Description of problem: in the latest fence package (fence-1.32.18), an empty fence-level will be treated as a successful execution of fencing rather than being skipped. Version-Release number of selected component (if applicable): fence-1.32.18 How reproducible: in a CS4 cluster, create a node and simply define an empty fence-level (without any fence device attached to it). in /etc/cluster/cluster.conf the node definition should look like this: <clusternode name="node1" votes="1"> <fence> <method name="1"/> </fence> </clusternode> If "node1" dies, it will be successfully fenced and services will migrate. Steps to Reproduce: 1. 2. 3. Actual results: an empty fence-level is reported as a successful fencing, even if no fencing actually happens. Expected results: an empty fence-level should be skipped and not treated as a success. in the aforementioned example, fencing should have failed. Additional info: re-initializing the "error" variable to -1 should correct this behaviour. --- fence-1.32.18/fenced/agent.c.orig 2006-05-04 12:01:57.000000000 +0100 +++ fence-1.32.18/fenced/agent.c 2006-04-28 08:44:24.000000000 +0100 @@ -287,6 +287,7 @@ if (error) continue; + error = -1; num_devices = count_devices(cd, victim, method); for (d = 0; d < num_devices; d++) {
fixed, thanks for the patch!
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 the 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-2006-0520.html