[Pacemaker] Problems shutting down server/services

Gianluca Cecchi gianluca.cecchi at gmail.com
Thu Jun 17 07:51:12 EDT 2010


On Thu, Jun 17, 2010 at 12:28 PM, Erich Weiler <weiler at soe.ucsc.edu> wrote:

> this is a problem ... without libvirtd running all management attempts via
>> "virsh" will fail (including the VirtualDomain RA) ... why is libvirtd
>> stopping before corosync?
>>
>
> Ah, because I forgot that I have libvirtd starting and stopping via init
> scripts.  ;)  That is the default shutdown sequence for libvirtd, as it is
> installed in REHL/CentOS 5.5.  I'm surprised more folks haven't noticed this
> issue...
>
> I totally forgot that I am supposed to disable init scripts for any service
> that pacemaker controls.  But in this case maybe I'll just move the libvirtd
> shutdown script to be much later in the shutdown process (after corosync).
>
> Thanks!!
>
>
Just in case, as this has bitten mee in the past ;-(

here "disable" means "chkconfig --del service_name"
because if you run a "chkconfig service_name off"
it will be stoppped during shutdown: the KNN links will stay in place when
you mark a service as off.

I like to notice this, as inside man page for chkconfig is is written to set
it off to disable it...
So that correct statement, from chkconfig man page point of view, would be:

disable ===> remove from chkconfig management
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.clusterlabs.org/pipermail/pacemaker/attachments/20100617/0b56c952/attachment-0001.html>


More information about the Pacemaker mailing list