Filesystem resource script is busted

heartbeat-2 is a transitional package.

Confirmed this bug in gutsy, heartbeat package version 2.1.2-1

Attached is a workaround script. Put it in resource.d while this bug isn`t resolved to get your cluster up and running. Be careful because the script only supports 2 parameters (device and mount point).

I ended up quickly porting the old Filesystem resource from heartbeat 1 and am using that for now. I'll use that until I see a patch for the OCF script. I'm wary of posting my ported file since it's a bit hackish and is certainly not a solution to the real problem.

Confirmed in gutsy.
This workaround script proposed by Vitor does work.

Beware that it's not a nice hack... it doesn't check for errors and I wouldn't recommend it for production.

Unfortunately does not seem to solve completely. After some tests I realized that.

When I turn off the primary server the secondary take control and mounts the /dev/drbd0 and everything seems fine, but about 10 seconds later I get this:

/var/log/ha-log:
ResourceManager[4421]: 2008/01/04_15:18:33 info: Running /etc/ha.d/resource.d/filesystem /dev/drbd0 /replicado start
ResourceManager[4421]: 2008/01/04_15:18:33 ERROR: Return code 32 from /etc/ha.d/resource.d/filesystem
ResourceManager[4421]: 2008/01/04_15:18:33 CRIT: Giving up resources due to failure of filesystem::/dev/drbd0::/replicado

And then heartbeat goes down stopping all services/filesystems, and then I got no working server.
I guess it is a problem with the filesystem script that Vitor posted (sorry, I know I am not supposed to rely on this one).
Any ideas of how to get an Filesystem script that works?

OK, got working!

Just add 'exit 0' at the end of Vitor's script and Filesystems is always return OK. I know, it is not a clever solution but it works for now.

I was in the linux-ha channel today, and talked with the guys over the bug, and they suggested not to use the above script, but to rather use the file from the current CVS/SVN. Here is what I have recorded.

[12:01] <milestone> nevermind, found https://bugs.launchpad.net/ubuntu/+source/heartbeat-2/+bug/174182
[12:01] [470] #java ##java Forwarding to another channel
[12:01] [Notice] -ChanServ- [##java] Please read the topic for basic channel guidelines. Thanks.
[12:32] <dejanm> milestone: that workaround is dreadful
[12:32] <milestone> dejanm: and what should i do instead? And why is it posted there?
[12:33] <dejanm> milestone: don't know why was it posted, suppose that anybody can post anything
[12:33] <dejanm> milestone: i think that that bug was fixed recently
[12:33] <dejanm> milestone: you can fetch Filesystem from the repository
[12:35] <dejanm> http://hg.linux-ha.org/dev/raw-file/667542dd39f2/resources/OCF/Filesystem
[12:40] <lmb> 2.1.2? They should use 2.1.3 anyway ;-)
[12:40] <dejanm> unfortunately, that bug is present in 2.1.3 too
[12:41] <lmb> build issue?
[12:42] <dejanm> no
[12:42] <dejanm> take a look at: http://hg.linux-ha.org/dev/rev/667542dd39f2
[12:46] <lmb> That's weird, how did that ever work?
[12:46] <dejanm> no idea
[12:47] * lmb 's very convinced it did work, because there's production systems running said code ...
[12:52] <dejanm> milestone: if you have an account over there, can you please ask people not to use that substitute script
[12:53] <lmb> beekhof: Hey, are you still updating the interim releases in OBS?
[12:53] <lmb> I'm thinking about what I should be doing with the daily builds now ...
[12:53] <dejanm> milestone: thx

Solved??
It seems to be a bash/dash issue.

Change #!/bin/sh to #!/bin/bash in Filesystem and all appears well!

Though If anyone can help with Mysql resource scripts I'd like to talk. /etc/ha.d/resource.d/mysql seems to be missing!

I confirm that it's a bash/dash issue.

It's working after I changed the first line of "/usr/lib/ocf/resource.d/heartbeat/Filesystem" from "#!/bin/sh" to "#!/bin/bash"

Patch attached (for the comfort of the maintainer :).

I'm sorry, Zumi, but changing "#!/bin/sh" to "#!/bin/bash" does not work for me.

I'm using:
Ubuntu 7.10 (server edition)
Heartbeat 2 (2.0.2-5)
-- I was mounting a partition from DRBD (8.0.3-2) but it does not work.

Kris

Again sorry, Zumi, your solution works well!
My mistake... I had edited a wrong file: /etc/ha.d/resource.d/Filesystem (the same, but in a other direcotry).

Thanks!
Kris

This bug was fixed in the package heartbeat - 2.1.3-6ubuntu1

---------------
heartbeat (2.1.3-6ubuntu1) intrepid; urgency=low

  * debian/patches/more_bashisms.patch:
    - Fix some bashisms (LP: #174182, #230653).
  * debian/control:
    - Update Maintainer field as per spec.

 -- Luca Falavigna <email address hidden> Thu, 19 Jun 2008 22:55:57 +0200