Am 30.09.2010 09:49, schrieb Jonathan Petersson:
> deb-fw1:~# ocf-tester -L -n conntrackd
> /usr/lib/ocf/resource.d/heartbeat/conntrackd
> Using lrmd/lrmadmin for all tests
> Beginning tests for /usr/lib/ocf/resource.d/heartbeat/conntrackd...
> * rc=0: Monitoring a stopped resource should return 7
> * Your agent does not support the notify action (optional)
> * Your agent does not support the demote action (optional)
> * Your agent does not support the promote action (optional)
> * Your agent does not support master/slave (optional)
> * rc=0: Monitoring a stopped resource should return 7
> * rc=0: Monitoring a stopped resource should return 7
> * rc=0: Monitoring a stopped resource should return 7
> Tests failed: /usr/lib/ocf/resource.d/heartbeat/conntrackd failed 4 tests
>
> Not 100% sure codewise what I should change.
>
> Maybe it would make sense to add master/slave to handle the
> state-transition and have start/stop handle whether the daemon is
> alive or not. However I'm not entirely sure how the crm configuration
> statement for this should look given that the daemon should be started
> on both nodes running in either master or slave mode, do you have an
> example for this?
Hi,
without having had a look onto your RA and its necessary parameters etc.
at all yet:
I had similiar issues with my Tomcat6- and Apache-RAs, though they
worked without problems in the cluster. When I had the time and ambition
to finally also make them ocf-tester-compliant (meanwhile they are
anyway, without special additional handlers for the ocf-tester), I could
track problems down to various timing issues. Not only that back at that
time I still used start-delay for the monitor-op in the cluster (because
back then I didn't check and wait for i.e. Tomcat being operational in
start-action respectively not anymore and all files being closed in
stop-action yet), but also exspecially that ocf-tester didn't set i.e. a
$OCF_RESKEY_CRM_meta_timeout, on which my RA back then still depended on
for its monitor-action (now it's not anymore, but back then I was still
eager about allways cleanly exiting with at least an OCF_ERR_* too,
instead of just simply letting it time out and having Pacemaker handle
that situation (-; ).
Also - since I don't know about your RA's parameters, but at least for
mine it helped, due to some not-up-to-date-anymore-defaults - it might
be necessary to tell ocf-tester to pass over some OCF-parameters, to
have them available as $OCF_RESKEY_*. You do that with the "-o"-option
to ocf-tester, like
-o log_level="debug"
to have ocf-tester set an env.-var. $OCF_RESKEY_log_level with the value
"debug".
This was my complete command for calling ocf-tester back then, long time
ago (i.e. still includes my switch for special ocf-tester-handler,
etc.): (-;
ocf-tester -n testTom -o log_debug="/tmp/debug-tomcat.log" -o
monitor_timeoutPerTry="5" -o monitor_precheckDB="1" -o log_level="debug"
-o stopith_enabled="1" -o
monitor_url="http://localhost:8080/opencms/opencms/test/cluster.html"; -o
agent_timebuffer="4000" -o ocftester="y" /usr/lib/ocf/resource.d/cj/tomcat
Hope that helps you a little on making ocf-tester accept your RA too. (-;
Cnut Jansen
_______________________________________________________
Linux-HA-Dev: Linux-HA-Dev@lists.linux-ha.org
http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev
Home Page: http://linux-ha.org/
