4

We are deploying Red Hat Enterprise Virtualization cluster and we faced a problem attempting to start clvmd service after the lvm partition creating. The following has been completed successfully:

vgchange -cy RHEVMVolGroup
lvmconf --enable-cluster
dracut...
reboot

Now we are trying to run:

service clvmd restart

and it failes with an error in /var/log/messages:

clvmd could not connect to cluster manager
Can't open cluster manager socket: No such file or directory

And strace -f clvmd notices about the following troubles:

....
connect(.../var/run/cman_client...) (No such file or directory)
corosync.ipc (Connection refused)
....

Afterwards we can see that pvs command reports about the problem from the other point of view:

...
connect () failed on local socket:
internal cluster lock. init. failed
...

Any help will be appreciated.

Community
  • 1
Vitaly Isaev
  • 159
  • 2
  • 5
  • 3
    Is cman running? Try service cman start (or status or restart) (based on the cman_client error). – ramruma Dec 24 '13 at 08:12
  • No, cman is not running because according to the **Setting up a RHEV-M 3.1 on a Highly Available Cluster (HA LVM)** we are expected to run cman only after configuring and starting the clvmd (Chapter **Choise #2: Set Up HA-LVM failover (CLVM))** – Vitaly Isaev Dec 24 '13 at 08:27
  • clvmd has need of cman, RHEV-M 3.1 Highly Available Cluster uses redhat cluster :) – c4f4t0r Dec 24 '13 at 08:38
  • What is this for? Are you trying to set up an HA engine? If so, why not simply use the hosted engine, it already is highly available without the extra work – dyasny Aug 26 '15 at 19:02

1 Answers1

0

Thank you. There was a bug in the RHEV documentation: cman, luci deployment and all the related work should have been done before the clvmd configuration.

c4f4t0r
  • 5,301
  • 3
  • 31
  • 42
Vitaly Isaev
  • 159
  • 2
  • 5