Gentoo Websites Logo
Go to: Gentoo Home Documentation Forums Lists Bugs Planet Store Wiki Get Gentoo!
Bug 198896 - app-admin/sysklogd - request for a new USE flag
Summary: app-admin/sysklogd - request for a new USE flag
Status: VERIFIED WONTFIX
Alias: None
Product: Gentoo Linux
Classification: Unclassified
Component: [OLD] Core system (show other bugs)
Hardware: All Linux
: High enhancement (vote)
Assignee: Gentoo Linux bug wranglers
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2007-11-12 08:40 UTC by Alun Jones
Modified: 2007-11-12 12:57 UTC (History)
0 users

See Also:
Package list:
Runtime testing required: ---


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Alun Jones 2007-11-12 08:40:46 UTC
sysklogd installs a cron job to rotate its logs. It would be very nice if the package had a USE flag which disabled installation of this script. Reading through previous bugs with this package, there seem to be a lot of people having trouble with this cron job and I, for one, would prefer to manage the logs myself. I'm aware that I can just remove the script or modify it, but it would be nice not to have to check this on every server I run after every update of sysklogd.

Reproducible: Always
Comment 1 Jakub Moc (RETIRED) gentoo-dev 2007-11-12 08:46:27 UTC
Use INSTALL_MASK if you don't want this installed, use flags for installing one plaintext file don't make sense.

Thanks.
Comment 2 Alun Jones 2007-11-12 11:59:55 UTC
(In reply to comment #1)
> Use INSTALL_MASK if you don't want this installed, use flags for 
> installing one plaintext file don't make sense.

Fair enough - I'd not realised that INSTALL_MASK could be set to specify full paths rather than patterns. 

It seems to be a bit of a brute-force approach to use a global setting to modify the behaviour of a single e-build, and that using INSTALL_MASK doesn't scale well if it's widely used for omitting single files from packages. It also doesn't work if a future update to sysklogd used a different name for the script).

When I made the request, I was thinking in terms of squid's "logrotate" option.
Comment 3 Jakub Moc (RETIRED) gentoo-dev 2007-11-12 12:07:24 UTC
(In reply to comment #2)
> When I made the request, I was thinking in terms of squid's "logrotate" option.

Squid uses this flag for switching between native logging and using logrotate to do the job (which is valid). Using the flag to control whether one plaintext file gets installed or not sucks, you can use INSTALL_MASK for this, or just don't install logrotate if you don't want the feature. :)

BTW, Bug 69391 claims that the rotation thing is basically no-op, so I wonder what's your real issue here. :) Also see Bug 40304.
Comment 4 Alun Jones 2007-11-12 12:57:34 UTC
(In reply to comment #3)
> 
Just to clarify, I'm not demanding that this is re-opened - your suggested solution is good enough for me. I'm just arguing that it's not an intuitive solution.

> Squid uses this flag for switching between native logging and using 
> logrotate to do the job (which is valid).

I suspect this is what I was asking for! sysklogd unconditionally installs a cron job which implements log rotation on a daily schedule whether or not you already use logrotate. Having a specific log rotation method imposed on me and having to use a global make.conf setting to change the behaviour of one package also sucks! Everything I've read about INSTALL_MASK suggests it was designed with a different purpose in mind.

Squid allows you to specify which method to use, but to do the same with sysklogd I have to modify a global make.conf setting. This seems counter-intuitive to me (actually, the fact that I had to ask this question in the first place shows how counter-intuitive it is - using INSTALL_MASK to do the job is using a sledgehammer to crack a nut!).

> or just don't install logrotate if you don't want the feature. :)

Ummm... that's the point - I *want* logrotate to do the job, but sysklogd insists on installing its own script. This script has, in the past, been problematic (e.g. the bugs you mention, plus the fact that they've sometimes used .bz2 and sometimes .gz). I can understand it's needed by default to stop you getting bug reports about sysklogd filling disks, but it's a nuisance for people who know what they're doing with regard to log rotation.

> BTW, Bug 69391 claims that the rotation thing is basically no-op, so I wonder
> what's your real issue here. :) Also see Bug 40304.

:-)

I don't have a huge issue with it - we'd already arrived at a solution here, but merging in a new version of /etc/cron.daily/syslog.cron and adding an "exit" for the 96th time just made me wonder whether it was worth making the request!