Experimental Physics and
Industrial Control System

"Jeff Hill" <[email protected]> · Tue, 6 Dec 2005 08:45:06 -0700

Erik,

> task 1aa53f8 CA event suspended
> ./taskwd.c" line number=175

o What version of EPICS is running?

O The next time this occurs can you email a stack trace for the suspended
thread. A stack trace can be readily obtained with the vxWorks (tt <task
id>) command. Allan can help with that.

> callbackRequest ring buffer full

In theory the callback request queue and the CA server's event queue are
independent entities. This leads one to guess that there might be
generalized corruption, but that is only a guess.

Jeff

> -----Original Message-----
> From: Erik Johansson [mailto:[email protected]]
> Sent: Monday, December 05, 2005 6:01 PM
> To: [email protected]
> Subject: Suspeneded task causes EPICS crash
> 
> Hi all,
> 
> We are getting the following message on one of our IOCs. The CA event
> task gets suspended, which effectively kills EPICS, as all channels
> become disconnected after that:
> 
> task 1aa53f8 CA event suspended
> ./taskwd.c" line number=175
> callbackRequest ring buffer full
> 
> From digging in the source code, it looks like enough callback requests
> are requested to cause the ring buffer to overflow. We use the default
> size of 2000 entries for the buffer.
> 
> Has anyone seen a failure like this before? Any advice on what to pursue
> as far as troubleshooting?
> 
> I am concerned that simply increasing the number of entries in the
> buffer may just be putting a band-aid on a bigger problem.
> 
> Thanks,
> 
> Erik
> 
> ****************************************
> Erik M. Johansson
> Adaptive Optics Software Engineer
> W. M. Keck Observatory
> 65-1120 Mamalahoa Hwy
> Kamuela, HI 96743
> Tel: 808-885-7887
> Fax: 808-885-4464

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System