[prev in list] [next in list] [prev in thread] [next in thread] 

List:       linux-ha-dev
Subject:    Re: [Linux-ha-dev] Status o2cb RA
From:       Lars Marowsky-Bree <lmb () suse ! de>
Date:       2008-01-11 19:55:38
Message-ID: 20080111195538.GZ16123 () marowsky-bree ! de
[Download RAW message or body]

On 2008-01-11T08:40:23, Serge Dubrouski <sergeyfd@gmail.com> wrote:

> > The "monitor" function is actually implemented as-intended at this
> > stage. There's nothing to monitor, and it shouldn't be run with a
> > periodic monitor.
> 
> It looks like that:
> 
> o2cb_monitor()
> {
> #       o2cb_init
>         exit $OCF_NOT_RUNNING
> }
> 
> 
> That means that CRM will never see it as running, right?

Correct. It doesn't have to - it will know its started after "start",
anyway. This is more-or-less a stateless resource - one could argue that
we could treat it as a pseudo-resource, but it would add no (functional)
value, it's not supposed to be monitored at runtime.

As I've said, it's mostly a (functional) proof-of-concept.

> > It's not supposed to offline anything itself, it just calls the init
> > script (assuming that the path wasn't wrong, that is).
> The problem is that /etc/init.d/o2cb stop fails of cluster wasn't
> offlined earlier. So in current version RA doesn't stop OCFS2

That is correct. You need to use it in conjunction with the Filesystem
RA to manage (umount, in this case) the resources on top - there's no
way for o2cb to forcibly offline the cluster.

> > > Unfortunately I'm not familiar with openAIS. Does it mean that
> > > Heartbeat won't support OCFS2? I mean combination of CRM + openAIS
> > > will work with OCFS2 but CRM + Heartbeat won't.
> > Right - there'll be no need for an agent such as o2cb with the new
> > stack, and probably the Filesystem RA will be able to drop most of the
> > special cases for the cluster filesystems as well.
> Does it mean that you are stopping Heartbeat support? That's not good at all.

Hm? OCFS2 will be fully supported.

It just so happens that some future version of OCFS2 will use a
different membership layer, openAIS, directly - and as Pacemaker will
use the same, less complexity will be needed to manage it, and so will
require a newer generation Filesystem RA (and no more o2cb RA).

This is part of the general convergence of the Linux cluster efforts;
the plus is that this future stack will support GFS2 and ClusterManager
as well, so no more mandatory combination of cluster file systems to
resource managers.

PaceMaker will continue to support running on top of heartbeat and
openAIS for awhile as well - only when you then want the new version of
OCFS2 you'll also need the new cluster infrastructure. (Or code the
necessary module for marrying it to the old version as well.)

For current RHEL or SLES customers, their current versions and
combinations will be continued to supported, migration paths will be
offered (SUSE for example is committed to a fully automatic conversion
of a cluster on upgrade to SLE11), and all will be well.

The unification of the Linux Open Source cluster stack is generally
considered a most happy thing. It implies more choice, more
standardization, more synergy, less waste. Customers and partners love
it, which is why RHAT, Oracle, Ubuntu, Novell, and SUSE are fully
backing it, and even IBM seems to like it quite a bit.


Regards,
    Lars

-- 
Teamlead Kernel, SuSE Labs, Research and Development
SUSE LINUX Products GmbH, GF: Markus Rex, HRB 16746 (AG Nürnberg)
"Experience is the name everyone gives to their mistakes." -- Oscar Wilde

_______________________________________________________
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/
[prev in list] [next in list] [prev in thread] [next in thread]