Mailing List Archive

On 4/11/2011 10:09 AM, Florian Haas wrote:
> Hi everyone,
>
> our last in-person meeting at LPC is now 5 months past, and I'm making
> an effort to see where we stand. Reference is the EtherPad that we
> created during the presentation; see
> http://etherpad.osuosl.org/lpc2010-high-availability-clustering
>
> * Definitions: web site glossary -- my bad, this has totally slipped.
> Where should this go? Linux-HA wiki or does someone have a better
> suggestion?
>
> * Kronosnet: Fabio, can you comment on the status here?

It is proceeding at slow speed, but the code is growing and since there
is no time pressure on this front, we are trying to do it right, rather
than just gluing code and hope it works. Code is up on
http://kronosnet.org/ together with mailing lists and stuff.

It is usable for basic testing, might not yet satisfy cluster
requirements in terms of features (active/active links are missing,
active/passive works) and user interface (it?s difficult to configure at
the moment and the CLI is a bit behind schedule).

> * How to do multi site: Andrew/Lars/Fabio, any updates on CIB
> replication and CTR?

No progress here atm.

Fabio

On 4/11/2011 at 06:09 PM, Florian Haas <florian.haas at linbit.com> wrote:
> * Language bindings: Tim, I believe you were planning to tackle that
> one. Any news on that?

I was going to look at doing SWIG bindings for libcib/libpengine to
make interfacing management tools (CLI, GUIs) with Pacemaker better,
stronger and/or faster. I don't yet have anything to report on that
front.

The etherpad mentions language bindings for RAs, to make it easier
to author them in languages other than shell, but I think that's
different than what I was talking about...?

> * CTDB/Samba integration with Pacemaker: Tim?

I've added ctdb_manages_samba and ctdb_manages_winbind parameters
to the CTDB RA, and done some other cleanup. The idea here is that
right now, you can turn these on and CTDB still manages Samba and
Winbind, but later (after we're all using the merged RAs, and have
the Samba RA available, and an as-yet unwritted Winbind RA), you'd
turn these off and have separate Samba and Winbind resources.

Bonus: being able to set ctdb_manages_winbind=no means joining an
AD domain isn't as horrible as it used to be (i.e. it's now actually
possible to do without hacking the RA).

Documentation at http://linux-ha.org/wiki/CTDB_%28resource_agent%29
pretty much reflects the current state of things, but is lacking in
examples (in particular I need to add some notes about using Active
Directory, get rid of the Clustered IP bit, and add something about
using ldirectord instead).

Regards,

Tim


--
Tim Serong <tserong at novell.com>
Senior Clustering Engineer, OPS Engineering, Novell Inc.

On Mon, Apr 11, 2011 at 10:09 AM, Florian Haas <florian.haas at linbit.com> wrote:
> Hi everyone,
>
> our last in-person meeting at LPC is now 5 months past, and I'm making
> an effort to see where we stand. Reference is the EtherPad that we
> created during the presentation; see
> http://etherpad.osuosl.org/lpc2010-high-availability-clustering
>
> * Definitions: web site glossary -- my bad, this has totally slipped.
> Where should this go? Linux-HA wiki or does someone have a better
> suggestion?
>
> * Kronosnet: Fabio, can you comment on the status here?
>
> * Language bindings: Tim, I believe you were planning to tackle that
> one. Any news on that?
>
> * CTDB/Samba integration with Pacemaker: Tim?
>
> * Community engagement: RA dev guide is done.
>
> * OCF RA schema/spec updates: Lars?
>
> * RA merge: Done as far as the repos are concerned. How do we proceed?
>
> * Storage based fencing: Lars, any updates on SBD? SBD/qdisk merge, any
> progress on that?
>
> * How to do multi site: Andrew/Lars/Fabio, any updates on CIB
> replication and CTR?

I don't recall agreeing to do CIB replication, just the PE bits that
make use of CTR attributes.
No progress on those though.

On 2011-04-11T10:09:01, Florian Haas <florian.haas at linbit.com> wrote:

> * Storage based fencing: Lars, any updates on SBD? SBD/qdisk merge, any
> progress on that?

SBD has just merged support for 2 or 3 devices, courtesy of Xin Wei and
mangled by myself. I don't think the sbd/qdisk merge made progress.

> * How to do multi site: Andrew/Lars/Fabio, any updates on CIB
> replication and CTR?

It's on the roadmap for us - i.e., the SUSE team - this year. It's been
disclosed, but not much has happened yet. I think Yan will be working on
the pacemaker features (deadman dependency, cluster-wide attributes),
and then we'll work on the rest.

> * OCF RA schema/spec updates: Lars?

I swear to god, it is next on my list. Also because we need it for
SUSE's next release cycle. (Which sort-of affects when I get to work on
stuff, unfortunately.)



Regards,
Lars

--
Architect Storage/HA, OPS Engineering, Novell, Inc.
SUSE LINUX Products GmbH, GF: Markus Rex, HRB 16746 (AG N?rnberg)
"Experience is the name everyone gives to their mistakes." -- Oscar Wilde