[Linux-HA] HA failover script

Alan Robertson alanr at unix.sh
Wed Jul 6 08:54:19 MDT 2005


Alexander Spanke wrote:
> 
> 
> Alan Robertson wrote:
>> Alexander Spanke wrote:
>>
>>> Hi all,
>>>
>>> just a quick question for my understanding. During a failover, 
>>> Heartbeat executes its failover script, which,  at my site, stops 
>>> NFS, NFS lockd and umounts the filesystems.
>>> Unfortunally I get 'filesystem busy' when it tries to umount the fs.
>>>
>>> I googled a bit and found a discussion about sending a SIGKILL signal 
>>> to the kernel lockd thread for releasing the locks, but I am stopping 
>>> the lockd before I try to umount the filesystem.
>>>
>>> Now, I got the impression the failover script is executing the 
>>> commands but is not waiting for the result of it. Which means NFS is 
>>> still in progress of stopping while Heartbeat tries to umount the 
>>> filesystem. This would explain the 'device busy' message for me, 
>>> because NFS is not yet stopped  fully at that time.
>>>
>>> Do I am right ? Other ideas ?
>>
>>
>>
>> It works for me :-).
>>
>> We ALWAYS wait for scripts to finish.  They don't always wait for 
>> their tasks to finish, though.
>>
>> And, if you're running a recent version of heartbeat the situation you 
>> described will cause the machine to reboot.  Is that what you're seeing?
> 
> No, on our failover systems, the script simply goes on and starts 
> finally on the new machine, also mounts the filesystem there and we have 
> a double mount situation ...


So, get a recent version -- like 1.2.3.


-- 
     Alan Robertson <alanr at unix.sh>

"Openness is the foundation and preservative of friendship...  Let me 
claim from you at all times your undisguised opinions." - William 
Wilberforce



More information about the Linux-HA mailing list