Hi Christiane,
same problems here. Our main compute servers are DEC and SGI
multiprocessors under Digital Unix 3.2 and Irix 6.1.
Here, we also suffer from AFS caused crashes on the
translator machine (either a rs_aix32 or sun4m_53). After the crash,
NFS clients will get stale NFS file handles on /afs, so all compute
servers will have to be rebooted.
I have reported this to Transarc under TR-16208.
Transarc apparently works on the problem, but response has been
very poor (my view), though I asked to increase the severity several
times.
It's really fun to explain users why your compute servers have to
be rebooted so often...
I wonder whether this is a strategy to push DFS... ;-)
Best regards,
Axel
