Re: Error running keepalived

[Jim Miller <jimm@xxxxxxxxxxxxxxx>]

Kjetil Torgrim Homme wrote:

> [lvs@xxxxxxxxxxxxx]:
>  
>
> >  I am hoping someone else has encountered this problem and solved
> >  it before.
> >  
> >  On 4 seperate installs of RedHat 7.3 (3 standard and 1 with LVS
> >  configured and running) when keepalived is started the following
> >  error apears
> >  
> >  Keepalived: Watchdog: Error connecting /tmp/.vrrp wdog socket
> >  
> >  The socket is created  
> >  #ls -la /tmp
> >  srwx------    1 root     root            0 Sep  2 14:51 .healthcheckers
> >  srwx------    1 root     root            0 Sep  2 14:16 .vrrp
> >    
> notice that the timestamps are different.  .vrrp was not recreated the
> last time you started keepalived.  when you started it at 14:16 it did
> work, no?
>
> there is a race condition in the code, I think.  in my setup, it
> "only" happens every fifth time or so I start keepalived.  my director
> is a Dell PowerEdge 2650, with dual CPUs.  I suspect the dual CPUs are
> aggravating the problem, but I haven't tried rebooting into a UP
> kernel.
>
> btw, upgrading to keepalived 1.1.1 will not fix this problem.  I do
> however have a patch for keepalived which will fix a segmentation
> fault which shows up as the same error in the log every time you
> reload the configuration file.
>
> also, consider joining the keepalived list -- this is more on-topic
> there.
>  
I've ran into the same problem with a dual PII 333 system.  If I 
completely power down the server and bring it up, it will create and be 
able to connect to the .vrrp wdog socket. But I have the problem if I do 
a soft-reset, or a stop-start of keepalived.  My other test box is a 
single cpu system and doesn't this problem at all.

--Jim