One of my Squid servers that runs on BSDI used to crash a lot and not dump
core. I managed to get a core by running Squid *in the foreground* and not
sending it into daemon-land with a &. When I had Squid running like this I
got a core every time.
-Bill
On Wed, 22 Oct 1997, Mikhail A. Sokolov wrote:
> Hello,
>
> I am watching segment violation/abort traps on squid's 1.1.[11,14,16,17] for
> several days after installed FreeBSD 3.0-SMP. It won't drop core as well :-(
> I tracked down the very last queries before it happens, one of such is attached
> below.
>
>
> Any input will be greately appreciated, since squid is crashing each 2/3
> minutes or maybe seconds sometimes. Sometimes it's steady for hours, though.
>
> Here's what may be significant:
>
> Compilation options: -DUSE_PULL, -DLOG_FULL_HEADERS, -DUSERAGENT_LOG (well,
> tried -DUSE_BIN_TREE also, - the squid.conf's is quite big).
>
> Spool: tried not to use -noatime,nosync; tried different spool sizes.
> now it's 15gb, was - 2gb and 9 gb, with swap_levels 256/16,
> 16/16, 16/256..
>
> Memory:
> squid.conf's - 64 vm allowed,
> login.conf's - unlimited
> RAM - 256 Mb
>
>
> Client's simultaneous average rate: ~290/300, maximum - 500.
>
> Still, I have another machine, running 3.0-current, where squid missbehaved
> as well, with the same signs, but now it's stable and happy with it's 1.1.11
> (sigh :-(). Hardware is the same, the only difference is it have 128 vm allowed.
>
> Thank you in advance,
>
> --
> -mishania, who is upset.
Received on Wed Oct 22 1997 - 17:08:44 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:37:19 MST