[Drbd-dev] Cstate stays in "Disconnecting" after Heartbeat link down and up

[Montrose, Ernest]

Phil,
Thanks!  I did not think that my patch was the proper fix :)  Thank you for explaining this.  This will help our understanding of the Cstate machine.

As for DRBD not noticing the link Up, I accept the current behavior.  Some external monitor should notice and wake up drbd. I wanted to confirm though.

EM--

-----Original Message-----
From: Philipp Reisner [mailto:philipp.reisner at linbit.com] 
Sent: Friday, October 06, 2006 9:19 AM
To: drbd-dev at linbit.com
Cc: Montrose, Ernest
Subject: Re: [Drbd-dev] Cstate stays in "Disconnecting" after Heartbeat link down and up

Am Donnerstag, 5. Oktober 2006 23:37 schrieb Montrose, Ernest:
> Hi all,
> If I do an ifdown and an ifup on the heartbeat link drbd goes into
> CS:Disconnecting .  I cannot manually recover until a reboot. Should the
> state be "StandAlone" instead? I include a patch that seems to work for
> me.
>

Hi Ernest,

You patch did not fix the root case of the problem. The source of the
problem was the missing dec_net() in drbd_try_connect().

That lead to the receiver thread hang in 
wait_event( ...,atomic_read(&mdev->net_cnt) == 0 ) in drbd_disconnect() .

Therefore the receiver thread did not terminate and could not set
the state finally to StandAlone.

see http://lists.linbit.com/pipermail/drbd-cvs/2006-October/001306.html

PS: Regarding the state-engine, you should always request a state
    of "Disconnecting", and let the exiting receiver set it finally
    to "StandAlone".

> A side question:
> Should drbd notice when the link comes up?  He does not seem to.
>

No, we do not do that. In case we discover that we can not bind()
our address we give up.

-phil
-- 
: Dipl-Ing Philipp Reisner                      Tel +43-1-8178292-50 :
: LINBIT Information Technologies GmbH          Fax +43-1-8178292-82 :
: Schönbrunnerstr 244, 1120 Vienna, Austria    http://www.linbit.com :