Hi all,
I HIGHLY appreciate all the responses I got from you guys.
Let me mention a couple of issues regarding this problem
1) The problem is not swapping. I'm continously checking hard disk activity, and it's normal (disk light is almost off 60% of the time, and memory is plentiful)
2) I had Linux 2.0.30 compiled with SYN-COOKIES and SYN_RESET options enabled. These are supposed to protect my Linux from SYN attacks. However, when I disabled them and recompiled the kernel, I got almost 60% increase in performance during peak times!!
3) Even though I have an old NE2000 compatible card, I don't beleive it's the problem. This is the way I check Squid:
* I telnet to port 8082 (cache port), wait for the "Escape character is '^]'." messae
* I issue a "GET" command (simply "GET"), and wait for the HTML error to appear
TTP/1.0 400 Cache Detected Error
Content-type: text/html
<HTML><HEAD><TITLE>ERROR: Invalid HTTP Request</TITLE></HEAD>
<BODY><H1>ERROR</H1>
<H2>Invalid HTTP Request</H2>
<HR>
<PRE>
GET
</PRE>
<P>
<HR>
<ADDRESS>
Generated by squid/1.1.10@squid.cyberia.net.lb
</ADDRESS></BODY></HTML>
This error message appears instantly when the load is rather low. Problem begins at peak hours (load ~ 70,000 hits/hour). During these times, the "Escape character is '^]'." shows rather quickly (Squid rarely hangs at this stage). However, when I issue the "GET" command, I sometimes wait 6-7 seconds for the error message to come up! and THIS IS THE PROBLEM!
I said the net card was not problem, because I also telnetted to 127.0.0.1 (the loopback address) and got the same delay.
Again, any would be appreciated.
PS: I have nothing against FreeBSD. I just don't have much experience with. My field is Solaris and Linux
TIA
Received on Fri May 23 1997 - 05:32:18 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:35:14 MST