[dm-devel] [PATCH 77/78] multipathd: asynchronous configuration

[Hannes Reinecke]

On 03/27/2015 06:58 AM, Benjamin Marzinski wrote:
> On Mon, Mar 16, 2015 at 01:37:04PM +0100, Hannes Reinecke wrote:
>> For initial configuration multipathd waits it has synchronized
>> with the existing setup. On larger systems this takes up quite
>> some time (I've measured 80 seconds on a system with 1024 paths)
>> causing systemd to stall and the system to fail booting.
>> This patch makes the initial configuration asynchronous, and
>> using the same codepath as the existing 'reconfigure' CLI
>> command.
> 
> If the issue is that configure takes too long, and is keeping multipath
> from resetting the watchdog timer, can't this problem still happen?
> 
> reconfigure still is called under the vecs lock, and checkerloop still
> locks the vecs lock, so a call to reconfigure can still stall the
> checker loop for just as long.
> 
> Or am I missing something?

The problem here is that on large configurations multipathd might
spend too much time in configure(), so the sd_notify() call with
'RUNNING' is would be later than the default systemd service timeout.
Hence multipathd will be killed by systemd with a job timeout, with
no chance to get it to run, ever.

Cheers,

Hannes
-- 
Dr. Hannes Reinecke		               zSeries & Storage
hare at suse.de			               +49 911 74053 688
SUSE LINUX GmbH, Maxfeldstr. 5, 90409 Nürnberg
GF: F. Imendörffer, J. Smithard, J. Guild, D. Upmanyu, G. Norton
HRB 21284 (AG Nürnberg)