On Mon, 26 Nov 2012 22:06:02 -0500
"Matthew M. DeLoera" <[email protected]> wrote:
> Hello,
>
> I searched the archive but only found some discussion regarding "wait
> indefinitely for responses" from 2010. I hope this isn't a redundant question!
>
> I have a CIFS entry in my fstab to a remote server on my VPN, so it's not
> available when I'm not connected to my VPN. When not connected, mount -a
> hangs for about 5 minutes before finally failing with "connection timed out".
> If I connect my VPN in the middle of that wait, and verify it's connected,
> mount still hangs for that entire timeout before failing. If I try again, it
> succeeds.
>
> Is there any issue with kill -9 for the mount -a process, in order to more
> quickly retry? I was planning to automate it, and to keep retrying until
> mount succeeds.
>
> Basically, I need quicker failure than 5 minutes, and I can't find any CIFS
> timeout settings. Any suggestions would be greatly appreciated!
>
> Thanks,
> - Matthew
>
>
>
>
What kernel is this? kill -9'ing the process likely won't hurt
anything, but it's not really a great solution.
That doesn't sound related to the fact that we wait indefinitely for
responses since the connection is presumably not even established yet.
cifs.ko is a fairly naive user of the socket APIs in the kernel and
does a blocking connect call to connect to the socket. That said, it
should be timing out a lot more quickly than that:
socket->sk->sk_rcvtimeo = 7 * HZ;
socket->sk->sk_sndtimeo = 5 * HZ;
A connect attempt shouldn't be hanging for 5 minutes. I'd suggest doing
a bit more investigation -- track down the [cifsd] kthread and see what
it's doing at the time. Something like:
# cat /proc/$(pidof cifsd)/stack
--
Jeff Layton <[email protected]>
--
To unsubscribe from this list: send the line "unsubscribe linux-cifs" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
