Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 4 Dec 2001 11:53:43 -0800
From:      Gary Kline <kline@tao.thought.org>
To:        Michal Mertl <mime@traveller.cz>
Cc:        questions@FreeBSD.ORG
Subject:   Re: problem with apache log (newsyslog proposal)
Message-ID:  <20011204115342.C21660@tao.thought.org>
In-Reply-To: <Pine.BSO.4.21.0112041116090.5444-100000@prg.traveller.cz>; from mime@traveller.cz on Tue, Dec 04, 2001 at 12:33:02PM %2B0100
References:  <Pine.BSO.4.21.0112041116090.5444-100000@prg.traveller.cz>

next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, Dec 04, 2001 at 12:33:02PM +0100, Michal Mertl wrote:
> The reason you couldn't realiably compress apache's logfiles is that after
> renaming the log, the daemon process still writes to the old log file.
> 
> Details: When you make it reopen the log (by sending USR1) new connections
> (ones handled with children who have already restarted) get logged to the
> new file. The ones which started before the signal was delivered are
> logged to the old file. With HUP delivered to master apache process the
> children are rather abruptly killed (thus all new log entries go the the
> new file). If you don't mind killing some downloads from your site
> (clients get only partial downloads) sending HUP should enable you to
> compress the logs. If you don't like interrupted downloads (imagine 10MB
> file over 56k modem interrupted at 9MB (client looses ~22 minutes) and
> want to compress the logs, you can either compress them separately from
> rotation much later (~hours to be sure) or you can rotate using some
> script which monitors if the old log file is still open by some children
> and start the compression only after there's no such children. I have
> small script for that if someone is interested.
> 
> I think there can be flag in newsyslog to make it check if there's a
> process which has open the file which is to be compressed and postpone the
> compression until either no such child exists or some timeout is hit. If
> anyone thinks it's a good idea, I can try to implement that functionality.
> 
> 

	Thanks for the in-depth insights here.    I think it would be
	great to have a flag that would [gb]zip or wait some -t N
	seconds.   (Yes, diskspace is getting inexpensive (&c), but
	some of us are wedged with older hardware.... )

	gary

-- 
   Gary Kline     kline@thought.org   www.thought.org     Public service Unix


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-questions" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20011204115342.C21660>