Bug#308675: request-tracker3.4: uses forever to start up with worker mpm
On Fri, May 13, 2005 at 06:08:03PM +0200, Steinar H. Gunderson wrote: > On Fri, May 13, 2005 at 03:49:29PM +0100, Stephen Quinney wrote: > > Thanks for the bug report. I have never used the Apache2 worker MPM > > only the prefork one so I'm not sure that I can do much to help you > > here. You didn't say how you are running RT, is it with fastcgi? That > > may well be very relevant here. > > It's with mod_perl2. Could you at least try to reproduce it, so I know if > it's specific to my machine or not? Simply swapping prefork for worker with > the same config triggers it here (aptitude install apache2-mpm-worker). Sorry for the delay, I finally managed to get hold of a Sarge box to test this on. I can reproduce the problem with apache2-mpm-worker, it flattens my machine for about 3 minutes while it starts up. It seems to be terrible in responding to requests after that as well. It does appear to be related to RT as starting up apache2 without that included works fine. I'm still not sure where to start with actually debugging the cause though. I'll report it to upstream at least and see what they say. Stephen Quinney -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#308675: request-tracker3.4: uses forever to start up with worker mpm
On Tue, May 17, 2005 at 02:26:58PM +0100, Stephen Quinney wrote: > Sorry for the delay, I finally managed to get hold of a Sarge box to > test this on. I can reproduce the problem with apache2-mpm-worker, it > flattens my machine for about 3 minutes while it starts up. It seems > to be terrible in responding to requests after that as well. It does > appear to be related to RT as starting up apache2 without that > included works fine. Should this perhaps be elevated to RC? If not, it should definitely be documented in the package (ie.: "use prefork or fastcgi"). /* Steinar */ -- Homepage: http://www.sesse.net/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#308675: request-tracker3.4: uses forever to start up with worker mpm
On Tue, May 17, 2005 at 03:30:10PM +0200, Steinar H. Gunderson wrote: > On Tue, May 17, 2005 at 02:26:58PM +0100, Stephen Quinney wrote: > > Sorry for the delay, I finally managed to get hold of a Sarge box to > > test this on. I can reproduce the problem with apache2-mpm-worker, it > > flattens my machine for about 3 minutes while it starts up. It seems > > to be terrible in responding to requests after that as well. It does > > appear to be related to RT as starting up apache2 without that > > included works fine. > > Should this perhaps be elevated to RC? If not, it should definitely be > documented in the package (ie.: "use prefork or fastcgi"). Well, this doesn't help you at all but we don't support mod_perl2 in Sid/unstable anymore and I'm expecting that Sarge will be the same soon. I'm well aware that being forced to go through a mod_perl2 API change at the last minute is going to annoy quite a lot of people... Maybe I should tag this as Sarge/important - important seems most appropriate to me - "a bug which has a major effect on the usability of a package, without rendering it completely unusable to everyone". Stephen -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#308675: request-tracker3.4: uses forever to start up with worker mpm
On Tue, May 17, 2005 at 02:44:09PM +0100, Stephen Quinney wrote: > Well, this doesn't help you at all but we don't support mod_perl2 in > Sid/unstable anymore and I'm expecting that Sarge will be the same > soon. I'm well aware that being forced to go through a mod_perl2 API > change at the last minute is going to annoy quite a lot of people... Hm, that's a bit unfortunate, but OK. (Is this a permanent change for sid, or something transient during the API change?) > Maybe I should tag this as Sarge/important - important seems > most appropriate to me - "a bug which has a major effect on the > usability of a package, without rendering it completely unusable to > everyone". Mm, seems right to me. /* Steinar */ -- Homepage: http://www.sesse.net/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#308675: request-tracker3.4: uses forever to start up with worker mpm
On Tue, May 17, 2005 at 04:01:18PM +0200, Steinar H. Gunderson wrote: > On Tue, May 17, 2005 at 02:44:09PM +0100, Stephen Quinney wrote: > > Well, this doesn't help you at all but we don't support mod_perl2 in > > Sid/unstable anymore and I'm expecting that Sarge will be the same > > soon. I'm well aware that being forced to go through a mod_perl2 API > > change at the last minute is going to annoy quite a lot of people... > > Hm, that's a bit unfortunate, but OK. (Is this a permanent change for sid, > or something transient during the API change?) > The mod_perl2 support will return to Sid/unstable in due course when upstream catches up with the new API. Sarge will ship without support for mod_perl2 in request-tracker3.4, I believe. It's all still up in the air since libapache2-mod-perl2 hasn't actually migrated to Sarge yet but I'm told it will. > > Maybe I should tag this as Sarge/important - important seems > > most appropriate to me - "a bug which has a major effect on the > > usability of a package, without rendering it completely unusable to > > everyone". > > Mm, seems right to me. ok, that's now done. Stephen -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#308675: request-tracker3.4: uses forever to start up with worker mpm
Package: request-tracker3.4 Version: 3.4.1-2 Severity: important Apache2 uses two or three minutes to start up (on a dual Athlon MP 2200+) when an RT-using vhost is enabled with the worker MPM. ltrace of the process at this time shows absolutely nothing; strace shows only a steady stream of increasing brk() calls, indicating that the process is allocating memory at a steady pace. With the prefork MPM, the problems disappear completely; Apache starts up in a few seconds. This happens on both start and restart. -- System Information: Debian Release: 3.1 APT prefers unstable APT policy: (500, 'unstable'), (1, 'experimental') Architecture: i386 (i686) Kernel: Linux 2.6.11.8 Locale: LANG=en_DK.UTF-8, LC_CTYPE=en_DK.UTF-8 (charmap=UTF-8) -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#308675: request-tracker3.4: uses forever to start up with worker mpm
On Wed, May 11, 2005 at 10:46:10PM +0200, Steinar H. Gunderson wrote: > Package: request-tracker3.4 > Version: 3.4.1-2 > Severity: important > > Apache2 uses two or three minutes to start up (on a dual Athlon MP > 2200+) when an RT-using vhost is enabled with the worker MPM. ltrace > of the process at this time shows absolutely nothing; strace shows only > a steady stream of increasing brk() calls, indicating that the process > is allocating memory at a steady pace. With the prefork MPM, the > problems disappear completely; Apache starts up in a few seconds. > > This happens on both start and restart. Thanks for the bug report. I have never used the Apache2 worker MPM only the prefork one so I'm not sure that I can do much to help you here. You didn't say how you are running RT, is it with fastcgi? That may well be very relevant here. The best advice I can offer right now is to contact the rt-users mailing list and see if anyone has had similar experiences. Details are at: http://lists.bestpractical.com/cgi-bin/mailman/listinfo/rt-users If you have any further information please followup to the bug report and I will try my best to track down the issue. Thanks, Stephen Quinney -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#308675: request-tracker3.4: uses forever to start up with worker mpm
On Fri, May 13, 2005 at 03:49:29PM +0100, Stephen Quinney wrote: > Thanks for the bug report. I have never used the Apache2 worker MPM > only the prefork one so I'm not sure that I can do much to help you > here. You didn't say how you are running RT, is it with fastcgi? That > may well be very relevant here. It's with mod_perl2. Could you at least try to reproduce it, so I know if it's specific to my machine or not? Simply swapping prefork for worker with the same config triggers it here (aptitude install apache2-mpm-worker). /* Steinar */ -- Homepage: http://www.sesse.net/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#308675: request-tracker3.4: uses forever to start up with worker mpm
On Fri, May 13, 2005 at 06:08:03PM +0200, Steinar H. Gunderson wrote: > On Fri, May 13, 2005 at 03:49:29PM +0100, Stephen Quinney wrote: > > Thanks for the bug report. I have never used the Apache2 worker MPM > > only the prefork one so I'm not sure that I can do much to help you > > here. You didn't say how you are running RT, is it with fastcgi? That > > may well be very relevant here. > > It's with mod_perl2. Could you at least try to reproduce it, so I know if > it's specific to my machine or not? Simply swapping prefork for worker with > the same config triggers it here (aptitude install apache2-mpm-worker). hmm, this is all made a lot more tricky for me in not having rt3.4 set up with mod_perl2 since it has recently gone through an evil last-minute API change on Sid/unstable. Currently rt3.4 does not support mod_perl2 on Sid at all which is very annoying. I will see if I can find a Sarge box on which I can install the necessary stuff to give it a try. Stephen -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]