[ClusterLabs] Resource is not started immediately regardless of require-all="false"

[Daniel]

Dear Pacemaker Users,

I would like to refresh the question of order constraints using resource
sets with require-all="false" option and the way they work.
The scenario is (tried with pacemaker v 1.1.11 and 1.1.14):
- create the following order constraint:
$ pcs constraint order set dummyRes1 set dummyRes21 dummyRes22
sequential=false require-all=false set dummyRes3
- when trying to start all 4 resources, dummyRes3 is started only after
both (dummyRes21 and dummyRes22) finish starting.
Indeed - one of dummyRes21 and dummyRes22 can fail to start (and this will
not block dummyRes3 from starting), but if one of them needs 10 seconds to
start and the other one 5 minutes - dummyRes3 will start only after 5
minutes.

Has anyone found some workaround for this or is aware of any plans to
adjust the implementation so that dummyRes3 can start earlier ?

Exactly the same question was asked 3 years ago here:
http://oss.clusterlabs.org/pipermail/pacemaker/2013-September/019730.html
but looks like there was no updates since then.

Any feedback is very appreciated.

Kind regards
Daniel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.clusterlabs.org/pipermail/users/attachments/20170117/27bdb4c3/attachment-0002.html>