodeid, cluster_ringid_seq);
node_history_cluster_remove(old_nodes[i].cn_nodeid);
del_configfs_node(old_nodes[i].cn_nodeid);
}
}
for (i = 0; i < cman_node_count; i++) {
if (cman_nodes[i].cn_member &&
!is_old_member(cman_nodes[i].cn_nodeid)) {
log_debug("cluster node %d added seq %u",
cman_nodes[i].cn_nodeid, cluster_ringid_seq);
rv = cman_get_node_addrs(ch, cman_nodes[i].cn_nodeid,
MAX_NODE_ADDRESSES,
&num_addrs, addrs);
if (rv < 0) {
log_debug("cman_get_node_addrs failed, falling back to single-homed. ");
num_addrs = 1;
addrptr = &cman_nodes[i].cn_address;
}
node_history_cluster_add(cman_nodes[i].cn_nodeid);
for (j = 0; j < num_addrs; j++) {
add_configfs_node(cman_nodes[i].cn_nodeid,
addrptr[j].cna_address,
addrptr[j].cna_addrlen,
(cman_nodes[i].cn_nodeid ==
our_nodeid));
}
} else {
/* look for any nodes that were members of both
* old and new but have a new incarnation number
* from old to new, indicating they left and rejoined
* in between */
old = get_node(old_nodes, old_node_count, cman_nodes[i].cn_nodeid);
if (!old)
continue;
if (cman_nodes[i].cn_incarnation == old->cn_incarnation)
continue;
log_debug("cluster node %d removed and added seq %u "
"old %u new %u",
cman_nodes[i].cn_nodeid, cluster_ringid_seq,
old->cn_incarnation,
cman_nodes[i].cn_incarnation);
/*
* remove (copied from above)
*/
node_history_cluster_remove(old_nodes[i].cn_nodeid);
del_configfs_node(old_nodes[i].cn_nodeid);
/*
* add (copied from above)
*/
rv = cman_get_node_addrs(ch, cman_nodes[i].cn_nodeid,
MAX_NODE_ADDRESSES,
&num_addrs, addrs);
if (rv < 0) {
log_debug("cman_get_node_addrs failed, falling back to single-homed. ");
num_addrs = 1;
addrptr = &cman_nodes[i].cn_address;
}
node_history_cluster_add(cman_nodes[i].cn_nodeid);
for (j = 0; j < num_addrs; j++) {
add_configfs_node(cman_nodes[i].cn_nodeid,
addrptr[j].cna_address,
addrptr[j].cna_addrlen,
(cman_nodes[i].cn_nodeid ==
our_nodeid));
}
}
}
}
