[OpenAFS] Re: Large files with 1.6.0pre2

Ryan C. Underwood nemesis@icequake.net
Fri, 4 Mar 2011 17:57:12 -0600


On Fri, Mar 04, 2011 at 04:19:48PM -0600, Andrew Deason wrote:
> On Fri, 4 Mar 2011 12:57:05 -0600
> "Ryan C. Underwood" <nemesis-lists@icequake.net> wrote:
> 
> > FileLog on source fileserver (this is showing two sequential tries):
> > Thu Mar  3 17:14:03 2011 Volume 536870966 now offline, must be salvaged.
> > Thu Mar  3 17:14:03 2011 Scheduling salvage for volume 536870966 on part /vicepa over SALVSYNC
> > Fri Mar  4 00:11:31 2011 Volume 536870966 now offline, must be salvaged.
> > Fri Mar  4 00:11:31 2011 Scheduling salvage for volume 536870966 on part /vicepa over SALVSYNC
> 
> Do you mean this is the destination fileserver? I thought you were
> copying from a 1.4.x fileserver to a 1.6.0pre2 fileserver. Those
> messages don't exist in 1.4.x.

No, both fileservers are 1.6.0pre2.  But I thought it might be
significant that the 6GB file resides on a volume created by 1.4.x while
the target volume is created by 1.6.0pre2.

> Also, what Simon said. nm dafileserver | grep open (or 'grep fstat')

# nm /usr/lib/debug/usr/lib/openafs/dafileserver |grep open
081188c0 b auditout_open
         U fdopen@@GLIBC_2.1
         U fopen64@@GLIBC_2.1
         U fopen@@GLIBC_2.1
         U freopen@@GLIBC_2.0
0809a330 T ih_open
0809b9e0 T namei_iopen
         U open64@@GLIBC_2.2
         U open@@GLIBC_2.0
080ad900 t open_file
080adaa0 t open_file
         U opendir@@GLIBC_2.0
         U openlog@@GLIBC_2.0
08099eb0 T stream_fdopen
0809a040 T stream_open

I do have an strace log of the fileserver when the client fails but have
no idea what to look for.  I didn't see any seek calls in it.  It is
about 120 megs of plaintext.  I can compress and post it somewhere if it
would help.

-- 
Ryan C. Underwood, <nemesis@icequake.net>