[squid-users] Squid 3.5.28 crash regulary when many requests occur

[Amos Jeffries]

On 22/03/19 6:07 am, Secinfo wrote:
> Hi we are using Squid as a Caching proxy for out patch deployment.
> 
> Since we moved to version 3.5.28 on windows Server ,we have problems in our
> main spots.
> 
> typical cache.log entry´s are

... snip a lot of normal startup messages ...


These following are side effects of crashing. That leaves any partially
received and open files with broken content. Squid has detected and
resolved these issues automatically.

> 2019/03/12 14:12:03 kid1| DiskThreadsDiskFile::openDone: (2) No such file or
> directory
> 2019/03/12 14:12:03 kid1|                  F:\LEMSS-Content/00/00/00000025
> 2019/03/12 14:12:03 kid1| WARNING: 1 swapin MD5 mismatches
> 2019/03/12 14:12:03 kid1| DiskThreadsDiskFile::openDone: (2) No such file or
> directory
...

When you have an actual crash situation (as opposed to a FATAL event,
exception, or assertion - where Squid shuts itself down cleanly) we need
a backtrace from the crash to identify what is going on.

I'm not sure exactly how one does that for Windows servers these days,
cygwin should in theory have the same or similar tools as Unix. Details
of what we need can be found at:
<https://wiki.squid-cache.org/SquidFaq/BugReporting#crashes_and_core_dumps>

Once you have a backtrace you may find existing bugs in our Bugzilla
(<https://bugs.squid-cache.org/>) about this issue. Remember to search
closed bugs as it may only be fixed in the current stable series. If
there is nothing already on record please report it.

Amos