[Pacemaker] cloned ping ressource with multiple target hosts fails

[Dejan Muhamedagic]

Hi,

On Tue, Jan 26, 2010 at 03:55:16PM +0100, Moritz Krinke wrote:
> Hello,
> 
> i'm trying to configure a 2-node cluster with one resource (ClusterIP) which should only be allowed to run if a cloned ping resource succeeds.
> After reading the pacemaker configuration explained document and some WIKI entries i came up with the following configuration:
> 
> node centos5srv1
> node centos5srv2
> primitive ClusterIP ocf:heartbeat:IPaddr2 \
>         meta target-role="started" is-managed="true" \
>         operations $id="ClusterIP-operations" \
>         op monitor interval="1s" timeout="5s" start-delay="0" \
>         params ip="192.168.0.65" cidr_netmask="32" nic="eth0"
> primitive ping ocf:pacemaker:ping \
>         params dampen="5s" multiplier="1000" host_list="192.168.0.60 10.100.64.50" name="pingval" \
>         operations $id="ping-operations" \
>         op monitor interval="10s" timeout="5s"
> clone Connected ping
> location ClusterIP_on_Connected ClusterIP \
>         rule $id="ClusterIP_on_Connected-rule" -inf: pingval lt 2000
> property $id="cib-bootstrap-options" \
>         dc-version="1.0.7-d3fa20fc76c7947d6de66db7e52526dc6bd7d782" \
>         cluster-infrastructure="openais" \
>         expected-quorum-votes="2" \
>         no-quorum-policy="ignore" \
>         stonith-enabled="false"
> rsc_defaults $id="rsc_defaults-options" \
>         resource-stickiness="100"
> 
> 
> but this does not work, the ClusterIP resource wont get started at all.
> after i removed all but one ping-target from the configuration, (and the constraint accordingly) it works.
> 
> primitive ping ocf:pacemaker:ping \
>         params dampen="5s" multiplier="1000" host_list="192.168.0.60" name="pingval" \
>         operations $id="ping-operations" \
>         op monitor interval="10s" timeout="5s"
> clone Connected ping
> location ClusterIP_on_Connected ClusterIP \
>         rule $id="ClusterIP_on_Connected-rule" -inf: pingval lt 1000

Not sure, but perhaps you ran into a string/integer issue, i.e.
by default the comparison is done as strings. That may have
changed later, but don't know when. At any rate, you can try
with "number:lt" instead of lt. 

Thanks,

Dejan

> as i understand the documentation i am allowed to put multiple targets into the "host_list" directive, and the value of "pingval" contains the number of reached hosts multiplied with the value of "multiplier"?
> 
> in the log i see errors complaining about a failing ping:
> Jan 26 15:42:15 centos5srv2 lrmd: [21076]: WARN: ping:0:monitor process (PID 885) timed out (try 1).  Killing with signal SIGTERM (15).
> Jan 26 15:42:15 centos5srv2 ping[885]: INFO: They use TERM to bring us down. No such luck.
> Jan 26 15:42:15 centos5srv2 ping[885]: ERROR: Unexpected result for 'ping -n -q -W 5 -c 5  10.100.64.50' 143: 
> Jan 26 15:42:15 centos5srv2 attrd_updater: [909]: info: Invoked: attrd_updater -n pingval -v 1000 -d 5s 
> Jan 26 15:42:15 centos5srv2 lrmd: [21076]: info: RA output: (ping:0:monitor:stderr) 2010/01/26_15:42:15 INFO: They use TERM to bring us down. No such luck. 2010/01/26_15:42:15 ERROR: Unexpected result for 'ping -n -q -W 5 -c 5  10.100.64.50' 143:  
> 
> 
> when i issue the stated ping cmd "ping -n -q -W 5 -c 5  10.100.64.50" myself, it succeeds.
> 
> 
> i probably could create a cloned ping resource for every host i want to have checked and than write corresponding rules for the location constraint, but this would result in a much more complex configuration that is needed as far as i understood the docs.
> 
> Any Help/Suggestions are appreciated :-)
> 
> Using:
> centos5.4 x86_64
> corosync-1.2.0-1.el5
> openais-1.1.0-1.el5
> pacemaker-1.0.7-2.el5
> 
> 
> Thanks & Regards,
> Moritz
> _______________________________________________
> Pacemaker mailing list
> Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker