From: Tilman Schmidt <>
Date: Fri, 14 Nov 1997 10:24:15 +0000

I have upgraded my Squid (running on a SPARCstation 5 with 32 Megs of RAM
and Solaris 2.4) from version 1.0.9 to 1.NOVM.18 and am quite happy with
it. Not only is it considerably faster, I also got rid of a lot of
annoying ICP error messages the old version had been sending to the syslog.

There is, however, a new error message now appearing frequently in
cache.log: ERR_NO_CLIENTS. (At least it doesn't clutter the syslog
anymore.) The explanation I found in src/errorpage.c says: "All Clients
went away before tranmission completed and the object would not be cached."
 However, in at least one case the client that had requested the URL in
question was still patiently waiting for it when the message appeared in
the cache.log. How come?

Also, I seem to get more ERR_READ_TIMEOUTs than before the upgrade. Has
the default read timeout value been reduced? Or is that just an illusion
because they appear in cache.log now in concentrated form?

Tilman Schmidt          E-Mail: (office)
SEMA GROUP GmbH        (private)
Koeln, Germany
Received on Fri Nov 14 1997 - 02:43:50 MST

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:37:32 MST