Red Hat Bugzilla – Bug 217703
clustered vg creation fails
Last modified: 2009-04-16 18:35:29 EDT
Description of problem:
5 node RHEL5 cluster, with all members joined, and clvmd running on each node,
hooked via fiber to a winsys storage array with 2 tb of storage listed as sdb1.
Note clvmd check below:
[root@maggie ~]# service clvmd status
clvmd (pid 2944) is running...
active volumes: LogVol00 LogVol01
When sdb1 is used in a new VG, the following error message is returned:
An error has occurred while commiting changes:
clvmd (clustered LVM daemon) is not running on maggie.lab.boston.redhat.com.
Start it and try it again.
Version-Release number of selected component (if applicable):
I tried making the VG on a couple of different nodes in the cluster - all had
clvmd running and were joined nodes, with the same result.
One other thing: it seems to me that
1) ricci should try and start clvmd for you. If it *is* trying to start it and
failing, then the error message should say as much.
2) I wish there was a way to know if clvmd is running on a system when I go to
that system page in the interface...I know this is not for now, but would be
nice to have down the road.
So...this bug actually has two things that need to be corrected in order to
1) The error msg above is returned if the locking_type does not equal '2' in the
lvm.conf file....but clvmd WAS running...locking_type was just set to '1'. I
think the error message needs to be modified to explain this better.
2) When installing lvm2-cluster rpms, the locking_type is no longer set to
clustered...the default action is not to touch the value. the lvmconf tool must
be run after installation with the arg --enable-cluster in order to set
locking_type correctly...HOWEVER: The acceptable values for clustered locking
type are now '2' and '3'...in fact, '3' is what maggie was set to after running
lvmconf on it...so ricci needs to be modified such that this code:
return get_locking_type() == "2";
allows for "2" or "3"
Devel ACK - Conga needs to be able to work with cluster volumes.
Changing to assigned as one more thing needs to be addressed:
if "enable shared storage" checkbox is selected during cluster deployment,
--enable-cluster should be run on every node (so user doesn't have to log into
each machine and run it by hand)
Commited to -HEAD and -RHEL5
A package has been built which should help the problem described in
this bug report. This report is therefore being closed with a resolution
of CURRENTRELEASE. You may reopen this bug report if the solution does
not work for you.