[Pacemaker] hbagent of pacemaker-mgmt

Yan Gao ygao at novell.com
Thu Jul 16 08:22:54 UTC 2009


On Thu, 2009-07-16 at 16:20 +0900, Junko IKEDA wrote:
> Hi,
> 
> Does openais need not "myuuid" but "myid"?
Usually, a node's uuid is same as its uname for openais. Please try the
new attached patch, it should work in most cases. As you know, System
uname could be changed later, So we'll seek a better way.

> I still get the following error.
> 
> Could not map uname=dom3 to a UUID: The object/attribute does not exist
> 
Hmm, hbagent was started so early that the node hadn't been added into
cib by that time.  Try the new patch then.


> 
> 
> On Thu, 16 Jul 2009 11:21:59 +0900, Junko IKEDA <ikedaj at intellilink.co.jp> wrote:
> 
> > Hi,
> >
> > I tried your patch,
> > and there was no problem with compiling,
> > but it complained during a start...
> >
> > # service openais start; export HA_cluster_type="openais"; /usr/lib64/heartbeat/hbagent -r 5 &
> > Starting OpenAIS daemon (aisexec): starting... rc=0: OK
> > [1] 2569
> > # Could not map uname=dom3 to a UUID: The object/attribute does not exist
> >
> > I tried to add some resources using cibadmin command after this,
> > but hbagent went to the segmentation fault.
> > I took its gdb and attached it.
> > It seems that myuuid for openAIS is still something wrong.
> >
> > By the way, hbagent(+ openAIS) became to be able to get LHAHBAgentOnline information with your patch.
> > Great!
> >
> > Thanks,
> > Junko
> >
> >
> > On Wed, 15 Jul 2009 21:49:05 +0900, Yan Gao <ygao at novell.com> wrote:
> >
> >> Please try the attached patch.
> >>
> >> On Wed, 2009-07-15 at 02:56 -0600, Yan Gao wrote:
> >>> Hi,
> >>>
> >>>  >>>On 7/15/2009 at  3:18 PM, "Junko IKEDA" <ikedaj at intellilink.co.jp>
> >>> wrote:
> >>> > Hi,
> >>> >
> >>> > hbagent detects whether or not the event occurred on that node if it
> >>> works
> >>> > with Heartbeat.
> >>> Indeed
> >>> > Can I get the "myuuid" of openais?
> >>> Theoretically, we could get openais nodeid. I'll revise hbagent later.
> >>>
> >>> >
> >>> >
> >>> > snmp_subagent/hbagentv2.c
> >>> >
> >>> >  280 static void
> >>> >  281 hbagentv2_update_diff(const char *event, crm_data_t *msg)
> >>> >  282 {
> >>> >
> >>> >  357 #if SUPPORT_HEARTBEAT
> >>> >  358         if (is_heartbeat_cluster()) {
> >>> >  359             if (STRNCMP_CONST(node_id, myuuid) != 0) {
> >>> >  360                 /* This change is not at my node */
> >>> >  361                 free_xml(diff);
> >>> >  362                 return;
> >>> >  363             }
> >>> >  364         }
> >>> >  365 #endif
> >>> >
> >>> > Thanks,
> >>> > Junko
> >>> >
> >>> > On Tue, 14 Jul 2009 16:18:03 +0900, Junko IKEDA
> >>> <ikedaj at intellilink.co.jp>
> >>> > wrote:
> >>> >
> >>> > > Sorry for missing attatchments.
> >>> > >
> >>> > > Thanks,
> >>> > > Junko
> >>> > >
> >>> > > On Tue, 14 Jul 2009 16:11:20 +0900, Junko IKEDA
> >>> <ikedaj at intellilink.co.jp>
> >>> > wrote:
> >>> > >
> >>> > >> Hi,
> >>> > >>
> >>> > >> I am trying to run hbagent with openais in the following way,
> >>> > >>
> >>> > >> # export HA_cluster_type="openais"
> >>> > >> # service openais start
> >>> > >> # /usr/lib64/heartbeat/hbagent -d
> >>> > >>
> >>> > >> hbagent could start successfully,
> >>> > >> but there is something wrong with it.
> >>> > >>
> >>> > >> When I run hbagent included in Heartbeat 2.1.4,
> >>> > >> trap events came from the node which the event occurred on.
> >>> > >> But with openais, all node throw the trap even if that event
> >>> didn't occur
> >>> > on it.
> >>> > >>
> >>> > >> For example,
> >>> > >> primitive resource "dummy1" and "dummy2" started on "dom1".
> >>> > >> In this case, only "dom1" was supposed to trap that these two
> >>> resources
> >>> > started on it,
> >>> > >> but "dom2" (stand-by node) also notice it.
> >>> > >> It looks like a duplicate.
> >>> > >> Please see the attached logs.
> >>> > >>
> >>> > >> Should I need any additional setups?
> >>> > >>
> >>> > >> * version
> >>> > >> openais-0.80
> >>> > >> pacemaker-1.0.4
> >>> > >> pacemaker-mgmt-1.99
> >>> > >>
> >>> > >> Thanks,
> >>> > >> Junko
> >>> > >
> >>>
> >>>
> >>
> >
> >
> > 
> 
> 
> _______________________________________________
> Pacemaker mailing list
> Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
-- 
Regards,
Yan Gao
China R&D Software Engineer
ygao at novell.com

Novell, Inc.
Making IT Work As One™
-------------- next part --------------
A non-text attachment was scrubbed...
Name: pacemaker-mgmt-snmptrap-node-0716.diff
Type: text/x-patch
Size: 7520 bytes
Desc: not available
URL: <https://lists.clusterlabs.org/pipermail/pacemaker/attachments/20090716/62913add/attachment-0005.bin>


More information about the Pacemaker mailing list