Re: [gentoo-cluster] CLVM on amd64 profile

[Brian Kroth]

Giacomo Bagnoli <g.bagnoli@××××××.com> 2009-05-18 17:06:
> Brian,
> first of all thanks for your answer
>
> Il giorno lun, 18/05/2009 alle 09.29 -0500, Brian Kroth ha scritto:
> > I remember looking at this a while ago as well and being fairly
> > confused.  I'm fairly certain that lvm2 with the use flags you mention
> > is the way to go, so you might have to unmask them in your
> > /etc/portage/profile/use.mask (add -clvm, -cman).
> >
> > OCFS2 generally uses it's own different cluster stack than CMAN (which
> > uses openais).  Since I believe 2.6.26 you can use the other one, but I
> > haven't found it to be as stable.
>
> I didn't understand here what you mean. Which one is "the other one"?
> (maybe it's just my english :P )

Since 2.6.26, OCFS2 has two cluster stacks available to it:
- ocfs2_stack_o2cb:  which is the original stack
- ocfs2_stack_user:  which is the one that requires cman and openais

> >   It also requires quite a bit of extra
> > configuration for proper fencing, whereas that's built into the OCFS2
> > stack.  Since you're doing this with DRBD it sounds like you're trying
> > to do it on the cheap so proper fencing probably isn't available
> > anyways.
>
> Yes, you're right. The only way I have to get a sort of STONITH (if it
> can be used, I've still to check this) is IPMI which is working well so
> far.

I believe that will work just fine.

> > If you aren't doing anything with LVM that needs to be cluster aware
> > (eg: just a concatenation of multiple PVs, no striping) then it should
> > be just fine without CLVM.  The key is that you /must/ take all but one
> > node offline (umount; vgchange -an) before you do any LVM operations
> > such as adding more extents, else they'll have an inconsistent view of
> > the VGs/LVs.  This also isn't technically supported but I've run it
> > without problems for a year or so now.
>
> The problem is that a node will use (mount) some LVs and the other one
> some others. So I can't do the fast way as you suggest, as I need to
> stop all services that use data on LVs before unmounting.

But, with OCFS2 you can (even temporarily) have a single node mount all
the LVs, move your service IPs over, and _then_ umount and stop them on
the other one while you do your changes and then remount them all.
That's the great things amount a clustered fs - more than one node can
play in the sandbox without problems.

Also, I'm gonna put a plug in for Heartbeat [1] for service management.
It's much more powerful than rgmanager in my opinion.  If you're dead
set on using the openais cluster stack you can also use pacemaker [2].

The model I typically use for Heartbeat, is to have it run a couple of
dummy monitoring scripts as resource agents clones that each set an
attribute as to the service's health on a machine.  Then using the value
of that attribute I setup rules to move service IPs to the "healthiest"
node, where your definition of "healthiest" is completely dependent upon
your monitoring scripts.

Cheers,
Brian

[1] http://www.linux-ha.org/
[2] http://clusterlabs.org/wiki/Main_Page