[DRBD-user] Upgrade from 0.7.15 to 0.7.21, advise to resync...

[Lars Ellenberg]

/ 2006-09-18 09:14:38 +0200
\ Sim:
> Hi,
> 
> I have update drbd from 0.7.15 to 0.7.21, in the beginning to second
> and then to the first.
> 
> This is log after upgrade.
> 
> Sep 15 13:20:38 mailing kernel: drbd0: drbd0_receiver [742]: cstate WFConnection --> WFReportParams
> Sep 15 13:20:38 mailing kernel: drbd0: Handshake successful: DRBD Network Protocol version 74
> Sep 15 13:20:38 mailing kernel: drbd0: Connection established.
> Sep 15 13:20:38 mailing kernel: drbd0: I am(S): 1:00000003:00000002:00000030:00000001:01
> Sep 15 13:20:38 mailing kernel: drbd0: Peer(S): 1:00000003:00000002:00000031:00000001:10
> Sep 15 13:20:38 mailing kernel: drbd0: drbd0_receiver [742]: cstate WFReportParams --> WFBitMapT
> Sep 15 13:20:38 mailing kernel: drbd0: Secondary/Unknown --> Secondary/Secondary
> Sep 15 13:20:38 mailing kernel: drbd0: drbd0_receiver [742]: cstate WFBitMapT --> SyncTarget
> Sep 15 13:20:38 mailing kernel: drbd0: Resync started as SyncTarget (need to sync 1052672 KB [263168 bits set]).
> Sep 15 13:20:50 mailing kernel: drbd0: Secondary/Secondary --> Secondary/Primary
> Sep 15 13:23:02 mailing kernel: drbd0: Resync done (total 143 sec; paused 0 sec; 7360 K/sec)
> Sep 15 13:23:02 mailing kernel: drbd0: drbd0_worker [3718]: cstate SyncTarget --> Connected
> 
> 
> 0: cs:SyncSource st:Primary/Secondary ld:Consistent
>    ns:705692 nr:0 dw:8240 dr:708393 al:222 bm:408 lo:0 pe:492 ua:0 ap:7
>        [=============>......] sync'ed: 66.3% (356844/1052672)K
>        finish: 0:00:48 speed: 7,336 (7,324) K/sec
> 
> 
> Sync, about 1052672K, has take less than one minutes. Is it possible?

I see
 Resync done (total 143 sec; paused 0 sec; 7360 K/sec)
so it took 2 minutes 23 seconds, average 7 MB/sec,
and resynced about one GB of data in total.

> I haven't moved this amount of data.

since 1052672 (the reported "need to sync" amount) is _exactly_
257 * 4096, which is suggested "al-size" (config parameter) times
al-extent-size (4MB, hardcoded), I suspect you "crashed" your primary,
or that at least to drbd it looked like such a crash, anyways.

> Is it advise to force resync from master to slave?

that you have to decide for yourself.

> If I do it, how I have to do it?

you go to the node you want to become sync target,
and tell it to "drbdadm invalidate all".
will cause a full resync.

-- 
: Lars Ellenberg                                  Tel +43-1-8178292-0  :
: LINBIT Information Technologies GmbH            Fax +43-1-8178292-82 :
: Schoenbrunner Str. 244, A-1120 Vienna/Europe   http://www.linbit.com :
__
please use the "List-Reply" function of your email client.