[982] in Hotline Meeting
rt disk i/o
daemon@ATHENA.MIT.EDU (Joe Harrington)
Thu Jun 28 02:01:52 1990
Date: Thu, 28 Jun 90 01:58:35 -0400
From: Joe Harrington <jh@ATHENA.MIT.EDU>
To: hotline@ATHENA.MIT.EDU
Cc: jh@ATHENA.MIT.EDU
Reply-To: jh@ATHENA.MIT.EDU
I have narrowed the possibilities for the crashes on our NFS server.
I logged into the server as root, and did
dd if=bigfile of=/dev/null
on the files which seemed to be causing the crashes. Then I tried it
with
bs=1k, bs=2k, bs=4k...
when I tried bs=8k, it crashed immediately. This indicates either
hardware: the physical memory, controller, etc. being addressed has a
problem, or software: something, possibly the device driver, is messed up.
Please send someone over with a complete set of diagnostic tools to
check the cpu, memory, bus, disk controllers, etc. I expect it's a
hardware problem because nfs always reads and writes with 1k blocks,
and these worked fine for me.
No errors are reported in the reboot, just in the crash.
I'll be here most of the afternoon tomorrow, but feel free to send
someone earlier. The room in 54-419; the phone number here is x7795,
and other people on the floor can let you in if I'm not here.
Thanks,
--jh--
