[111] in Info-AFS_Redistribution
Large files
daemon@ATHENA.MIT.EDU (David Brelsford)
Tue May 7 03:45:52 1991
Date: Mon, 6 May 91 13:42:06 GMT
From: brels4d@minnie.nic.kingston.ibm.com (David Brelsford)
To: -v@minnie.nic.kingston.ibm.com, info-afs@transarc.com
Has anyone run into a problem similar to this?
I'm running the release version of AFS 3.1 with three
file servers, one RISC System/6000 running AIX 3.1 and
two RTs running AIX 2.2.1. The 6000 and most of my
20+ clients belong to a Tokenring while the two RT
servers are on an eithernet connected by a PS/2 router
running AIX.
Running on a 6000 AFS client, we are unable to copy
files that are around 10Meg or larger into AFS directories
located on the RT servers.
We've typically gotten "connection timed out" messages.
Transarc suggested reducing the chunksize. We went from
64K to 32K to 16K and are still experimenting. Lowering
the chunksize, so far, has not resolved the problem, but
has changed the symptoms. At 64K, we'd get the "connection
timed out message" and the destination file would
show 0 bytes. At both 32K and 16K chunksizes, we see no
error message from the copy, but the destination file
contains only 4Meg when copying a 32meg file.
I think a related but different problem is that we are also
unable to move AFS volumes which contain files of around
10Meg or greater, from the 6000 server to either of the
RTs. The following message appears in VolserLog:
Volser: DumpVolume: Rx call failed during the dump, error 14922325122
