I have a SLES 11SP3 setup with pacemaker and ocfs2. I’m currently trying to understand the entire setup (didn’t install it) and ran into some issues understanding, where the mount-options of the ocfs2 come from. As far as I understand pacemaker all the configuration is done by the corresponding Ressource-Agents. In this case the ocf:heartbeat:Filesystem which lists “acl,noatime” in the crm configuration as file system options, which is fine.
However looking into /proc/mounts I see (of course) much more info, which seems to be ocfs2 specific. e.g.
localalloc=2,cluster_stack=pcmk,coherency=full
Now to my question: I understand the parameters and I’m aware of what they are used from/why necessary. What I don’t know is, where are theses parameters set besides of the crm configuration? Are they set automatically (by which daemon or script)?
It would be great if someone could lighten me up a bit.
On 06/17/2014 03:44 AM, jhaemmer wrote:[color=blue]
I have a SLES 11SP3 setup with pacemaker and ocfs2. I’m currently trying
to understand the entire setup (didn’t install it) and ran into some
issues understanding, where the mount-options of the ocfs2 come from. As
far as I understand pacemaker all the configuration is done by the
corresponding Ressource-Agents. In this case the
ocf:heartbeat:Filesystem which lists “acl,noatime” in the crm
configuration as file system options, which is fine.
However looking into /proc/mounts I see (of course) much more info,
which seems to be ocfs2 specific. e.g.
localalloc=2,cluster_stack=pcmk,coherency=full
Now to my question: I understand the parameters and I’m aware of what
they are used from/why necessary. What I don’t know is, where are theses
parameters set besides of the crm configuration? Are they set
automatically (by which daemon or script)?[/color]
I am guessing they are automatic for a couple of reasons:
coherency=full is a fault per the mount.ocfs2 manpage.
I also have cluster_stack=pcmk on my system. The full line from my
/proc/mounts “file” follows:
Checking the ‘mount’ command output most of this is hidden. I do not
know how the localalloc=2 setting was implemented in yours, but perhaps
it’s a default based on versions.
Looking on my system I also see a reference to ‘pcmk’ in
/usr/lib/ocf/resource.d/ocfs2/o2cb so maybe that will help you.
–
Good luck.
If you find this post helpful and are logged into the web interface,
show your appreciation and click on the star below…
thanks for your reply. The o2cb-Resource Agent helped a bit
However I’m still stuck with the localalloc=2, which to me is currently the most important one.
So anyone any ideas where this option is/can be set appart from the CRM configuration
or are there any procedures which set this option automatically?
thanks for your reply. The o2cb-Resource Agent helped a bit
However I’m still stuck with the localalloc=2, which to me is currently the most important one.
So anyone any ideas where this option is/can be set appart from the CRM configuration
or are there any procedures which set this option automatically?[/QUOTE]
as far as I can tell, the localalloc parameter is determined dynamically by the FS driver (unless specified explicitly during mount). If you want to override the value, do so in the CIB.
Hi Jens,
thanks for your reply. This definitely supports my opservations from today. Without any interaction the localalloc changed from “2” to “3” and back on two different nodes. So I guess this explains it
