[30372] in Hotline Meeting
Repeated crash dumps
daemon@ATHENA.MIT.EDU (John Hawkinson)
Sun Oct 15 15:01:46 1995
Date: Sun, 15 Oct 1995 15:01:40 -0400
To: bug-sgi@MIT.EDU
Cc: hotline@MIT.EDU
From: John Hawkinson <jhawk@MIT.EDU>
From my perspective, the number one failure mode of cluster SGIs is
panicing (from afs) due to running out of space. Invariably the
largest consumer of disk space is crash dumps from other panics. This
state makes the machines unusable, and they need to be booted single
user and space made.
Since no one seems to be looking at these crash dumps, could they either:
1) be disabled
2) cleaned up periodically
3) have minfree set up so that crash dumps will not overfill
the partition
Thanks!
--jhawk
