Red Hat Bugzilla – Bug 460909
Two new nodes with dirty flag don't see each other
Last modified: 2010-10-23 00:14:19 EDT
Description of problem:
If two nodes start up in a partitioned network - ie they can't see each other, often because the switch separating them hasn't sorted out its multicast routes yet - and have the dirty flags set, when the network connection is restored the two nodes don't see each other. No node is fenced or even marked "disallowed".
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Separate the two nodes eg:
on 131: iptables -A INPUT -s 10.15.84.132 -p udp -j DROP
on 132: iptables -A INPUT -s 10.15.84.131 -p udp -j DROP
2. Join them to the cluster.
3. Set the dirty flag, I think fenced does this.
4. Join the cluster nodes with iptables -D INPUT 1
cman_tool nodes on both systems shows only the local node, and not the other. syslog shows that openais/Clm can see both nodes.
The other node shows up as "disallowed" or there is a fence race to kill one node.
This bug is similar to bz#443358
See also bz#460190
Created attachment 315566 [details]
Small program to set the dirty flag in cman
Created attachment 315567 [details]
This patch should fix the problem. I'd like to see it tested rather a lot before releasing it though.
This request was evaluated by Red Hat Product Management for inclusion in a Red
Hat Enterprise Linux maintenance release. Product Management has requested
further review of this request by Red Hat Engineering, for potential
inclusion in a Red Hat Enterprise Linux Update release for currently deployed
products. This request is not yet committed for inclusion in an Update
Author: Christine Caulfield <email@example.com>
Date: Wed Sep 10 09:06:25 2008 +0100
cman: honour the dirty flag on a node we haven't seen before
An advisory has been issued which should help the problem
described in this bug report. This report is therefore being
closed with a resolution of ERRATA. For more information
on therefore solution and/or where to find the updated files,
please follow the link below. You may reopen this bug report
if the solution does not work for you.