klogd 1.5 takes an excessively long time to fork.
This causes my system to hang on boot when using sysklogd 1.5 for excessively long amounts of time to the point that it's faster to restart my system into a livecd and edit the init scripts to subvert the problem. Attached is a workaround rc, which does klogd -c 4 & disown to subvert the problem of it-taking-forever-to-fork by forking it to the background, I'm not sure if klogd actually does it's proper work after being forked, but this at least makes the system bootable. -- ~predatorfreak GnuPG Public key: http://pred.dcaf-security.org/dcafsec-pub-gpgkey.asc
Brett Goulder [2007-09-03 15:46]: Hi Brett,
This causes my system to hang on boot when using sysklogd 1.5 for excessively long amounts of time to the point that it's faster to restart my system into a livecd and edit the init scripts to subvert the problem.
Attached is a workaround rc, which does klogd -c 4 & disown to subvert
Thanks for noticing, it doesn't even happen with my laptop. However, I'd like to NAK the patch for now. The release notes of sysklogd 1.5 say, among other things: "Improved daemonise routine in klogd to stabilise startup" which makes me think that something went really wrong there :) I'll try to figure out what's going on. Thanks, Tilman -- A: Because it messes up the order in which people normally read text. Q: Why is top-posting such a bad thing? A: Top-posting. Q: What is the most annoying thing on usenet and in e-mail?
On Mon, Sep 03, 2007 at 03:46:38PM -0400, Brett Goulder wrote: Hi Brett,
This causes my system to hang on boot when using sysklogd 1.5 for excessively long amounts of time to the point that it's faster to restart my system into a livecd and edit the init scripts to subvert the problem.
thanks for your report. I cannot reproduce that behaviour here on 3 different boxes and nobody else has reported that problem so far, so I guess it might be specific for your setup only. Could you please provide more informations and try to debug the problem ? regards Juergen -- Juergen Daubert | mailto:jue@jue.li Korb, Germany | http://jue.li/crux
On Tue, 4 Sep 2007 15:02:40 +0200 Juergen Daubert wrote:
On Mon, Sep 03, 2007 at 03:46:38PM -0400, Brett Goulder wrote:
Hi Brett,
This causes my system to hang on boot when using sysklogd 1.5 for excessively long amounts of time to the point that it's faster to restart my system into a livecd and edit the init scripts to subvert the problem.
thanks for your report. I cannot reproduce that behaviour here on 3 different boxes and nobody else has reported that problem so far, so I guess it might be specific for your setup only.
Could you please provide more informations and try to debug the problem ?
regards Juergen
Strangely, I'm unable to reproduce this problem on my system now, despite the fact nothing has changed since the last time I built sysklogd 1.5, before I narrowed it down to the fact klogd's child was segfaulting, but now klogd seems to run fine... strange. -- ~predatorfreak GnuPG Public key: http://pred.dcaf-security.org/dcafsec-pub-gpgkey.asc
Brett Goulder [2007-09-04 23:20]:
On Tue, 4 Sep 2007 15:02:40 +0200 Juergen Daubert wrote:
On Mon, Sep 03, 2007 at 03:46:38PM -0400, Brett Goulder wrote:
This causes my system to hang on boot when using sysklogd 1.5 for excessively long amounts of time to the point that it's faster to restart my system into a livecd and edit the init scripts to subvert the problem.
thanks for your report. I cannot reproduce that behaviour here on 3 different boxes and nobody else has reported that problem so far, so I guess it might be specific for your setup only.
Could you please provide more informations and try to debug the problem ?
Strangely, I'm unable to reproduce this problem on my system now, despite the fact nothing has changed since the last time I built sysklogd 1.5, before I narrowed it down to the fact klogd's child was segfaulting, but now klogd seems to run fine... strange.
Brett and me tracked the segfault down, I committed a fix in 1.5-2. The culprit was the CONFIG_KALLSYMS kernel option -- the klogd child tries to open /proc/kallsyms, which is only available if that kernel option is set. If klogd cannot open the file, it would still try to fclose(NULL), which will lead to a segfault. I submitted a patch upstream. Regards, Tilman -- A: Because it messes up the order in which people normally read text. Q: Why is top-posting such a bad thing? A: Top-posting. Q: What is the most annoying thing on usenet and in e-mail?
participants (3)
-
Brett Goulder -
Juergen Daubert -
Tilman Sauerbeck