Description of problem: Cluster IP resource brings up an IP address without prior checking. If the IP remains live on other cluster member it leads to IP address collision. Version-Release number of selected component (if applicable): rgmanager-2.0.52-1.el5 How reproducible: Starting a cluster service with an IP resource while that IP address exists on the network. Steps to Reproduce: 1. Create cluster service with IP resource 2. Bring up the same IP address manually on nodeB 3. Start the service on nodeA Actual results: Service brings up the IP address and IP address collision occurs on the network. Expected results: /usr/share/cluster/ip.sh should check an IP address presence before bringing up an IP address. If the IP address is responding it should block the service avoiding IP address collision. Additional info:
Created attachment 365025 [details] suggested patch for ip.sh to avoid IP address collisions This patch inserts ping check into ip.sh prior to IP address activation.
I would change the 'notice' to 'debug', otherwise the patch is good.
Would you like me to replace 'notice' to 'debug' and upload the modified one?
No, it's fine. If you're okay with that small change I will just commit the patch with the minor change with you as the author.
http://git.fedorahosted.org/git/?p=cluster.git;a=commit;h=3ae5f5cea15c8198141308d0fc8443853c45faf4 http://git.fedorahosted.org/git/?p=cluster.git;a=commit;h=d767b979e3841b01d39e498e15ae72d446c12f78
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. http://rhn.redhat.com/errata/RHBA-2010-0280.html