This content has been marked as final. Show 9 replies
Updating the cluster configuration means editing the cluster configuration file and propagating it to each node in the cluster.
Have you tried the following:
1. Edit /etc/cluster/cluster.conf on any node and increment config_version.
2. Run cman_tool version -r to propagate the the configuration.
You might want to run "ccs_config_validate" to verify your configuration. Depending on what resource you changed you may have to restart the cluster software.
Restarting has no effect.
I believe either named is,
a) Disabled deliberately somehow in the definition (not 'production ready?)
b) has an error in the definition (xml code?) causing it to fail to parse correctly and as a result 'vanish'.
Unfortunately, I lack the knowledge of exactly how luci gets it's data for resources. I THINK it is from /usr/share/cluster/*
Many people think they have discovered a bug which turned out to be something else. Maybe it is a bug, ok.
However, if you want to report a bug, then detailed information to be able to reproduce or troubleshoot the issue is typically required. The current information may be your own and valid assumption and conclusion, but I'm afraid it is not sufficient to confirm a bug.
Perhaps someone else has run into the very same or similar issue and knows the solution. Good luck.
Well, that's pretty arrogant..
"However, if you want to report a bug, then detailed information to be able to reproduce or troubleshoot the issue is typically required. The current information may be your own and valid assumption and conclusion, but I'm afraid it is not sufficient to confirm a bug."
What part of the previous statement,
"With a brand new empty cluster, the Resource "named" does not appear in the pull down lists of luci."
fails to meet any of the requirements to report a bug?
Did you try that simple test on any cluster system you have access to? It is harmless to check.
As a follow up, I have also verified the lack of named in the pull down of a similar Redhat cluster, suggestion there may be a 'common issue' in the base code somewhere. The named.sh and named.metadata files DO exist in /usr/share/cluster/.