[Linux-HA] cluster-delay and action timeouts

[Keisuke MORI]

"Andrew Beekhof" <beekhof at gmail.com> writes:
> On 2/23/07, Keisuke MORI <kskmori at intellilink.co.jp> wrote:
>> Hi,
>>
>> If I configured an action timeout to quite a long time,
>> my guess is that cluster-delay (or transition_idle_timeout in 2.0.7)
>> must be a larger value than it.
>> Is this correct?
>
> shouldn't be.  the action's timeout and the cluster delay are
> (supposed to be) added together to produce the actual value used.
>
> from what you're seeing, this is not happening or at least is getting confused
>
> can you send me the complete logs please (zipped if possible)

Attached.

It includes two sets of cib.xml/ha-debug and a modified Dummy script.

  - cib.xml-action-timeout-2-OK / ha-debug-2 ; worked as expected.
    start timeout: 70s, cluster-delay: 120s

  - cib.xml-action-timeout-3-NG / ha-debug-3 ; not worked as expected
    start timeout: 70s, cluster-delay: 60s

  - Dummy-test; the modified Dummy to induce start timeout.


When I used cib.xml-action-timeout-3-NG, start seems to be
called 3 times and finally moved to the secondary node.

I'm running 2.0.8(stable release) on RedHat 4.

>
>> (or documented somewhere?)
>
> only in "pengine metadata" so far

Thanks.
It mentions that it excludes the action execution as you said...
 
Keisuke MORI
NTT DATA Intellilink Corporation
-------------- next part --------------
A non-text attachment was scrubbed...
Name: cluster-delay-20060226.tar.gz
Type: application/octet-stream
Size: 22317 bytes
Desc: not available
Url : http://lists.community.tummy.com/pipermail/linux-ha/attachments/20070226/612d397c/cluster-delay-20060226.tar-0001.obj