From: David Teigland <teigland@redhat.com>
To: cluster-devel.redhat.com
Subject: [Cluster-devel] [PATCH 1/3] fenced: full check for member changes
Date: Wed, 28 Sep 2011 11:37:26 -0500 [thread overview]
Message-ID: <1317227848-8323-1-git-send-email-teigland@redhat.com> (raw)
cman members are queried in response to a callback,
and members sometimes leave and rejoin between queries
(e.g. when they leave and rejoin before corosync
detects they left.)
This means that simply checking if a node is a member
in consecutive queries sometimes misses events. We
need to compare the incarnation numbers of members
from consecutive queries to avoid this.
bz 663397
Signed-off-by: David Teigland <teigland@redhat.com>
---
fence/fenced/member_cman.c | 36 ++++++++++++++++++++++++++++++++++++
1 files changed, 36 insertions(+), 0 deletions(-)
diff --git a/fence/fenced/member_cman.c b/fence/fenced/member_cman.c
index 0919b8e..ee879be 100644
--- a/fence/fenced/member_cman.c
+++ b/fence/fenced/member_cman.c
@@ -33,6 +33,17 @@ void kick_node_from_cluster(int nodeid)
}
}
+static cman_node_t *get_node(cman_node_t *node_list, int count, int nodeid)
+{
+ int i;
+
+ for (i = 0; i < count; i++) {
+ if (node_list[i].cn_nodeid == nodeid)
+ return &node_list[i];
+ }
+ return NULL;
+}
+
static int is_member(cman_node_t *node_list, int count, int nodeid)
{
int i;
@@ -149,6 +160,7 @@ int name_to_nodeid(char *name)
static void update_cluster(void)
{
cman_cluster_t info;
+ cman_node_t *old;
int quorate = cluster_quorate;
int removed = 0, added = 0;
int i, rv;
@@ -197,6 +209,30 @@ static void update_cluster(void)
node_history_cluster_add(cman_nodes[i].cn_nodeid);
added++;
+ } 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);
+
+ node_history_cluster_remove(cman_nodes[i].cn_nodeid);
+ removed++;
+
+ node_history_cluster_add(cman_nodes[i].cn_nodeid);
+ added++;
}
}
--
1.7.6
reply other threads:[~2011-09-28 16:37 UTC|newest]
Thread overview: [no followups] expand[flat|nested] mbox.gz Atom feed
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1317227848-8323-1-git-send-email-teigland@redhat.com \
--to=teigland@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).