[OpenAFS-devel] volserver hangs possible fix

[Horst Birthelmer]

On Apr 12, 2005, at 5:16 AM, Derrick J Brashear wrote:

> On Mon, 11 Apr 2005, Horst Birthelmer wrote:
>
>> The Problem we still have are those spurious volserver hangs in some 
>> FSYNC operations.
>
> Backtrace?

No, I don't have any since it weren't my machines that hang. Mine never 
hang :-), which also means, I wasn't able to reproduce the problem.

In all the backtraces I saw volserver was in FSYNC_askfs, which doesn't 
mean very much.
So I looked into the processing of that call in the fileserver. There 
weren't any callback calls for hours.
Those guys had volservers hanging for almost a day (or until they 
restarted them with the appropriate bos command), which dosen't fit 
into the theory of breaking callbacks, does it??

This has always been on AIX and in the beginning I thought it's a 
problem related to AIX. Now since others got the same problems on Linux 
or maybe other OSes I think maybe we have a completely different 
problem.
The possibility that there are more than one bug out there and we're 
tracing ghosts is also more than nonexistent, isn't it?

Horst