Description of problem: A. Get the kernel cluster up [root@morph-04 root]# cat /proc/cluster/nodes Node Votes Exp Sts Name 1 1 6 M morph-06 2 1 6 M morph-05 3 1 6 M morph-03 4 1 6 M morph-01 5 1 6 M morph-04 6 1 6 M morph-02 [root@morph-04 root]# head /etc/cluster/cluster.conf <?xml version="1.0"?> <cluster name="morph-cluster" config_version="1"> B. Edit and update /etc/cluster/cluster.conf on another node and then let cman know on this node (and all others in cluster) [root@morph-04 root]# cman_tool version -r 2 [root@morph-04 root]# head /etc/cluster/cluster.conf <?xml version="1.0"?> <cluster name="morph-cluster" config_version="2"> [root@morph-04 root]# cat /proc/cluster/status Protocol version: 4.0.1 Config version: 2 Cluster name: morph-cluster Cluster ID: 41652 Membership state: Cluster-Member Nodes: 6 Expected_votes: 6 Total_votes: 6 Quorum: 4 Active subsystems: 0 Node addresses: 192.168.44.64 C. Edit and update /etc/cluster/cluster.conf on this node [root@morph-04 root]# head /etc/cluster/cluster.conf <?xml version="1.0"?> <cluster name="morph-cluster" config_version="3"> D. Attempt of ccs_tool update causes the version to be bumped back down and fail [root@morph-04 root]# ccs_tool update /etc/cluster/cluster.conf Proposed updated config file does not have greater version number. Current config_version :: 2 Proposed config_version:: 2 Failed to update config file. [root@morph-04 root]# head /etc/cluster/cluster.conf <?xml version="1.0"?> <cluster name="morph-cluster" config_version="2"> However, if I once again edit /etc/cluster/cluster.conf and then update, it will work [root@morph-04 root]# ccs_tool update /etc/cluster/cluster.conf Config file updated from version 2 to 3 Update complete. Version-Release number of selected component (if applicable): [root@morph-04 root]# ccsd -V ccsd 0.17. (built Jan 13 2005 18:56:06) Copyright (C) Red Hat, Inc. 2004 All rights reserved. How reproducible: Always Additional info: if you do not edit /etc/cluster/cluster.conf and instead use a different file for updates, it will work just fine.
debugged this further and found that for the update attempt to fail, it has to be on another node than where the first update was done from.
fixed.
fix verified.