[Pacemaker] stop/demote order during node standby

Andrew Beekhof andrew at beekhof.net
Mon Sep 13 13:04:33 UTC 2010


On Mon, Sep 13, 2010 at 8:30 AM, Patrick Zwahlen <paz at navixia.com> wrote:
> Hi all (already posted to linux-ha, but this seems a better place),
>
> I am building a 2-nodes iSCSI target, using the following:
> - Fedora 12 x86_64
> - drbd 8.3.8.1 (RPM built from source)
> - pacemaker 1.0.9.1 (clusterlabs RPM)
> - corosync 1.2.7 (clusterlabs RPM)
> - scst/iscsi-scst rev 2071 (compiled from source)
>
> I have built a custom RA (attached) that handles device/target/lun
> configuration. My pacemaker CIB defines a drbd MS resource, as well as a
> group containing my iSCSI target resource and a floating IP. Everything
> is configured x2 for load sharing between the two nodes (see my
> pacemaker CIB attached).
>
> I have defined simple drbd:promote->group:start order constraints as
> explained is the many howtos. I can successfully migrate/unmigrate my
> group, and in that case I see group:stop->drbd:demote in the right
> order.
>
> However, if I issue a "node standby", then both actions happen at the
> same time, and in most cases, my group:stop fails because the underlying
> (drbd) device has been removed already.
>
> Andrew mentions in this thread
> (http://www.mail-archive.com/pacemaker@oss.clusterlabs.org/msg05629.html
> ) that this is the normal behavior, because pacemaker hasn't been told
> to behave differently.

I think I was wrong in that thread, until now I didn't see the
dangling appyul1:start and thought it wasn't involved in any ordering
constraints.

>
> Question is: How to tell pacemaker how resources should be
> stopped/demoted during a node standby ?

Your attached config looks right to me.
Could you include a hb_report for this scenario and I'll take a closer look.


> I have tried these 2 order constraints, with no luck:
>  order ord_grp_iscsi_a_before_drbd0_1 inf: grp_iscsi_a:stop
> ms_drbd0:demote
>  order ord_grp_iscsi_a_before_drbd0_2 inf: grp_iscsi_a:stop
> ms_drbd0:stop
>
> Any help appreciated. Thanks a ton, cheers. - Patrick -
>
>
> **************************************************************************************
> This email and any files transmitted with it are confidential and
> intended solely for the use of the individual or entity to whom they
> are addressed. If you have received this email in error please notify
> the system manager. postmaster at navixia.com
> **************************************************************************************
>
> _______________________________________________
> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>
> Project Home: http://www.clusterlabs.org
> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
> Bugs: http://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker
>
>




More information about the Pacemaker mailing list