Re: squid.1.NOVM.20+retry memory usage report bug

Michael Pelletier wrote:
>
> On Wed, 25 Feb 1998, Lionel Bouton wrote:
>
> > Since I upgraded from NOVM.17+retry to NOVM.20+retry, I noticed that the
> > amount of memory reported by cachemgr.cgi for 'Pool for disk I/O'
> > started to grow. ...
> Based on what I recall Mike Mitchell saying, the 1.NOVM.20+retry patch is
> pretty much identical to the 1.NOVM.17+retry patch, just with a few tweaks
> to make it apply properly. I'm sifting through the details of the new
> #ifdef-ified NOVM patch to insure that there's no unintentional
> differences right now... Nothing aside from the extra #ifdef lines and a
> couple of spaces here and there are present in the new ifdef patch.
>
> Has anyone running the pre-February 23 version of the 1.NOVM.20+retry
> patch observed this problem?

I just noticed a similar problem, with 1.NOVM.20 and a copy of the (NOVM
version of the) retry patch fetched on 10th Feb.

I'm currently seeing

Pool for disk I/O 85870 x 8192 bytes = 686960 KB ( 0 free)

and a total accounted memory usage of 730799KB for a Squid process which
according to "top" is using 117MB, on a system where trying to use 700+MB
would exceed the combined memory+swap available (so it clearly has to be
wrong...). Squid has been running since 11th Feb. Not sure if it's been sent
any signals to reload the configuration in that time (probably has); it will
certainly have been signalled to rotate its logs daily.

                                John Line

-- 
University of Cambridge WWW manager account (usually John Line)
Send general WWW-related enquiries to webmaster@ucs.cam.ac.uk