[Linux-HA] HA NFS fun

Wally Winchester wally_winchester at fastmail.fm
Mon Feb 13 09:13:11 MST 2006


The problem was solved by using a 2.4 kernel (2.4.27).
Both the 2.6 kernel in Debian and the 2.6 kernel in Ubuntu had the
failback delay.

On Thu, 09 Feb 2006 16:06:58 +0000, "Wally Winchester"
<wally_winchester at fastmail.fm> said:
> I've got a bit further.
> 
> On box 1, I run heartbeat start. Box 2 has heartbeat stopped.
> I mount the nfs share on the client, and start writing to it.
> Then, on box 1, I stop heartbeat. On box 2 I start heartbeat.
> 
> It doesn't work either.
> 
> All heartbeat is doing is running a single script. When it does this,
> failover takes 15 minutes, when I do this manually - without heartbeat -
> it takes around a minute.
> However, if I run my script (again, manually, without heartbeat) after
> heartbeat has run, I get the broken behaviour. Running the script
> without running heartbeat first works perfectly.
> heartbeat seems to be breaking something.
> 
> This is very very strange.
> 
> 
> On the server I see:
> tcp        0      1 serverip:2049     clientip:800            FIN_WAIT1 
> -
> for ten minutes.
> 
> Send-Q continues to grow during this time.
> 
> After just over five more minutes, the file gets written to disk.
> -- 
>   Wally Winchester
>   wally_winchester at fastmail.fm
> 
> -- 
> http://www.fastmail.fm - Or how I learned to stop worrying and
>                           love email again
> 
> _______________________________________________
> Linux-HA mailing list
> Linux-HA at lists.linux-ha.org
> http://lists.linux-ha.org/mailman/listinfo/linux-ha
> See also: http://linux-ha.org/ReportingProblems
-- 
  Wally Winchester
  wally_winchester at fastmail.fm

-- 
http://www.fastmail.fm - The way an email service should be




More information about the Linux-HA mailing list