Hello all,
We have a squid server which, for no apparent reason, is suddenly running
with an average CPU usage for Squid of 90%. Before today, the average was
around 8-9%.
Our caches, four separate file systems running the Veritas File System,
have all reached 100%. We reset the high-low water mark for LRU at 89-90,
due to the size of our caches. We defrag each file system over night and on
the weekends.
The only error in the cache.log file, that is constantly occurring is:
sslReadClient: FD 143: read failure: (131) Connection reset by peer
We have another server, exactly the same ( load sharing the network ) and
it's load is around 8-9%.
What could be causing this???
Any thoughts?
Is it worth increasing the debug level??
The server seems to be servicing all requests at the moment without delays,
but for how long??
Received on Thu Dec 20 2001 - 18:31:24 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 17:05:27 MST