• If you are still using CentOS 7.9, it's time to convert to Alma 8 with the free centos2alma tool by Plesk or Plesk Migrator. Please let us know your experiences or concerns in this thread:
    CentOS2Alma discussion

Apache sometimes fails to restart

Hi IgorG,

your updated patch seems to work on centos 5.5 - 64bit. Before this patch I had to manually restart apache several times while migrating domains and it took over 30 min for migrating a domain. Now it's under 3 min und no need to restart apache manually and there was no downtime. This patch solved a big bug for me and hopefully also for other server admins.

Thanks
 
Last edited:
centos 5.5. 64bit

Hi,

I have also used new binary to test. confirming - patch works.
 
IgorG said:
Try to replace websrvmng with attached and check results.
Do not forget create backup of original websrvmng!

http://download1.sw-soft.com/Plesk/Autoupdate/websrvmng.tar.bz2
Click to expand...

It this patch specific to 10.1.1? I just upgraded a server that had been a fresh install of 10.0.1 to 10.1.1 last night, centos 5.5, and now I'm having the same exact issue as everyone else in this thread; a change to a plan or a new plan and apache doesn't restart on its own.

Edit: Actually after putting that new file in place, the watchdog seems to kill apache off every minute and then can't restart it, so much for that idea.
 
Last edited:
Hostasaurus.Com said:
It this patch specific to 10.1.1? I just upgraded a server that had been a fresh install of 10.0.1 to 10.1.1 last night, centos 5.5, and now I'm having the same exact issue as everyone else in this thread; a change to a plan or a new plan and apache doesn't restart on its own.

Edit: Actually after putting that new file in place, the watchdog seems to kill apache off every minute and then can't restart it, so much for that idea.
Click to expand...

Following up, 10.0.1 we were stable, and watchdog was not running. After upgrading to 10.1.1, any change in plesk that requires an apache restart leaves apache not running. Trying to use watchdog to fix the issue results in watchdog causing apache to die off and NOT restarting, umm, yeah, kind of the opposite of what the program is supposed to do. With watchdog active, you get this in the logs when apache dies off:

[Tue Feb 01 19:00:24 2011] [emerg] (43)Identifier removed: couldn't grab the accept mutex
[Tue Feb 01 19:00:24 2011] [emerg] (43)Identifier removed: couldn't grab the accept mutex
[Tue Feb 01 19:00:24 2011] [emerg] (43)Identifier removed: couldn't grab the accept mutex
[Tue Feb 01 19:00:24 2011] [emerg] (43)Identifier removed: couldn't grab the accept mutex
[Tue Feb 01 19:00:24 2011] [emerg] (43)Identifier removed: couldn't grab the accept mutex
[Tue Feb 01 19:00:24 2011] [emerg] (43)Identifier removed: couldn't grab the accept mutex
[Tue Feb 01 19:00:24 2011] [emerg] (43)Identifier removed: couldn't grab the accept mutex
[Tue Feb 01 19:00:24 2011] [emerg] (43)Identifier removed: couldn't grab the accept mutex
[Tue Feb 01 19:00:24 2011] [emerg] (43)Identifier removed: couldn't grab the accept mutex
[Tue Feb 01 19:00:25 2011] [emerg] (22)Invalid argument: couldn't grab the accept mutex
[Tue Feb 01 19:00:25 2011] [emerg] (22)Invalid argument: couldn't release the accept mutex
[Tue Feb 01 19:00:26 2011] [emerg] (22)Invalid argument: couldn't grab the accept mutex


With watchdog disabled, at least your web server continues running, until someone changes something that requires an apache restart, then its down until your monitoring system pages you so you can manually restart it; that's great.
 
We are having the same problem here with Plesk 10.1.1 on Debian 5 64bit os.

I can't enable the watchdog to, getting a message that the watchdog could not be started and I need to contact technical support.

Parallels, any update on solving this issue? not only we, but customers running dedicated servers with plesk 10 are having a lot of trouble and stress..
 
mkoninkx said:
We are having the same problem here with Plesk 10.1.1 on Debian 5 64bit os.

I can't enable the watchdog to, getting a message that the watchdog could not be started and I need to contact technical support.

Parallels, any update on solving this issue? not only we, but customers running dedicated servers with plesk 10 are having a lot of trouble and stress..
Click to expand...

Did you try provided above patch?
 
Not yet since this was not confirmed to work with Plesk 10.1.1

I'll try the patch right away and post the results here.
 
The patch seems to solve the problem on Debian 5 x64!
I'm still having a lot of the following messages in the apache error log:
[Thu Feb 03 10:55:32 2011] [warn] Init: SSL server IP/port conflict: domain.nl:443 (/var/www/vhosts/domain.nl/conf/12961358770.86166400_httpd.include:10) vs. horde.webmail:443 (/opt/psa/admin/conf/generated/12964160230.68876900_horde.include:52)
 
IgorG said:
Did you try provided above patch?
Click to expand...

I tried the patch and watchdog still killed apache every minute. Is the patch supposed to fix watchdog or supposed to fix Plesk's problem leaving apache down every time a change is made? And is it for 10.1.1 or 10.0.1?
 
Sorry, I don´t know how to apply the patch to my system (opensuse 11.1, Plesk 10.1.1).
Can please anyone tell me what to do?

I know that you can run websrvmng at /usr/local/psa/admin/bin/ , but that´s not the path, where I have to put the patched file, right?
 
It should be something like /opt/psa/admin/sbin/ for opensuse. But I'm not sure. Check it.
 
I applied the patch for opensuse 11.1, but it didn´t solve the problem.
I can reproduce the error by typing 'logrotate -f logrotate.conf'.
After running this statement, Apache stops and I have to start the service manually.
 
IgorG said:
Yes, it is patch for 10.1.1 for all oses.
Click to expand...

I've found one bug in the patch so far.

If you change a domain's IP address as the server admin, it never updates the config files for apache.

To reproduce the issue, you log into the server as an admin, go to Subscriptions, click on the domain name, click on "Change Hosting Settings", pick a new IP, click Ok. Plesk will show the new IP at this point, but the config file for apache will be the same as it was before and apache has not been restarted.

To get it to work correctly, at this point you have to click Open Control Panel to log in as the customer, click Websites & Domains, click Web Hosting Settings and then click the OK button without actually changing anything. Plesk will now create the new httpd.conf file with the correct IP address.
 
IgorG said:
http://kb.odin.com/en/9659
Click to expand...

Perhaps you can expand on your post by telling us what exactly the new article represents? Is this a later patch version that fixes the bug I pointed out? Is this a later patch version than the one posted previously in the thread but still doesn't fix the other bug? Does this fix additional issues not yet documented? etc.
 
You must log in or register to reply here.

Similar threads

H
Issue ipv6 install error failed (99: Cannot assign requested address)
Replies
1
Views
582
Peter Debik
P
Ehud
Resolved Plesk nginx conf files, seem to bring to an error notification when running 'nginx -s reload'
Replies
2
Views
944
Ehud
Ehud
Tim_Wakeling
Resolved nginx protocol options redefined
2
Replies
39
Views
9K
TobiasP
T
C
Issue New Configuration not created
Replies
3
Views
367
Peter Debik
P
S
Issue Apache Default Module
Replies
1
Views
879
IgorG
IgorG
Back
Top