• 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

dennisV17

New Pleskian
Hello,

We have upgraded to the last plesk panel 10.0.1 BUT We have a strange problem now. Somethimes when i restart i get a fault. But sometimes i restart it all goes right without any errors. The start and stop commands work properly. But when i change settings to plesk, and plesk restart apache then it doesn't go right every time.

The problem is that apache procces needs more time to stop i think. Because when i wait a litte i can start it again with the start command.

We have it running on ubuntu 8
The fault message is
/etc/init.d/apache2 restart
* Restarting web server apache2 (98)Address already in use: make_sock: could not bind to address 0.0.0.0:80
no listening sockets available, shutting down
Unable to open logs

When i look at netstat -nap when it happens i see 1 apache child running, a few seconds later it's gone and i can start apache again.

I thought to be smart and set the value of sleep in the initscript a little bit higher but that doesn't help.
i see also this errors
[Sat Nov 13 17:33:22 2010] [warn] Init: SSL server IP/port conflict: DOMAIN.be:443 (/var/www/vhosts/DOMAIN.be/conf/12896582090.59740300_httpd.include:10) vs. horde.webmail:443 (/opt/psa/admin/conf/generated/12896658090.71577100_horde.include:146)
[Sat Nov 13 17:33:22 2010] [warn] Init: SSL server IP/port conflict: default-94_103_156_221:443 (/opt/psa/admin/conf/generated/12896658090.71577100_server.include:265) vs. horde.webmail:443 (/opt/psa/admin/conf/generated/12896658090.71577100_horde.include:240)
[Sat Nov 13 17:33:22 2010] [warn] Init: SSL server IP/port conflict: default-94_103_156_222:443 (/opt/psa/admin/conf/generated/12896658090.71577100_server.include:227) vs. horde.webmail:443 (/opt/psa/admin/conf/generated/12896658090.71577100_horde.include:193)
[Sat Nov 13 17:33:22 2010] [warn] Init: SSL server IP/port conflict: default-94_103_156_207:443 (/opt/psa/admin/conf/generated/12896658090.71577100_server.include:189) vs. horde.webmail:443 (/opt/psa/admin/conf/generated/12896658090.71577100_horde.include:146)
[Sat Nov 13 17:33:22 2010] [warn] Init: SSL server IP/port conflict: default-192_168_1_222:443 (/opt/psa/admin/conf/generated/12896658090.71577100_server.include:151) vs. horde.webmail:443 (/opt/psa/admin/conf/generated/12896658090.71577100_horde.include:99)
[Sat Nov 13 17:33:22 2010] [warn] Init: SSL server IP/port conflict: default-94_103_156_208:443 (/opt/psa/admin/conf/generated/12896658090.71577100_server.include:113) vs. horde.webmail:443 (/opt/psa/admin/conf/generated/12896658090.71577100_horde.include:52)
[Sat Nov 13 17:33:22 2010] [warn] Init: You should not use name-based virtual hosts in conjunction with SSL!!
Anyone else with this problem?

thanks
Dennis
 
same probleme here. always fails to restart apache2. or psa. depending on the changes made. very annoying!
 
I'm getting the same error! Please help parallels! I'm going to try to remove Horde as an option...


[Tue Nov 16 10:55:57 2010] [warn] Init: SSL server IP/port conflict: default-66_117_230_175:443 (/opt/psa/admin/conf/generated/12898791910.32823200_server.include:313) vs. horde.webmail:443 (/opt/psa/admin/conf/generated/12898791910.32823200_horde.include:289)
[Tue Nov 16 10:55:57 2010] [warn] Init: SSL server IP/port conflict: default-66_117_230_174:443 (/opt/psa/admin/conf/generated/12898791910.32823200_server.include:274) vs. horde.webmail:443 (/opt/psa/admin/conf/generated/12898791910.32823200_horde.include:242)
[Tue Nov 16 10:55:57 2010] [warn] Init: SSL server IP/port conflict: default-66_117_230_173:443 (/opt/psa/admin/conf/generated/12898791910.32823200_server.include:235) vs. horde.webmail:443 (/opt/psa/admin/conf/generated/12898791910.32823200_horde.include:195)
[Tue Nov 16 10:55:57 2010] [warn] Init: SSL server IP/port conflict: default-66_117_230_172:443 (/opt/psa/admin/conf/generated/12898791910.32823200_server.include:197) vs. horde.webmail:443 (/opt/psa/admin/conf/generated/12898791910.32823200_horde.include:148)
[Tue Nov 16 10:55:57 2010] [warn] Init: SSL server IP/port conflict: default-66_117_230_171:443 (/opt/psa/admin/conf/generated/12898791910.32823200_server.include:158) vs. horde.webmail:443 (/opt/psa/admin/conf/generated/12898791910.32823200_horde.include:101)
[Tue Nov 16 10:55:57 2010] [warn] Init: SSL server IP/port conflict: default-66_117_230_217:443 (/opt/psa/admin/conf/generated/12898791910.32823200_server.include:119) vs. horde.webmail:443 (/opt/psa/admin/conf/generated/12898791910.32823200_horde.include:54)
 
Steps for Ubuntu 8.04LTS

Removed Horde
# /usr/local/psa/admin/bin/webmailmng --remove --name=horde

Tried Plesk Reconfigure All
./usr/local/psa/admin/httpdmng --reconfigure-all

but apache2 wouldn't restart error, "Can not restart web server:"

Finally, Plesk "Reconfigure Server" worked!
./usr/local/psa/admin/httpdmng --reconfigure-server

Well, this worked once. I tried to make a php.ini change and I couldn't restart. same error:
/etc/init.d/apache2 restart
* Restarting web server apache2 (98)Address already in use: make_sock: could not bind to address 0.0.0.0:80
no listening sockets available, shutting down
Unable to open logs

I did
killall apache2

and i was able to restart... man this is FRUSTRATING!
 
Last edited:
I am having the same problems with Plesk Panel 10.0.1 on Ubuntu 8.04 LTS (2.6.24-28-generic x86_64) and latest updates (apt-get update / apt-get upgrade -> no updates) since I updated the Plesk Panel from 9.5.*

/etc/init.d/apache2 reload
has no problems

ps -A|grep apache2
lists 10 instances of apache2

after "/etc/init.d/apache2 stop" and at least 5 Seconds waiting I still get 5 instances with "ps -A|grep apache2"

of course "/etc/init.d/apache2 start" would result in an error "(98)Address already in use: make_sock: could not bind to address..."

"/etc/init.d/apache2 restart" would do the same: stop, wait 10 seconds (with some instances still running), start (with the same error because of that running instances)

after "killall apache2" and at least 5 Seconds waiting "ps -A|grep apache2" shows no result
so "/etc/init.d/apache2 start" is working again ("Starting web server apache2 [ OK ]")

not just "/usr/local/psa/admin/bin/httpdmng --reconfigure-all" but also changing of any hosting setting or creating a new (sub-)domain always results in "Can not restart web server:" and sending me that message via e-mail

Edit:
I have modified some of the template-files in "/usr/local/psa/admin/conf/templates/default/" and "/etc/apache2/ports.conf" because I use https with SNI and OpenVPN 2.1 with port-share on my public IP on port 443 and redirect all non-vpn connections to 127.0.0.1:443. The configurations files are generated correctly - but I always have to manually kill and start the apache2 when I change the configuration
I'm using a custom "/usr/lib/apache2/modules/mod_ssl.so" (from Apache 2.2.17) with SNI-Support but the rest of my apache is still 2.2.8. (It might not be the best way to use apache modules of newer versions with an older apache install but it works so far.)
I also use a custom "/usr/lib/apache2/modules/libphp5.so" (PHP 5.3.3).

But I had the problems before I modified the template files and before I changed the mod_ssl module so I think it is not related to this modifications.
 
Last edited:
hey igor,

We found that as Temp fix already. But this is not THE solution. Because it happends Every time someone makes changes to plesk or create new domains/subdomains.

it only occurs once.
root@crush:~# grep -ir Listen /etc/apache2/*
/etc/apache2/ports.conf:Listen 80
/etc/apache2/ports.conf: Listen 443

thanks
 
i have the same problem. it is driving me insane. i have to manually start apache every time an account is suspended or created. why oh why would these actions be time based?

really poorly done.
 
parallels, upgrading to 10 is a mess!!!!!!!!!!!!!!!!!!!!!!!

I've been running plesk since version 6 and have had a few problem when upgrading. Nothing to the extent of this mess I'm experiencing with my painful upgrade to v10.x! You're seriously sucking days out of my life trying to get apache2 to remain running after ANY changes are made to a domain. I.e., adding a sub-domain, enabling SSL, etc... things my "customers" not "clients" (that's an entirely different rant) can do in their new convuluted, messed up, wtf were you thinking parallels - control panel! Thanks for taking a **** on my server.

I regress... From my apache2 error logs i believe something has to be happening with the SSL certs.


So, I disabled SSL from every domain, turned off Horde and AtMail (they both use SSL), removed all certificates and made a self-signed certificate for my server. And voila, I was able to restart apache inside Plesk. As soon as I enable SSL on a domain the ball of apache2 wax melts all over the server and I have to start plesk. I wrote a shell script "apache2-restart.sh" because I have to do this process everytime I make a change

apache2-restart.sh
/etc/init.d/apache2 stop
#
# restart - because Parallels blew up Apache
#

# sleep
sleep 5

# kill running instances
killall apache2

# start apache
/etc/init.d/apache2 start

after you make your script. make sure you chmod +x to the file to make it executable.

chmod +x apache2-restart.sh

sometimes I have to run the script twice because of the fail to bind to port error appears...

that should save you a few keystrokes. If I was good at writing shell scripts maybe I could write a cron job to check if Apache is running properlly and restart. The problem is the pid / apache2 daemon are actually running but all sites are down....................

so, what is SSL doing?

[Wed Dec 15 20:52:36 2010] [warn] Init: You should not use name-based virtual hosts in conjunction with SSL!!

After google'n this one for a while. I believe the vhost config should not include the port ":443" for the virtual domain

These are samples I've found...
incorrect conf
<VirtualHost xxx.xxx.xxx.xxx:443>
ServerName www.domain.com:443
DocumentRoot /home/directory/html
SSLEnable
SSLCertificateFile /www.domain.crt
SSLCertificateKeyFile /www.domain.key
</VirtualHost>

The VirtualHost example below is the correct method of using an SSL enabled VHost. As you can see :443 has been removed from the ServerName parameter.

<VirtualHost xxx.xxx.xxx.xxx:443>
ServerName www.domain.com
DocumentRoot /home/directory/html
SSLEnable
DocumentRoot /home/directory/html
SSLEnable
SSLCertificateFile /www.domain.crt
SSLCertificateKeyFile /www.domain.key
</VirtualHost>

instead of the VirtualHost IP, maybe use the domain?

<VirtualHost domain.com:443>
ServerName domain.com
DocumentRoot /home/directory/html
SSLEnable
DocumentRoot /home/directory/html
SSLEnable
SSLCertificateFile /www.domain.crt
SSLCertificateKeyFile /www.domain.key
</VirtualHost>

As pschneider pointed out, I've also modified these files to remove the port :443

I have modified some of the template-files in "/usr/local/psa/admin/conf/templates/default/" and "/etc/apache2/ports.conf"

However, I *HAVE NOT HAD SUCCESS* changing these files and rebuilding / reconfigure the domain using httpdmng in your_psa/admin/bin/

I'm wondering if anyone else could shed light? Also, what OS are you having difficulty with? I'm on ubuntu 8.04 LTS.

Thanks!
 
ubuntu 8.04 LTS here too

i'm considering jumping ship. i can't have this constant chaos.
 
jumping ship

Yes, cpanel is looking more attractive everyday. I went from having a stable server running version 6-9 and now with 10 I have to baby sit the machine. IMHO, Parallels, you blew it with this upgrade!
 
I've contacted the support team. I paid my $75 for a "Support Incident." This problem was created by Parallels and their shoddy upgrade. I have no clue as to why I'm forced to pay, when Parallels broke my machine. Oh yeh, my web service has been offline for about 6 hours now! This sucks!
 
here's my email proof:

From: Abhiram, Parallels <[email protected]>
Subject: [Parallels #1068694] Ticket escalated -- OFS: Per Incident Support - 38K6HN3-G8ZX
Date: December 22, 2010 6:40:11 PM EST
To: james-paul <[email protected]>
Reply-To: [email protected]

Dear Sir/Madam,

Your ticket has been escalated to our Secondline support engineers for the further detailed investigation. We will address your request as soon as possible and kindly ask for your patience and cooperation. We will contact you as soon as any investigation updates will be available.

There is no need to reply to this message right now.

Sincerely,
Parallels Support Team

Why do you ask? It doesn't sound like I'm the only user who has a messed up server after upgrading to v10. AND NOW WE'RE FORCED TO PAY SO YOU CAN FIX IT... Seems like you're the glass shop throwing rocks through our windows and then we come to you so you can fix it.......................................
 
This message is just notification about moving your ticket to more experienced engineers. It is not proof.
The problem can be in some misconfiguration on your server but not in Plesk bug and only detailed investigation on your server can define the real reason of a problem. Why you immediately speak about a bug in Plesk? Why you don't admit, what the problem can be caused by your incorrect actions or by some server misconfiguration? Could you please provide me ticket ID and I will look itself what support engineers have found?
 
Who's at fault; customer or developer...

This message is just notification about moving your ticket to more experienced engineers. It is not proof.
The problem can be in some misconfiguration on your server but not in Plesk bug and only detailed investigation on your server can define the real reason of a problem. Why you immediately speak about a bug in Plesk? Why you don't admit, what the problem can be caused by your incorrect actions or by some server misconfiguration? Could you please provide me ticket ID and I will look itself what support engineers have found?

I thought you were asking for proof that they were working on my problem. Sometimes, It's difficult for me to understand your English.

Do you have proofs? Whether was bug submitted regarding this issue?

My "proof" is the fact that my server, specifically the "Web Services" was running fine and without ANY ISSUES for years before I upgraded to v10.x. Immediately after the upgrade my "Web Services" stopped working. This is all the proof I need that Parallels is the source of the problem. I run a vanilla installation of Plesk and my OS in an attempt to avoid any configuration issues.

I'm waiting for the support team to respond to my emails. Since they started working on my machine after hours, I was unable to reach *anyone* through your maze of phone connections and forwarding. I've sent multiple emails to the support team without any response. I would really like to know what's going on.

In the end, if Parallels points to the fact that I have setup my server improperly I will eat my words. Until then, I will continue to point blame on this shoddy v10.x upgrade!
 
As far as know support has submitted request to developers regarding your problem. Let's wait results of investigation.
 
Back
Top