Re: squid-1.2.beta23 and falce (?) parent timeouts

From: Duane Wessels <wessels@dont-contact.us>
Date: Fri, 24 Jul 1998 08:40:40 -0600

Matija Grabnar writes:

>I installed squid-1.2.beta23 on one machine and gave it my main cache
>as the only parent. (The parent does NOT have the default option set).
>
>I am seeing something funny: all the operations involving the parent
>report a timeout condition, like this:
>
>> 901263403.867 131 193.2.1.232 TCP_REFRESH_HIT/304 249 GET http://
>> www.cnn.com/ images/9706/pointcast.gif - TIMEOUT_PARENT_HIT/cache.arnes.si imag
e/gif
>
>At first I thought my main cache was too busy, but there were NO non-timeout
>reports. To experiment, I set
>icp_query_timeout 15000
>
>which would make the timeout visible, if it occured.
>The result is that I still have all parent actions report timeouts,
>EVEN THOUGH the pages load quicker than in 15 seconds.
>
>So I suspect the timeouts are reported falsely.

Indeed. Here's a fix:

Index: access_log.c
===================================================================
RCS file: /surf1/CVS/squid/src/access_log.c,v
retrieving revision 1.39
diff -w -u -r1.39 access_log.c
--- access_log.c 1998/07/22 20:36:56 1.39
+++ access_log.c 1998/07/24 14:39:33
@@ -178,7 +178,7 @@
         al->private.method_str,
         al->url,
         al->cache.ident,
- al->hier.icp.timeout ? "TIMEOUT_" : "",
+ al->hier.icp.timedout ? "TIMEOUT_" : "",
         hier_strings[al->hier.code],
         al->hier.host,
         al->http.content_type);

Duane W.
Received on Fri Jul 24 1998 - 07:42:06 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:41:15 MST