|
Home > Archive > Debian Developers > October 2004 > NMU on sysklogd
You are viewing an archived Text-only version of the thread.
To view this thread in it's original format and/or if you want to reply to
this thread please [click here]
|
|
| Jerome Warnier 2004-10-28, 7:49 am |
| Hello guys,
I'm having a problem with sysklogd on Sarge: everytime it rotates logs,
it fails to log in the new file, it continues in the previous, renamed
one.
I introduced bug #275111 about that.
It is really annoying since every log analysis tool is failing on this
every week at least? By "log analysis tool" I mean anything relying on
files in /var/log to do something.
I notice that that package has a huge number of bugs, with many having a
patch attached, but many bugs are really old.
Could someone go through the list and NMU this? I'm willing to help, if
necessary.
Thanks
--
Jerome Warnier <jwarnier@beeznest.net>
BeezNest s.a r.l.
--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
| |
| Petter Reinholdtsen 2004-10-28, 5:52 pm |
| [Jerome Warnier]
> Could someone go through the list and NMU this? I'm willing to help,
> if necessary.
The maintainer of sysklogd have a problematic relationship with NMUs.
Have a look at bug #225895 for an ironic view on this. 
--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
| |
| Jerome Warnier 2004-10-28, 5:52 pm |
| On Fri, 2004-10-29 at 00:46 +0200, Petter Reinholdtsen wrote:
> [Jerome Warnier]
>
> The maintainer of sysklogd have a problematic relationship with NMUs.
> Have a look at bug #225895 for an ironic view on this.
So what? Am I stuck with my problem like so many people are already? And
a "friendly" takeover of the package?
I already have to problem on at least 4 machines, with things as
POP-before-SMTP and log analysers, packaged in Debian.
--
Jerome Warnier <jwarnier@beeznest.net>
BeezNest s.a r.l.
--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
| |
| Petter Reinholdtsen 2004-10-29, 2:49 am |
| [Jerome Warnier]
> So what? Am I stuck with my problem like so many people are already? And
> a "friendly" takeover of the package?
I suspect you will discover and get stuck in the power games in Debian.
> I already have to problem on at least 4 machines, with things as
> POP-before-SMTP and log analysers, packaged in Debian.
I would recommend changing to another syslogd. There are several, and
you could switch to one with a closer match to your needs.
--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
| |
| Jerome Warnier 2004-10-29, 7:48 am |
| On Fri, 2004-10-29 at 09:02 +0200, Petter Reinholdtsen wrote:
> [Jerome Warnier]
>
> I suspect you will discover and get stuck in the power games in Debian.
I always found that funny, but I know the game. ;-)
>
> I would recommend changing to another syslogd. There are several, and
> you could switch to one with a closer match to your needs.
Well, I already tried others, but they have more dependencies or are
more difficult to configure to do what I want. And sysklogd is in
section "base", so in my point of view, it should just _work_ out of the
box. Moreover it is the default syslogd installed for now.
In fact, I dug into the problem, and found that the initscript
(/etc/cron.daily/sysklogd) had a 'reload-or-restart' argument, which is
used by the cronjob and does not always work (in my case, on several
machines, never) nor do the restart one (I'm not sure anymore I tried
it, though). So, I replaced the last line:
/etc/init.d/sysklogd reload-or-restart > /dev/null
with:
kill -HUP `cat /var/run/syslogd.pid` > /dev/null
One more thing to say: because of this bug (#275111), the logs do not
get rotated right, which can cause big trouble on heavily-loaded servers
(or using Quagga, which generates a lot of messages because of failed
communication with SNMP, but that's another story).
--
Jerome Warnier <jwarnier@beeznest.net>
BeezNest s.a r.l.
--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
| |
| Javier Fernández-Sanguino Peña 2004-10-29, 7:48 am |
| On Fri, Oct 29, 2004 at 11:56:46AM +0200, Jerome Warnier wrote:
> In fact, I dug into the problem, and found that the initscript
> (/etc/cron.daily/sysklogd) had a 'reload-or-restart' argument, which is
> used by the cronjob and does not always work (in my case, on several
> machines, never) nor do the restart one (I'm not sure anymore I tried
> it, though). So, I replaced the last line:
(...)
I think it would be best if you sent this information to the BTS, as a
patch for the current package. That way others (including the maintainer,
which needs not be following this thread) can test it too.
Please provide as much information in order to reproduce this issue as
possible too.
Those are the very first steps you should take before even considering a
NMU. Also, in this particulary case, read the README.NMU file in the
sources.
I've been reviewing these packages' bugs. Even if some patches could be
introduced to fix some of the issues I'm not sure all of them are
appropiate. Even so, this package is of base priority, so it's frozen in
sarge, only RC bugs should be fixed there. Maybe a cleanup of the package
could be considered (following the NMU procedures above) and I'm open to
doing it myself (as I've been doing it for other base packages).
I haven't yet contacted Joey, however, since I'm still considering this
option and how to do it best. One option, for example, is to do a NMU to
experimental so other people can test the package (and verify that the bugs
are actuall fixed) before uploading to sid. I have done this for other base
packages (like ifupdown). I suggest you think about it (and the
consequences of a broken^Wuntested upload at this point) before talking
with him.
Regards
Javier
| |
| Andreas Barth 2004-10-29, 7:48 am |
| * Jerome Warnier (jwarnier@beeznest.net) [041028 15:25]:
> It is really annoying since every log analysis tool is failing on this
> every week at least? By "log analysis tool" I mean anything relying on
> files in /var/log to do something.
> [..]
> Could someone go through the list and NMU this? I'm willing to help, if
> necessary.
For Bug 275111 - I think this bug needs a proper explanaition / patch,
because that's the only way to fix it.
In general: sysklogd is already frozen for release of sarge. So, please
don't upload _anything_ which won't make it for sarge, i.e. only RC-bug
fixes. Changes like using /etc/defaults might be nice, but are not
possible any more for sarge. (And, I guess that the maintainer might do
a cleanup round after release of sarge.)
Cheers,
Andi
--
http://home.arcor.de/andreas-barth/
PGP 1024/89FB5CE5 DC F1 85 6D A6 45 9C 0F 3B BE F1 D0 C5 D1 D9 0C
--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
| |
| Jerome Warnier 2004-10-29, 5:52 pm |
| On Fri, 2004-10-29 at 13:28 +0200, Javier Fern=E1ndez-Sanguino Pe=F1a wrote=
:
> On Fri, Oct 29, 2004 at 11:56:46AM +0200, Jerome Warnier wrote:
>=20
> (...)
>=20
> I think it would be best if you sent this information to the BTS, as a=20
> patch for the current package. That way others (including the maintainer,=
=20
> which needs not be following this thread) can test it too.=20
Well, the problem is that I'm not happy with that solution. In fact, the
initscript should be fixed, which maybe involves bug #211858 (relates to
--pidfile), which has already a non-applied patch.
The fact is that both 'reload-or-restart' and 'restart' do not work as
expected.
> Please provide as much information in order to reproduce this issue as=20
> possible too.
It is very simple: install a Sarge, let it run and you'll see at some
point that /var/log/syslog (this is not the only file affected of
course) is empty and keeps empty and syslog.0 continues to be filled by
the new messages.
All my Sarge systems show the same behaviour.
> Those are the very first steps you should take before even considering a=20
> NMU. Also, in this particulary case, read the README.NMU file in the=20
> sources.
I cannot do a NMU myself, so don't worry, I won't ;-)
> I've been reviewing these packages' bugs. Even if some patches could be
> introduced to fix some of the issues I'm not sure all of them are
> appropiate. Even so, this package is of base priority, so it's frozen in
> sarge, only RC bugs should be fixed there. Maybe a cleanup of the package
I'm aware of that too.
> could be considered (following the NMU procedures above) and I'm open to
> doing it myself (as I've been doing it for other base packages).
Well, thanks but it is not quite ready for that.
> I haven't yet contacted Joey, however, since I'm still considering this
> option and how to do it best. One option, for example, is to do a NMU to
> experimental so other people can test the package (and verify that the bu=
gs
> are actuall fixed) before uploading to sid. I have done this for other ba=
se
> packages (like ifupdown). I suggest you think about it (and the
> consequences of a broken^Wuntested upload at this point) before talking
> with him.
We'll see that in due time. I have lots of other things to do.
> Regards
>=20
> Javier
--=20
Jerome Warnier <jwarnier@beeznest.net>
BeezNest s.a r.l.
| |
| Javier Fernández-Sanguino Peña 2004-10-29, 8:47 pm |
| On Fri, Oct 29, 2004 at 03:50:31PM +0200, Jerome Warnier wrote:
> On Fri, 2004-10-29 at 13:28 +0200, Javier Fernández-Sanguino Peña wrote:
> We'll see that in due time. I have lots of other things to do.
I'm sure you do. I have been telling you (and you've ignored) some
steps you can do to help. Since you previously said:
> Could someone go through the list and NMU this? I'm willing to help, if
> necessary.
If you have information, please send it to the BTS, that's the place
maintainers look for information related to bugs. You shouldn't expect
people to review the -devel archives everytime they want to find
information related to a bug, do you?
If you can provide more help (besides adding information to the bugs in the
BTS) you could prepare or test patches for those bugs you care for.
Replying in a public mailing list is not the kind of help that would move
forward a fix.
Regards
Javier
ObBug: 41801 54967 54999 69754 70557 76209 90486 110732 116162 118647
132401 140776 146819 150423 150423 150423 159708 160454 165098 166531
166713 166943 173746 173938 174181 194502 203578 206205 210516 215855
219206 220382 225087 228031 228258 228688 230020 243681 253508 262941
263224 267112 272581
[Modutils NMU to experimental, yes, I'm breaking the rule since those
aren't RC and I'm not fixing them in sid, but you get the point]
|
|
|
|
|