On Sun, Apr 26, 2020 at 07:24:33AM +0000, Gang He wrote:
> Hello List,
>
> In libdlm code, cluster_ringid_seq variable is defined with uint32_t in dlm_controld/dlm_daemon.h,
> the corosync API returns uinit64_t ring_id, in the current code, we use type cast to get the low-32bit ring-id.
> But, in some case, the corosync returns a huge ring-id (greater than 32bit), the DLM daemon does not work normally (looks stuck).
> Then, I want to know if we can change cluster_ringid_seq in libdlm from uint32_t to uint64_t?
That looks ok, please try the attached patch.
Dave
diff --git a/dlm_controld/cpg.c b/dlm_controld/cpg.c
index e4b27fac3cd5..5b5c52fc09bc 100644
--- a/dlm_controld/cpg.c
+++ b/dlm_controld/cpg.c
@@ -450,21 +450,24 @@ static int check_ringid_done(struct lockspace *ls)
but that's probably not guaranteed.) */
if (ls->cpg_ringid_wait) {
- log_group(ls, "check_ringid wait cluster %u cpg %u:%llu",
- cluster_ringid_seq, ls->cpg_ringid.nodeid,
+ log_group(ls, "check_ringid wait cluster %u cpg %llu:%llu",
+ (unsigned long long)cluster_ringid_seq,
+ ls->cpg_ringid.nodeid,
(unsigned long long)ls->cpg_ringid.seq);
return 0;
}
- if (cluster_ringid_seq != (uint32_t)ls->cpg_ringid.seq) {
- log_group(ls, "check_ringid cluster %u cpg %u:%llu",
- cluster_ringid_seq, ls->cpg_ringid.nodeid,
+ if (cluster_ringid_seq != ls->cpg_ringid.seq) {
+ log_group(ls, "check_ringid cluster %llu cpg %u:%llu",
+ (unsigned long long)cluster_ringid_seq,
+ ls->cpg_ringid.nodeid,
(unsigned long long)ls->cpg_ringid.seq);
return 0;
}
- log_limit(ls, "check_ringid done cluster %u cpg %u:%llu",
- cluster_ringid_seq, ls->cpg_ringid.nodeid,
+ log_limit(ls, "check_ringid done cluster %llu cpg %u:%llu",
+ (unsigned long long)cluster_ringid_seq,
+ ls->cpg_ringid.nodeid,
(unsigned long long)ls->cpg_ringid.seq);
return 1;
diff --git a/dlm_controld/dlm_daemon.h b/dlm_controld/dlm_daemon.h
index 3221e19cb7be..5b9a52dae750 100644
--- a/dlm_controld/dlm_daemon.h
+++ b/dlm_controld/dlm_daemon.h
@@ -179,7 +179,7 @@ EXTERN uint64_t fence_delay_begin;
EXTERN uint64_t cluster_quorate_monotime;
EXTERN uint64_t cluster_joined_monotime;
EXTERN uint64_t cluster_joined_walltime;
-EXTERN uint32_t cluster_ringid_seq;
+EXTERN uint64_t cluster_ringid_seq;
EXTERN char cluster_name[DLM_LOCKSPACE_LEN+1];
EXTERN int our_nodeid;
EXTERN uint32_t control_minor;
diff --git a/dlm_controld/member.c b/dlm_controld/member.c
index 10351ec41d6d..da3a1f5b0e90 100644
--- a/dlm_controld/member.c
+++ b/dlm_controld/member.c
@@ -122,10 +122,10 @@ static void quorum_callback(quorum_handle_t h, uint32_t quorate,
cluster_quorate_monotime = now;
cluster_quorate = quorate;
- cluster_ringid_seq = (uint32_t)ring_seq;
+ cluster_ringid_seq = ring_seq;
- log_debug("cluster quorum %u seq %u nodes %u",
- cluster_quorate, cluster_ringid_seq, node_list_entries);
+ log_debug("cluster quorum %u seq %llu nodes %u",
+ cluster_quorate, (unsigned long long)cluster_ringid_seq, node_list_entries);
old_node_count = quorum_node_count;
memcpy(&old_nodes, &quorum_nodes, sizeof(old_nodes));
@@ -139,8 +139,8 @@ static void quorum_callback(quorum_handle_t h, uint32_t quorate,
for (i = 0; i < old_node_count; i++) {
if (!is_cluster_member(old_nodes[i])) {
- log_debug("cluster node %u removed seq %u",
- old_nodes[i], cluster_ringid_seq);
+ log_debug("cluster node %u removed seq %llu",
+ old_nodes[i], (unsigned long long)cluster_ringid_seq);
rem_cluster_node(old_nodes[i], now);
del_configfs_node(old_nodes[i]);
}
@@ -148,8 +148,8 @@ static void quorum_callback(quorum_handle_t h, uint32_t quorate,
for (i = 0; i < quorum_node_count; i++) {
if (!is_old_member(quorum_nodes[i])) {
- log_debug("cluster node %u added seq %u",
- quorum_nodes[i], cluster_ringid_seq);
+ log_debug("cluster node %u added seq %llu",
+ quorum_nodes[i], (unsigned long long)cluster_ringid_seq);
add_cluster_node(quorum_nodes[i], now);
fence_delay_begin = now;
_______________________________________________
linux-lvm mailing list
linux-lvm@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/
