RE: epoll patch on squid2.5stable13 logging way too much debug

From: Steven Wilton <swilton@dont-contact.us>
Date: Wed, 5 Apr 2006 15:29:26 +0800

These messages are harmless, and the debug code can safely be raised to a
higher number (to stop them being output). The reason that I have left them
in at debug level 1 is because they indicate that squid has somehow sent all
the data it has in memory to all clients, but it has marked the reading
filedescriptor (ie the socket to the web server, or the on-disk object) as
deferred. If left unchecked, this would cause squid to never send any more
data to this client.

In theory this should never happen, as squid will normally mark a reading
filedescriptor as deferred when the object is using too much memory, and
will resume reading from the filedescriptor when data is sent to the client
(and the memory is freed). If a transfer is causing these debug messages,
it means that something unusual has happened in another part of the squid
code.

I personally see about 150 of these messages a day in our cache logs (out of
12 million requests). I don't know if there are any normal conditions that
would cause squid to behave this way, so I have left this code at debug
level 1. If there is a definite way to hit this error, it can be replicated
and debugged further.

Steven

> -----Original Message-----
> From: john allspaw [mailto:jallspaw@yahoo.com]
> Sent: Saturday, 1 April 2006 6:10 AM
> To: squid-dev@squid-cache.org
> Subject: epoll patch on squid2.5stable13 logging way too much debug
>
> ...or, if it's not debug info, then how would it be useful:
>
> I'm getting boatloads (as many as 4 every second) in my
> cache.out file:
>
> 2006/03/31 03:05:37| storeClientCopy3: http://(URL) -
> clearing ENTRY_DEFER_READ
>
> the patch I applied was taken from:
> http://devel.squid-cache.org/cgi-bin/diff2/epoll-2_5.patch?s2_5
>
> and it says:
> This patch is generated from the epoll-2_5 branch of s2_5 in squid
> Tue Jan 10 22:35:19 2006 GMTI saw someone mention this on the
> mailing-list sometime last year, but no real resolution came from it.
> thoughts ?
>
> thanks,
> John
>
>
>
> --
> No virus found in this incoming message.
> Checked by AVG Free Edition.
> Version: 7.1.385 / Virus Database: 268.3.5/301 - Release
> Date: 4/04/2006
>
>

-- 
No virus found in this outgoing message.
Checked by AVG Free Edition.
Version: 7.1.385 / Virus Database: 268.3.5/301 - Release Date: 4/04/2006
 
Received on Wed Apr 05 2006 - 02:08:54 MDT

This archive was generated by hypermail pre-2.1.9 : Mon May 01 2006 - 12:00:03 MDT