When clusvcmgrd gets a request to start a service and it can't because of a failover-domain restriction, it currently simply returns failure with no action taken. This patch alters clusvcmgrd's behavior to mimic normal service start failures: it attempts to relocate the service to cluster members which are allowed run it.
Created attachment 99955 [details] Patch which fixes problem.
I. Requirements/setup A. Requirements 1. 2 node cluster (ex: {magenta, yellow}) 2. a restricted failover domain comprised of one of the members (ex: {magenta}) 3. a service bound to the above domain. (ex: "testing") B. Setup 1. Start cluster software on both members. 2. Fire up redhat-config-cluster on member NOT specified in I.A.2. (in our case, yellow) II. Old behavior A. Steps 1. Click Service defined in I.A.3. 2. Click "Disable" button. 3. Click "Enable" button. B. Result 1. Service will not start. 2. clusvcadm output: Member yellow enabling testing...failed III. New Behavior A. Ditto. B. Result 1. Service starts on member in RFD defined in I.A.2. 2. clusvcadm reports: Member yellow trying to enable testing...success Note: Service testing failed to start on yellow. Service testing is now running on magenta.
An errata 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/RHEA-2004-239.html
Fixing product name. Clumanager on RHEL3 was part of RHCS3, not RHEL3