[Linux-HA] Heartbeat 1.2.0 random segfaults on node restart

[Alan Robertson]

Umberto Nicoletti wrote:
> Hi Alan,
> thanks for the prompt reply and sorry for cross-posting with Mike, but I
> thought that maybe the two issues were related, since they both dealt
> with tty errors and I hoped that it was only a problem with version
> 1.2.0, something that can be quickly fixed with a downgrade.

No one else has reported this problem before...


>>Is it possible you have other processes trying to use these lines for 
>>something?
> 
> 
> I checked that already and I am sure heartbeat is the only process
> accessing the serial line:
> 
> what I did to check this was:
> robin:/var/log # fuser -v /dev/ttyS0
>  
>                      USER        PID ACCESS COMMAND
> /dev/ttyS0           root        736 f....  heartbeat
>                      root        839 f....  heartbeat
>                      root        840 f....  heartbeat
>                      root        841 f....  heartbeat
>                      root        842 f....  heartbeat
>                      root        843 f....  heartbeat
> 
> checked that no gpm or getty was running.
> 
> I tried to cat /dev/ttyS0 for some time, but the only output I got was
> related to hertbeat.

This is a perfect example of interfering with heartbeat ;-).  If you read 
these characters, then heartbeat can't.

> It should not be cable problem , as the cluster ran heartbeat 0.4.9 for
> almost a month and never had problems.

Your problem is definitely not cables.  It's a bug.  But, I'm afraid I'll 
need a core file to diagnose this...  The hbread processes are very 
simple...  :-(

-- 
     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