- Original Message -
From: Ulrich Windl ulrich.wi...@rz.uni-regensburg.de
To: linux-ha@lists.linux-ha.org
Sent: Monday, April 15, 2013 1:39:34 AM
Subject: [Linux-HA] Q: limiting parallel execution of resource actions
Hi!
I have a wuestion: If I want to limit parallel execution of some resources,
how can I configure that?
Background: Some resources may hit some OS bug when demanding high I/O (e.g.
starting several Xen-VMs residing on a OCFS2 filesystem that is mirrored by
cLVM). The I/O performance will drop approximately to zero when cLVM is also
mirroring. That's not because of the I/O channel being saturated, but
because of terrible programming regarding cLVM... anyway:
Would it work to define an advisory ordering of resources (in addition to a
different mandatory ordering), so if crm would schedule alle those resources
at one (in parallel), it would actually schedule the sequentially?
If I stop one resource, will those logically after the one being stopped be
also stopped due to advisory ordering then?
Maybe a new mechanism is needed to restrict parallelism based on resources
(i.e. a new type of constraint).
hey,
'batch-limit' cluster option might help.
http://clusterlabs.org/doc/en-US/Pacemaker/1.1-pcs/html-single/Pacemaker_Explained/index.html#_available_cluster_options
-- Vossel
Regards,
Ulrich
___
Linux-HA mailing list
Linux-HA@lists.linux-ha.org
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems
___
Linux-HA mailing list
Linux-HA@lists.linux-ha.org
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems