posix locking on OCFS2

[steve]

On 21/11/14 19:28, Richard Sharpe wrote:
> On Fri, Nov 21, 2014 at 5:06 AM, steve <steve at steve-ss.com> wrote:
>> On 21/11/14 13:46, Ralph Böhme wrote:
>>>
>>> You just need a userspace cluster stack:
>>>
>>
>> OK. So is CTDB a 'Userspace cluster stack'? I don't think it has anything in
>> the kernel does it? But, 'stack'? Why stack? Why do we add the word 'stack'
>> to so many terms these days? Stack.
>
> Sounds like they are saying you need to run Corosync and their
> lock-manager daemon that uses the Closed Process Group abstraction ...
>
> And, upon reading this link:
> http://comments.gmane.org/gmane.comp.file-systems.ocfs2.user/4641
>
> I find that it says:
>
> "fcntl() locking does not pair at all with dlm locking. So both ocfs2 and gfs2
> use a special scheme just for this lock. Both use the ordered messaging
> provided by the corosync cluster engine. "
>

I wonder. I still wonder what 'stack' is. We have been up on 2 nodes 
with ctdb, ocfs2, drbd and 4.1.7 for around 3 months. We've had a sb 
only once which the drbd config mailed me about and we were able to 
recover from although I had to go to the box and choose the node myself. 
Now it seems we must have Corosync for ocfs2, I suspect that even though 
it seems to be working OK, our (dare I say) stack is wrong.

Does anyone know?
Cheers and have a great w/e,
Steve