Hello,
We've only been using squid for about three weeks,
so I'm still a novice. We're a small ISP (1600 customers, 300 lines)
using squid primarily to improve modem performance, which works
quite well. It really helps to eliminate the sawtooth pattern of
modem performance by keeping the pipe to the user full.
However, we're seeing periods of up to two hours where
squid just max's out our T-1. Between 1 and 3 am today, our
outbound from the squid machine was around 90kb/sec, while the
inbound was 180k.
I've been following this list and set the following parameters:
quick_abort_min 0
quick_abort_max 0
quick_abort_pct 95
half_closed_clients off
refresh_pattern . 30 50% 604800
The machine is setup with 2.8 gb of space for cache and
typically has a max object life of 1.1 to 1.2 days.
What am I missing here?
If you want to see the graph, it is at
http://stats.bit-net.com/mrtg/reflect1.1.html
Thanks in advance,
Jon
Received on Wed Dec 16 1998 - 10:16:55 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:43:39 MST