vdsmd fails after reboot

Latest response

Hi,

 

I've added a rhel-server 6.2 in rhev manager. Everything seems to works fine, it installs succesfully, but if I boot the machine, the vsdmd deamon will not start:

 

$ service vdsmd start
vdsm: libvirt already configured for vdsm                  [  OK  ]
Starting iscsid:
initctl: Unknown job: libvirtd
vdsm: one of the dependent services did not start, error co[FAILED]  

It looks like libvirtd fails

 

$ initctl start libvirtd
initctl: Unknown job: libvirtd

I can resolve the issue, doing a inictl reload-configuration, but wonder why this happens and how to resolve properly?

 

$ initctl reload-configuration
$ service vdsmd start
vdsm: libvirt already configured for vdsm                  [  OK  ]
Starting iscsid:
Starting up vdsm daemon:
vdsm start                                                 [  OK  ]

Regards, Brian

Responses

Hi Brian,

 

What build of vdsm are you talking about there? We have had such a problem in the early beta and it should be fixed long ago, so if you're using old code, you definitely need to update

 

Hi Dan,

 

Version on rhel-server is vdsm-4.9-110.el6.x86_64, which to my understanding is the latest?

 

Thanks, Brian

 

Yes, it is the latest. I will need to investigate this a bit further, will get back to you next week, after talking to engineering.

Could you put s host sosreport somewhere I can download?  https://access.redhat.com/kb/docs/DOC-61039 might be of help if you have no facility of your own

Hi Dan,

 

sosreport uploaded to dropbox.redhat.com

 

sosreport-Bchristensen.12345678...........

 

Thanks, Brian

Looks like the sosreport is missing libvirtd.log, can you upload it as well? Should be under /var/log/libvirt

I think I see where it might have failed  by looking at the code, and I also see that it has been fixed in vdsm minor build 112 (you're using 110).

 

Please wait for an update, and upgrade the packages, in the meantime, you can add the config reload command in rc.local as a workaround

Ok, thanks Dan,

 

I'll wait for the update.

 

Regards, Brian

The RHEV log collector was designed to capture all of the logs that a support person needs to investigate an issue.  It has various command-line options for collecting logs from hypervisor(s), and it has the the ability to upload directly to the dropbox. You might want to give it a try. 

 

Example - Collect from one hypervisor:

[root@rhevm src]# rhevm-log-collector --hosts=node1.example.com collect
About to collect information from 1 hypervisors. Continue? (Y/n):

 

Example - Collect logs from all hosts in a DC and upload when done:

[root@rhevm src]# rhevm-log-collector -d Default --upload=ftp://dropbox.redhat.com/incoming collect
Please provide the password for RHEV-M (CTRL+D to skip):
About to collect information from 4 hypervisors. Continue? (Y/n): 

For more examples, see manpage.