Hi,

On Wed, Jan 16, 2008 at 03:39:50PM +0100, Yves Schumann wrote:
> Hi Dejan
> 
> 
> [EMAIL PROTECTED] wrote on 11.01.2008 16:38:46:
> > There's an unfortunate bug in 2.1.2 which makes lrmd eat quite a
> > bit of memory and cpu if you have often monitor operations. Or,
> > the communication is lighter now that the compression is used, in
> > particular if your CIB is big. Probably quite a few other fixes.
> > Take a look at the changelog and pick your favourite argument :)
> 
> OK, these arguments did a nice job and so I've been running the 2.1.3 now.
> 
> > > > > Is it possible that the var "OCF_RESKEY_CRM_meta_clone" is not set
> and
> > > so
> > > > > expr runs into a error? I did not found any other occourence of
> >
> > This was introduced in the changeset 784fcb44d688.
> > cc-ing the relevant parties :) Is OCF_RESKEY_CRM_meta_clone always
> > set by v2?
> 
> But this problem is still there and actually I have no idea where to look
> for this!? Any ideas or hints?

It's broken. It should check if the CRM_meta_clone* is defined
first and only then try to use it. I'll try to fix it. Looks like
nobody but you is using IPaddr2 ;-)

> > > > Looks like it. BTW, why do you use the heartbeat and not OCF RA?
> > >
> > > Hm, maybe I understand the (not existing :-| ) documentation wrong, but
> I
> > > defined the resource like this:
> > >
> > > <primitive id="rsc_IPaddr" class="ocf" type="IPaddr2"
> provider="heartbeat">
> > >  <instance_attributes id="rsc_IPaddr_instance_attrs">
> > >   <attributes>
> > >    <nvpair id="96252422-60e0-4e3f-bcd6-4d1c3996d10e" name="ip"
> > > value="172.17.206.153"/>
> > >   </attributes>
> > >  </instance_attributes>
> > >  <meta_attributes id="rsc_IPaddr_meta_attrs">
> > >   <attributes>
> > >    <nvpair id="rsc_IPaddr_metaattr_target_role" name="target_role"
> > > value="started"/>
> > >   </attributes>
> > >  </meta_attributes>
> > >  <operations>
> > >   <op id="2df96287-97cf-4d2e-b87b-66bb72090c33" name="monitor"
> > > interval="10s" timeout="20s" disabled="false" role="Started"/>
> > >  </operations>
> > > </primitive>
> > >
> > > This is _not_ a ocf resource?
> >
> > Yes, it is. But in this case the wrapper script
> > /etc/ha.d/resource.d/IPaddr2 is not run. That's what made me
> > think that you're using the Heartbeat class RA.
> 
> Can you explain this a little bit more please? Actually I can't follow...

Different RA classes support different semantics. The wrapper is
there only to allow an OCF class RA to be used as a Heartbeat RA.
Does that make it more clear now?

Thanks,

Dejan

> Here is what crm_mon shows with the above resource configuration:
> 
> --------------->8 snip 8<----------------------------
> 
> ============
> Last updated: Wed Jan 16 15:38:13 2008
> Current DC: oan-redundanz (f2ddf7a6-1f3a-4184-8ddc-e82605ef77d9)
> 2 Nodes configured.
> 1 Resources configured.
> ============
> 
> Node: oan-redundanz (f2ddf7a6-1f3a-4184-8ddc-e82605ef77d9): online
> Node: oan-master (1ef0ac81-4865-4d40-8152-bf054d02745c): online
> 
> Resource Group: oan_rsc_group
>     rsc_IPaddr  (heartbeat::ocf:IPaddr2):       Started oan-master
> 
> --------------->8 snap 8<----------------------------
> 
> 
> 
> Yours
> 
> Yves Schumann
> Softwareentwicklungsingenieur
> Security Solutions Division
> ______________________________
> Ascom (Schweiz) AG
> Eichtal, CH-8634 Hombrechtikon
> Phone: +41 55 254 66 84
> [EMAIL PROTECTED]
> http://www.ascom.com
> 
> _______________________________________________
> Linux-HA mailing list
> [email protected]
> http://lists.linux-ha.org/mailman/listinfo/linux-ha
> See also: http://linux-ha.org/ReportingProblems
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

Reply via email to