[OpenAFS] Fileserver: frequent crashes

Derrick J Brashear shadow@dementia.org
Fri, 15 Oct 2004 13:58:35 -0400 (EDT)


On Fri, 15 Oct 2004, Erwin Broschinski wrote:

> Hi
>
> we are running all (but one) fileservers on Solaris 8 with OpenAFS-1.2.11. The
> software is from openafs.org's website.
> For a few weeks now, we experience frequent fs crashes, after months of living
> very comfortably.

This first core backtrace is nonsense; rx_NewService isn't called from 
DestroyConnectionNoLock. try dbx?

> Thread 16 (process 586978    ):
> #0  0xff19f474 in ?? ()
> #1  0xff0c93ac in ?? ()
> #2  0xff0c81b4 in ?? ()
> #3  0xff0c8078 in ?? ()
> #4  0x0007748c in rx_NewService ()
> #5  0x00076c24 in rxi_DestroyConnectionNoLock ()
> #6  0x000744f8 in rxi_InitPeerParams ()
> #7  0x00073e58 in rx_GetIFInfo ()
>

>    1630676 accesses in the past day (i.e., vnode references)

Well, I can guess what the problem might be, since you say Windows 
software.

> Clients in the student labs are 1.3.71

You'll want 1.3.72 when it's released, but, can you get a better backtrace 
so I know if I also have a relevant patch for the fileserver core issue?

> Frequently accessing a volume should not crash a fileserver anyhow??

Correct.

> Anything else I can do?

Get a real backtrace, for starters.