waiting for MNT_DETACH

waiting for MNT_DETACH

[Ian Turner]

Hello all,

I administer a two-node high-availability cluster using DRBD and heartbeat. 
One of my concerns is that at failover time, the replicated filesystem might 
be in use. I can stop the official services that use the filesystem, but some 
other transient scripts might have reason to probe it for a second or two.

umount -l seems like the answer to this question, in that one can ensure no 
new accesses will be made to the filesystem, while those that exist can take 
their time to finish. The catch is, there does not appear to be a way to tell 
if the unmount has really completed or not.

So, here's the question: Is there a way to check if a call to umount2(..., 
MNT_DETACH) has completed or not?

Thanks,

--Ian Turner

Re: waiting for MNT_DETACH

[Ingo Oeser]

Hi Ian,

On Sunday, 2. July 2006 20:51, Ian Turner wrote:
> I administer a two-node high-availability cluster using DRBD and heartbeat. 
> 
> So, here's the question: Is there a way to check if a call to umount2(..., 
> MNT_DETACH) has completed or not?

Check the usage count (refcount) of the DRBD module. It should be as low as
you would expect with a normal umount.

"cat /sys/module/drbd/refcnt" should be the right command.


Regards

Ingo Oeser