[DRBD-user] Error using SDP in DRBD8.4

[Nuno Fernandes]

A Quarta, 21 de Outubro de 2015 13:09:07 Lars Ellenberg escreveu:
> On Sat, Oct 17, 2015 at 11:41:22AM +0100, Nuno Fernandes wrote:
> > Hello,
> > 
> > We had drbd 8.3 replicating using mellanox infiniband cards using SDP. It
> > worked fine. After upgrading to 8.4 the SDP replication doesn't work.
> > Using plain IP over Infiniband works but using sdp i get the following
> > logs in the "secondary host":
> > 
> > Oct 17 11:36:27 s2 -bash: (4415) [root.root] |.| /etc/init.d/drbd start
> > Oct 17 11:36:27 s2 kernel: drbd: applying 16k kernel stack fix up
> 
> Uh?

Don't know.... Found this link http://elrepo.org/bugs/view.php?id=583 where it 
reports:

"A quick google search seems to suggest this related to the expanded kernel 
stack size (to 16k) in kernel-3.16. So I guess we can assume RedHat have 
backported that to their 6.7 kernel.

Doesn't seem to cause any issues AFAICS, but it's probably not a bad idea to 
rebuild modules against the 6.7 kernel. We currently build nvidia packages 
against the RHEL-6.5 kernel for increased backward compatibility."

 
> > Oct 17 11:36:28 s2 kernel: drbd infiniband: conn( StandAlone ->
> > Unconnected ) Oct 17 11:36:28 s2 kernel: drbd infiniband: Starting
> > receiver thread (from drbd_w_infiniba [21676]) Oct 17 11:36:28 s2 kernel:
> > drbd infiniband: receiver (re)started
> > Oct 17 11:36:28 s2 kernel: drbd infiniband: conn( Unconnected ->
> > WFConnection ) Oct 17 11:36:39 s2 kernel: drbd infiniband: sock_recvmsg
> > returned -11 Oct 17 11:36:39 s2 kernel: drbd infiniband: conn(
> > WFConnection -> BrokenPipe ) Oct 17 11:36:39 s2 kernel: drbd infiniband:
> > short read (expected size 8) Oct 17 11:36:39 s2 kernel: drbd infiniband:
> > Connection closed
> > Oct 17 11:36:39 s2 kernel: drbd infiniband: conn( BrokenPipe ->
> > Unconnected ) Oct 17 11:36:40 s2 kernel: drbd infiniband: conn(
> > Unconnected -> WFConnection )
> > 
> > I'm using latest centos 6 with 2.6.32-573.7.1.el6.x86_64 kernel and drbd84
> > from elrepo.
> Last time I checked, it "worked for me".

It works with drbd 8.3. How can i continue in the debug process? Is there any 
loglevel i can increase in drbd kernel module?


> > Oct 17 11:36:54 s2 multipathd: drbd0: remove path (uevent)
> 
> Multipath on top of DRBD? really?

I'm not using multipath on top of drbd. Will exclude it from 
/etc/multipath.conf so that this log doesn't appear.

Thanks
Nuno Fernandes