[Pacemaker] drbd under pacemaker - always get split brain

Andrew Beekhof andrew at beekhof.net
Tue Jul 10 22:50:57 EDT 2012 

On Wed, Jul 11, 2012 at 8:06 AM, Andreas Kurz <andreas at hastexo.com> wrote:
> On Tue, Jul 10, 2012 at 8:12 AM, Nikola Ciprich
> <nikola.ciprich at linuxbox.cz> wrote:
>> Hello Andreas,
>>> Why not using the RA that comes with the resource-agent package?
>> well, I've historically used my scripts, haven't even noticed when LVM
>> resource appeared.. I switched to it now.., thanks for the hint..
>>> this "become-primary-on" was never activated?
>> nope.
>>
>>
>>> Is the drbd init script deactivated on system boot? Cluster logs should
>>> give more insights ....
>> yes, it's deactivated. I tried resyncinc drbd by hand, deleted logs,
>> rebooted both nodes, checked drbd ain't started and started corosync.
>> result is here:
>> http://nelide.cz/nik/logs.tar.gz
>
> It really really looks like Pacemaker is too fast when promoting to
> primary ... before the connection to the already up second node can be
> established.

Do you mean we're violating a constraint?
Or is it a problem of the RA returning too soon?

> I see in your logs you have DRBD 8.3.13 userland  but
> 8.3.11 DRBD module installed ... can you test with 8.3.13 kernel module
> ... there have been fixes that look like addressing this problem.
>
> Another quick-fix, that should also do: add a start-delay of some
> seconds to the start operation of DRBD
>
> ... or fix your after-split-brain policies to automatically solve this
> special type of split-brain (with 0 blocks to sync).
>
> Best Regards,
> Andreas
>
> --
> Need help with Pacemaker?
> http://www.hastexo.com/now
>
>>
>> thanks for Your time.
>> n.
>>
>>
>>>
>>> Regards,
>>> Andreas
>>>
>>> --
>>> Need help with Pacemaker?
>>> http://www.hastexo.com/now
>>>
>>> >
>>> > thanks a lot in advance
>>> >
>>> > nik
>>> >
>>> >
>>> > On Sun, Jul 08, 2012 at 12:47:16AM +0200, Andreas Kurz wrote:
>>> >> On 07/02/2012 11:49 PM, Nikola Ciprich wrote:
>>> >>> hello,
>>> >>>
>>> >>> I'm trying to solve quite mysterious problem here..
>>> >>> I've got new cluster with bunch of SAS disks for testing purposes.
>>> >>> I've configured DRBDs (in primary/primary configuration)
>>> >>>
>>> >>> when I start drbd using drbdadm, it get's up nicely (both nodes
>>> >>> are Primary, connected).
>>> >>> however when I start it using corosync, I always get split-brain, although
>>> >>> there are no data written, no network disconnection, anything..
>>> >>
>>> >> your full drbd and Pacemaker configuration please ... some snippets from
>>> >> something are very seldom helpful ...
>>> >>
>>> >> Regards,
>>> >> Andreas
>>> >>
>>> >> --
>>> >> Need help with Pacemaker?
>>> >> http://www.hastexo.com/now
>>> >>
>>> >>>
>>> >>> here's drbd resource config:
>>> >>> primitive drbd-sas0 ocf:linbit:drbd \
>>> >>>     params drbd_resource="drbd-sas0" \
>>> >>>     operations $id="drbd-sas0-operations" \
>>> >>>     op start interval="0" timeout="240s" \
>>> >>>     op stop interval="0" timeout="200s" \
>>> >>>     op promote interval="0" timeout="200s" \
>>> >>>     op demote interval="0" timeout="200s" \
>>> >>>     op monitor interval="179s" role="Master" timeout="150s" \
>>> >>>     op monitor interval="180s" role="Slave" timeout="150s"
>>> >>>
>>> >>> ms ms-drbd-sas0 drbd-sas0 \
>>> >>>    meta clone-max="2" clone-node-max="1" master-max="2" master-node-max="1" notify="true" globally-unique="false" interleave="true" target-role="Started"
>>> >>>
>>> >>>
>>> >>> here's the dmesg output when pacemaker tries to promote drbd, causing the splitbrain:
>>> >>> [  157.646292] block drbd2: Starting worker thread (from drbdsetup [6892])
>>> >>> [  157.646539] block drbd2: disk( Diskless -> Attaching )
>>> >>> [  157.650364] block drbd2: Found 1 transactions (1 active extents) in activity log.
>>> >>> [  157.650560] block drbd2: Method to ensure write ordering: drain
>>> >>> [  157.650688] block drbd2: drbd_bm_resize called with capacity == 584667688
>>> >>> [  157.653442] block drbd2: resync bitmap: bits=73083461 words=1141930 pages=2231
>>> >>> [  157.653760] block drbd2: size = 279 GB (292333844 KB)
>>> >>> [  157.671626] block drbd2: bitmap READ of 2231 pages took 18 jiffies
>>> >>> [  157.673722] block drbd2: recounting of set bits took additional 2 jiffies
>>> >>> [  157.673846] block drbd2: 0 KB (0 bits) marked out-of-sync by on disk bit-map.
>>> >>> [  157.673972] block drbd2: disk( Attaching -> UpToDate )
>>> >>> [  157.674100] block drbd2: attached to UUIDs 0150944D23F16BAE:0000000000000000:8C175205284E3262:8C165205284E3263
>>> >>> [  157.685539] block drbd2: conn( StandAlone -> Unconnected )
>>> >>> [  157.685704] block drbd2: Starting receiver thread (from drbd2_worker [6893])
>>> >>> [  157.685928] block drbd2: receiver (re)started
>>> >>> [  157.686071] block drbd2: conn( Unconnected -> WFConnection )
>>> >>> [  158.960577] block drbd2: role( Secondary -> Primary )
>>> >>> [  158.960815] block drbd2: new current UUID 015E111F18D08945:0150944D23F16BAE:8C175205284E3262:8C165205284E3263
>>> >>> [  162.686990] block drbd2: Handshake successful: Agreed network protocol version 96
>>> >>> [  162.687183] block drbd2: conn( WFConnection -> WFReportParams )
>>> >>> [  162.687404] block drbd2: Starting asender thread (from drbd2_receiver [6927])
>>> >>> [  162.687741] block drbd2: data-integrity-alg: <not-used>
>>> >>> [  162.687930] block drbd2: drbd_sync_handshake:
>>> >>> [  162.688057] block drbd2: self 015E111F18D08945:0150944D23F16BAE:8C175205284E3262:8C165205284E3263 bits:0 flags:0
>>> >>> [  162.688244] block drbd2: peer 7EC38CBFC3D28FFF:0150944D23F16BAF:8C175205284E3263:8C165205284E3263 bits:0 flags:0
>>> >>> [  162.688428] block drbd2: uuid_compare()=100 by rule 90
>>> >>> [  162.688544] block drbd2: helper command: /sbin/drbdadm initial-split-brain minor-2
>>> >>> [  162.691332] block drbd2: helper command: /sbin/drbdadm initial-split-brain minor-2 exit code 0 (0x0)
>>> >>>
>>> >>> to me it seems to be that it's promoting it too early, and I also wonder why there is the
>>> >>> "new current UUID" stuff?
>>> >>>
>>> >>> I'm using centos6, kernel 3.0.36, drbd-8.3.13, pacemaker-1.1.6
>>> >>>
>>> >>> could anybody please try to advice me? I'm sure I'm doing something stupid, but can't figure out what...
>>> >>>
>>> >>> thanks a lot in advance
>>> >>>
>>> >>> with best regards
>>> >>>
>>> >>> nik
>>> >>>
>>> >>>
>>> >>>
>>> >>> _______________________________________________
>>> >>> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
>>> >>> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>>> >>>
>>> >>> Project Home: http://www.clusterlabs.org
>>> >>> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
>>> >>> Bugs: http://bugs.clusterlabs.org
>>> >>>
>>> >>
>>> >>
>>> >>
>>> >>
>>> >>
>>> >
>>> >
>>> >
>>> >> _______________________________________________
>>> >> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
>>> >> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>>> >>
>>> >> Project Home: http://www.clusterlabs.org
>>> >> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
>>> >> Bugs: http://bugs.clusterlabs.org
>>> >
>>> >
>>> >
>>> >
>>> > _______________________________________________
>>> > Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
>>> > http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>>> >
>>> > Project Home: http://www.clusterlabs.org
>>> > Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
>>> > Bugs: http://bugs.clusterlabs.org
>>> >
>>>
>>>
>>>
>>>
>>>
>>
>>
>>
>>> _______________________________________________
>>> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
>>> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>>>
>>> Project Home: http://www.clusterlabs.org
>>> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
>>> Bugs: http://bugs.clusterlabs.org
>>
>>
>> --
>> -------------------------------------
>> Ing. Nikola CIPRICH
>> LinuxBox.cz, s.r.o.
>> 28.rijna 168, 709 00 Ostrava
>>
>> tel.:   +420 591 166 214
>> fax:    +420 596 621 273
>> mobil:  +420 777 093 799
>> www.linuxbox.cz
>>
>> mobil servis: +420 737 238 656
>> email servis: servis at linuxbox.cz
>> -------------------------------------
>>
>> _______________________________________________
>> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
>> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>>
>> Project Home: http://www.clusterlabs.org
>> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
>> Bugs: http://bugs.clusterlabs.org
>>
>
>
>
>
> _______________________________________________
> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>
> Project Home: http://www.clusterlabs.org
> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
> Bugs: http://bugs.clusterlabs.org
>

More information about the Pacemaker mailing list