On Monday 27 January 2003 16.03, Schelstraete Bart wrote:
> Quoting trainier@kalsec.com:
> > I was having that same problem. A few differences but generally
> > the same thing. I just dropped into google and searched for:
> > (Without the quotes)
> >
> > "Squid Parent: child process exited due to signal 25"
>
> Hello ,
>
> When does this error message occur? Does it occur at startup, or
> does it appear after a little while?
On Linux X86 signal 25 is filesize exceeded and indicates one of the
log files have grown above the magical 2GB file size limitation of 32
bit applications..
Once Squid hits this problem you will be stuck there until the problem
has been corrected, crashing shortly after start (almost immediately,
or on the first request or on first cache miss depending on which log
file is too large).
If you use external log rotation then the external log rotation
process is quite likely to clear up the error when it runs. If not
Squid will be most likely be stuck in the situation until you
manually fix it.
Files to look out for are
1. squid/var/logs/cache.log
2. squid/var/logs/access.log
3. cache_dir_name/swap.state
The way to protect yourself from this happening is to set up periodic
log rotation often enough. For a low/medium traffic Squid server once
per day is suitable. For very high traffic Squid server more often
may be needed.
Related Note: Personally I prefer the use of external log rotation
programs such as logrotate. Not only are they more error resilent,
but often also provide enhanced functionality like compression and/or
archival.
Regards
Henrik
Received on Mon Jan 27 2003 - 14:48:15 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 17:12:53 MST