Dear Managers,
I seem to have a subborn update daemon. Up to now, I have been able to
establish that the update daemon's CPU utilization is directly proportional
to disk activity. Then, the cycling of CPU usage is around the (default) 30
second heartbeat that triggers the sync operations. In a reasonably
configured system, the update daemon takes about 2-3 seconds to flush
information to disk (with about 70% of CPU utilization) and then bottoms
down at 0% for the rest of the cycle.
However, I have been wittnessing that the CPU utilization of the update
daemon now bottoms down at 30% and, more often than not, it is above 70%.
This could indicate a disk performance bottleneck, except that during this
period (which lasts for several minutes) there is not a significant amount
of disk activity! Given the high priority of this operation, very little CPU
time is left for the rest of my users.
When the update daemon is in this mode, I have tried to send a kill signal
to it, and it takes minutes to receive it and die.
Any suggestions as to what could be happening or similar events would be
appreciated. I will summarize.
Thanks in advance,
Carlos Martinez-Mascarua
ALSTOM ESCA
Received on Tue Apr 04 2000 - 16:39:25 NZST