[Linux-HA] Core Heartbeat process exited with SIGXCPU

[Peter Luciak]

Dejan Muhamedagic  wrote / napísal(a):
> Hi,
> 
> On Fri, Jan 22, 2010 at 08:39:35AM +0100, Peter Luciak wrote:
>> Hi,
>>
>> I'm running into weird problems on a Heartbeat v1 cluster: Heartbeat 
>> restarts itself with the message:
>>
>> heartbeat[2419]: 2010/01/22_06:30:35 WARN: Exiting HBREAD process 3272 
>> killed by signal 24 [SIGXCPU - CPU limit exceeded].
>> heartbeat[2419]: 2010/01/22_06:30:35 ERROR: Exiting HBREAD process 3272 
>> dumped core
>> heartbeat[2419]: 2010/01/22_06:30:35 ERROR: Core heartbeat process died! 
>> Restarting.
> 
> The read process CPU usage is limited to 10 percent. According to
> ha.cf below, heartbeats are every 5 seconds which is quite low.

Quite low? So you suggest to increase the interval? I wonder what is the 
recommended interval for heartbeats?

>> setserial /dev/ttyS0
>> /dev/ttyS0, UART: 16550A, Port: 0x03f8, IRQ: 4
>>
>> I turned off the serial line in ha.cf (interestingly I stopped seeing 
>> serial in /proc/interrupts afterwards) to see if that will help.
> 
> So, did it?

Yup, after stopping the serial comms, heartbeat didn't crash at all in 
the past 4 days. So it was definitely something with the serial line...

Thanks
Peter
-- 
Peter LUCIAK (Peter.Luciak at iblsoft.com)
IBL Software Engineering, http://www.iblsoft.com/
Mierová 103, 82105 Bratislava, Slovakia
Phone: +421-2-32662111, Fax: +421-2-32662110
Direct: +421-2-32662175