The following URL http://nagoya.apache.org/bugzilla/show_bug.cgi?id=21737 describes how running cgi programs via suExec in Apache 1.3.28 results in defunct processes. A problem which I've noticed occurs in the now labelled "stable" Apache 1.3.28 release in portage.
Hey, many things are labelled "stable" and then have things called "bugs" and/or "issues" discovered. Deal. http://nagoya.apache.org/bugzilla/showattachment.cgi?attach_id=7407 is the patch most folks are reccomending (note for myself)
If you don't want bug reports then I won't bother in future and I think you know the problem that arrises when people don't submit bug reports. I indicated it was the stable release as to next get mixed up with any beta's of the package which I DON'T use. What an attitude towards someone who's just trying to HELP. The system is here for a reason and you are a part of that system by choice. I'm simply reporting it so that it's something that can be fixed in subsequent releases of the package, if people don't report bugs then they'd not get fixed. Deal.
It's a good report, and I'm certainly glad you brought it up. It seemd as though you were criticizing the fact that it was marked "stable" (quotations yours). If that was not your intention, my mistake. At any rate, you did do the right thing by filing a bug.
I can see your point of view as to how it seemed that way, I'm sorry for not being a tad more careful in my wording. But I did that to try and remove ambiguity as to exactly which package I'm reffering to (E.g. In case a new r1 was in beta) and to try and indicate I was reffering to the gentoo ebuild as opposed to whatever Apache have marked as stable. It's not yours or anyone else on the Gentoo teams fault that the package got marked stable since that's what happens when nobody submits reports at the beta stage. And yes I understand that not every bug's going to get squashed at the beta stage either. To be honest I'm more surprised at the Apache project for having let something which to myself seems a rather obvious problem slip through :-/ But I'm happy to but this down to a simple misunderstanding between us. Keep up the good maintanence work, it's nice when a maintainer makes the latest versions available to us so we can all stay sat on our lazy arses not making custom ebuilds :-) Cheers.
have you by chance tried that patch out? i dont have apache-1.3.x on any machines at the moment to test with...
I have tested the patch on the current 1.3.28 ebuild script. Having applied it just after the apachectl patch within the src_unpack section it cured the zombie issue. From what I could tell the other patches still managed to be applied correctly and no problems have yet arrisen, all cgi processes are running correctly. I'll keep you posted if I notice any faults occur.
Please consider adding the patch and making it the stable ebuild for 1.3.28 It appears to work fine for Forester for almost three weeks and I guess I'm not the only one who stumbled over this bug after upgrading to 1.3.28 a couple of weeks ago. The only possible issues with the patch are for other unix OS, so it should be Linux only.
Recommended fix currently in CVS: wget -O apache-zombie-cgi.diff "http://cvs.apache.org/viewcvs.cgi/apache-1.3/src/main/alloc.c.diff?r1=text&tr1=1.145&r2=text&tr2=1.147&diff_format=u" patch apache_1.3.28/src/main/alloc.c < apache-zombie-cgi.diff
Right-o, thanks for reminder. I'll try hard to make a new release this week.
added this patch to apache-1.3.28-r1, thank you
