Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
On Fri, Feb 17, 2012 at 11:15:45AM +0200, Lawrence Strydom wrote:
> Hi List,
>
> I used DRBD in dual primary mode with ocfs2 for my load balancing
If you use dual primary DRBD with cluster file systems,
you *MUST* have *working* and *tested* fencing in place.
That is DRBD fencing policy has to be "resource-and-stonith",
and the "fence-peer" handler is supposed to trigger, or at least wait
for, a node-fencing as well.
This is necessary because otherwise, as soon as the replication
connection breaks, the data on both nodes could diverge.
(aka split brain, or even just resource-internal split brain).
I also recommend to upgrade to 8.3.12.
With 8.3.7, you'd probably still have to configure after-split-brain
auto recovery policies, even if you got the fencing right.
> web server cluster. I didn't encounter any errors during setup and
> when I put the web site on the DRBD device on the primary node, it
> replicated without any errors. It has been running fine during the
> week of testing but this morning when we updated code located on the
> DRBD device we noticed it was not replicating to the secondary node.
> the DRBD device was mounted on both nodes but /proc/drbd output this:
>
> *version: 8.3.7 (api:88/proto:86-91)
> GIT-hash: ea9e28dbff98e331a62bcbcc63a6135808fe2917 build by
> root at web01.junkmail.co.za, 2012-01-10 09:54:40
> 0: cs:StandAlone ro:Primary/Unknown ds:UpToDate/DUnknown r----
> ns:0 nr:0 dw:5960937 dr:5047235 al:1490 bm:1363 lo:0 pe:0 ua:0
> ap:0 ep:1 wo:b oos:8840028*
>
>
> I restarted drbd and ocfs2 but still the result was the same. Next I
> rebooted the misbehaving node and noticed when it came back up that
> the DRBD device was no longer mounted.
>
> Trying to mount the device manually returns this error:
> *mount /dev/drbd0
> mount.ocfs2: I/O error on channel while opening device /dev/drbd0*
>
>
> A tail of the log file shows nothing but an earlier entry shows this:
>
> *Feb 17 10:47:54 web02 kernel: [ 13.531600] block drbd0: disk(
> Attaching -> UpToDate )
> Feb 17 10:47:54 web02 kernel: [ 13.535865] block drbd0: conn(
> StandAlone -> Unconnected )
> Feb 17 10:47:54 web02 kernel: [ 13.535889] block drbd0: Starting
> receiver thread (from drbd0_worker [1484])
> Feb 17 10:47:54 web02 kernel: [ 13.535998] block drbd0: receiver
> (re)started
> Feb 17 10:47:54 web02 kernel: [ 13.536006] block drbd0: conn(
> Unconnected -> WFConnection )
>
>
> *This is my r1.res file:
>
> *===============================================================
> resource r1 {
> meta-disk internal;
> device /dev/drbd0;
> disk /dev/vol01/docroot;
>
> syncer { rate 1000M; }
> net {
> allow-two-primaries;
> after-sb-0pri discard-zero-changes;
> after-sb-1pri discard-secondary;
> after-sb-2pri disconnect;
> }
> startup { become-primary-on both; }
>
> on web01.junkmail.co.za { address 10.0.0.111:7789; }
> on web02.junkmail.co.za { address 10.0.0.112:7789; }
> }*
> *===============================================================*
>
>
>
> Here is /etc/ocfs2/cluster.conf:
>
> ===============================================================
> *cluster:
> node_count = 2
> name = jbm_web
>
> node:
> ip_port = 7777
> ip_address = 10.0.0.111
> number = 1
> name = web01
> cluster = jbm_web
>
> node:
> ip_port = 7777
> ip_address = 10.0.0.112
> number = 2
> name = web02
> cluster = jbm_web
> ================================================================
>
>
>
> *Any help/ideas much appreciated - the pressure is on here.
>
> Thanks
> *
>
>
>
> *
> *-- *
> *Lawrence Strydom*
> *Linux System Administrator*
> Junk Mail Publishing Group
> Tel : (+27) 12 342 3840 Ext 2811
> Fax : 0000
> Email : lawrence at junkmail.co.za
>
>
>
> *NOW FREE . NOW LIVE . BUY & SELL IN MINUTES*
>
> Join us on Facebook <http://www.facebook.com/junkmailclassifieds>
> Place Free Ad <http://www.junkmail.co.za/post-free-ad?utm_source=inhouse&utm_medium=email&utm_campaign=EmailSignature>
> Browse Classifieds <http://www.junkmail.co.za/r-southafrica-classifieds-QZQYRgnQX0005?utm_source=inhouse&utm_medium=email&utm_campaign=EmailSignature>
>
> *Link to MAP: http://maps.google.co.za/maps?f=q&source=s_q&hl=en&geocode=&q=1312+Pretorius+Str,+Hatfield,+Pretoria&sll=- <http://maps.google.co.za/maps?f=q&source=s_q&hl=en&geocode=&q=1312+Pretorius+Str,+Hatfield,+Pretoria&sll=->*
>
> Please visit: <http://maps.google.co.za/maps?f=q&source=s_q&hl=en&geocode=&q=1312+Pretorius+Str,+Hatfield,+Pretoria&sll=->www.junkmail.co.za
> <http://www.junkmail.co.za> www.jobmail.co.za
> <http://www.jobmail.co.za> www.capeads.co.za
> <http://www.capeads.co.za> www.lovemail.co.za
> <http://www.lovemail.co.za>
> www.saautomart.co.za <http://www.saautomart.co.za>
> www.truckandtrailer.co.za <http://www.truckandtrailer.co.za>
> www.buyandsell4x4.co.za <http://www.buyandsell4x4.co.za>
> www.bikeandquadmart.co.za <http://www.bikeandquadmart.co.za>
> Legal Disclaimer: This e-mail and its attachments may contain
> information that is confidential and that may be
> subject to legal privilege and copyright. If you are not the
> intended recipient you may not peruse, use, disclose,
> distribute, copy or retain this message. If you have received this
> message in error, please notify the sender
> immediately by e-mail, facsimile or telephone and return and
> thereafter destroy the original message. Please note
> that e-mails are subject to viruses, data corruption, delay,
> interception and unauthorised amendment, and that the
> sender does not accept liability for any damages that may be
> incurred as a result of communication by e-mail. No
> employee or intermediary is authorised to conclude a binding
> agreement on behalf of the sender by e-mail without
> express written confirmation by a duly authorised representative of
> the sender. By transmitting this e-mail message
> over the Internet the sender does not intend to allow the contents
> hereof to become part of the public domain, and
> the confidential nature of the contents shall not be altered or
> diminished from by such transmission.
> _______________________________________________
> drbd-user mailing list
> drbd-user at lists.linbit.com
> http://lists.linbit.com/mailman/listinfo/drbd-user
--
: Lars Ellenberg
: LINBIT | Your Way to High Availability
: DRBD/HA support and consulting http://www.linbit.com
DRBD® and LINBIT® are registered trademarks of LINBIT, Austria.
__
please don't Cc me, but send to list -- I'm subscribed