| Author |
|
Kanashii
Joined: 17 Jul 2006
Posts: 155
Location: Porando
|
 Posted: Wed 24 Nov '10 21:59 Post subject: |
 |
|
So
Mod FCGID 2.3.6 for Apache 2.2.x -- i use this
Mod FCGID 2.3.6b for Apache 2.2.x -- hmm add [ autdate ]
 |
|
| Back to top |
|
glsmith
Moderator
Joined: 16 Oct 2007
Posts: 2268
Location: Sun Diego, USA
|
| Posted: Wed 24 Nov '10 22:11 Post subject: |
|
|
I'd suggest using the "b" version. Although there were quite some changes done to 2.3.6, our specific case was not dealt with. I was able to reproduce the problem in 2.3.6, it just wasn't as easy.
If you were just offline because of mod_fcgid, yes, I suggest the "b" version.
We keep an unpatched version as well since this just does not affect everyone, and if someone is not seeing the problem, they prefer one not tampered with.
We have a bug report in and I believe Mario has discussed this with the guy who basically drives the development. As to whether it ever makes it into a ASF release, your guess is as good as mine.
I do not understand what you mean by
hmm add [ autdate ] |
|
| Back to top |
|
Kanashii
Joined: 17 Jul 2006
Posts: 155
Location: Porando
|
| Posted: Thu 25 Nov '10 2:28 Post subject: |
|
|
means outdate - version witch buq
Ok i upgrade to this fixed b |
|
| Back to top |
|
bbwc
Joined: 04 Jul 2012
Posts: 3
Location: Germany, Kassel
|
| Posted: Wed 04 Jul '12 12:23 Post subject: the bug is back in 2.3.7 |
|
|
| when I use mod_fcgid 2.3.7 with Apache 2.4 win32, the bug is back, I wonder if someone could provide a patch version like 2.3.7b, or download link for older version like 2.3.6b/2.3.5b, thanks in advance. |
|
| Back to top |
|
James Blond
Moderator
Joined: 19 Jan 2006
Posts: 7404
Location: EU, Germany, Next to Hamburg
|
| Posted: Wed 04 Jul '12 17:17 Post subject: |
|
|
The patch is in that version included. I made sure of that http://svn.apache.org/viewvc?view=revision&revision=1234169 
I never had this issue again, even with 2.3.7 which is more patched than 2.3.6b
It would be very cool, if you could set the log level from your apache to debug, past it at pastebin.com and provide the url, so I can take a look at it. |
|
| Back to top |
|
bbwc
Joined: 04 Jul 2012
Posts: 3
Location: Germany, Kassel
|
|
| Back to top |
|
glsmith
Moderator
Joined: 16 Oct 2007
Posts: 2268
Location: Sun Diego, USA
|
| Posted: Thu 05 Jul '12 21:05 Post subject: |
|
|
That's not bug 50309 you are seeing, I can see Apache is recovering and not "going down permanently," which was the problem pre-50309. I can provide you a 2.3.6d which was the last of the patched 2.3.6 modules if you want to give it a try anyway.
https://www.apachehaus.net/modules/~mod_fcgid-2.3.6d/
Of course, James states on the previous page of this topic
| James Blond wrote: | For me it works fine on Win7 and w2k8R2.
For each fcgid process I get a line like FastCGI process 424 still did not exit, terminating forcefully when using -k restart |
|
|
| Back to top |
|
James Blond
Moderator
Joined: 19 Jan 2006
Posts: 7404
Location: EU, Germany, Next to Hamburg
|
| Posted: Fri 06 Jul '12 17:59 Post subject: |
|
|
Yepp that is exactly what the patch does! It kills all the process started from fcgid which didn't stopped graceful. Without the patch the stopping failed and that caused apache to crash.
Maybe the kill should be changed from error to warn or notice. |
|
| Back to top |
|
glsmith
Moderator
Joined: 16 Oct 2007
Posts: 2268
Location: Sun Diego, USA
|
| Posted: Fri 06 Jul '12 20:53 Post subject: |
|
|
both those still pollute the error log and there's no way to stop APLOG_NOTICE, APLOG_DEBUG sounds good to me tho I agree that NOTICE or WARN sound much less dire.
The bigger poop is it's forcing a restart of Apache, it's obvious in the pastebin post, looking at the time stamps. Not that restarting every so often is a bad thing, on 2.4 it keeps the acceptex errors from PILING UP AND making the thing quit responding. |
|
| Back to top |
|
James Blond
Moderator
Joined: 19 Jan 2006
Posts: 7404
Location: EU, Germany, Next to Hamburg
|
| Posted: Sat 07 Jul '12 21:28 Post subject: |
|
|
| glsmith wrote: | | both those still pollute the error log and there's no way to stop APLOG_NOTICE, |
Of cause you can stop that with the Apache 2.4 log level config and just ban the fcgid errors
But I agree that debug it wiser than the other ones |
|
| Back to top |
|
bbwc
Joined: 04 Jul 2012
Posts: 3
Location: Germany, Kassel
|
| Posted: Wed 11 Jul '12 11:10 Post subject: |
|
|
| glsmith wrote: | both those still pollute the error log and there's no way to stop APLOG_NOTICE, APLOG_DEBUG sounds good to me tho I agree that NOTICE or WARN sound much less dire.
The bigger poop is it's forcing a restart of Apache, it's obvious in the pastebin post, looking at the time stamps. Not that restarting every so often is a bad thing, on 2.4 it keeps the acceptex errors from PILING UP AND making the thing quit responding. |
it seems like I misunderstood the meaning of [fcgid:error], the patch works and server do restart after all. I also tested the 2.3.6d version, it generate the same 'misleading' output.
btw, the fcgid do not force apache process to restart, actually I did manually to see if php-cgi quit with apache, sometimes it will stay after apache stop, is that normal? |
|
| Back to top |
|
glsmith
Moderator
Joined: 16 Oct 2007
Posts: 2268
Location: Sun Diego, USA
|
| Posted: Wed 11 Jul '12 22:36 Post subject: |
|
|
| That is bug 50309 and causes Apache not to start back up. All version of 2.3 <= 2.3.6 have the bug except patched versions (like 2.3.6d) |
|
| Back to top |
|
