Help needed, Permission denied after upgrade 11 to 12

[gijsbert]

I need urgent help. Last night I've been upgrading 4 servers from plesk 11 to plesk12. These servers are identical or at least they should be.

On 2 servers I receive the following messages when logging in:

Internal error: Permission denied
Message Permission denied
File Hierarchical.php
Line 185
Type Exception

I've searched the forum and found a thread with the following suggestion:

rm -rf /var/lock/lmlib/container_locks
rm -f /usr/local/psa/var/cache/*
service sw-engine restart

This seems to do the trick, at least I'm able to login to the panel.

However after a reboot of the server the same problem occurs again. I have checked and compared a lot of things but I can't find any difference between the 2 servers that work well and the 2 servers that has this problem.

Attached is the last entries of the panel.log with the error. I need to have these 2 servers up and running correctly as soon as possible. Please help me out here!

Kind regards,

Gijsbert

UPDATE:

After doing the following command:

rpm -Uvh http://autoinstall.plesk.com/PSA_12...re-12.0.18-cos6.build1200140606.15.x86_64.rpm --force

It seems that the panel is working fine now. However, I can I verify that things work again the way it should be?

 

[abdi]

Hi,
Tried disabling SELinux?

 

[gijsbert]

SELinux is disabled. I noticed that when running "stat /var/lock/lmlib/container_locks" (see thread: http://kb.odin.com/en/122288) the uid was not root but psaadm. I did a chown root:lock-manager on this directory.

But after making the owner root I followed this thread: http://kb.odin.com/en/125173 and recreated the container_locks directory like this:

rm -rf /var/lock/lmlib/container_locks
rm -f /usr/local/psa/var/cache/*
service sw-engine restart

When plesk creates the new container_locks, it is again owned by psaadm and not by root. There is something wrong, but I cannot find out where or how this ownership psaadm comes from.

Help will be appreciated.

 

[IgorG]

Have you seen this article?
https://kb.odin.com/en/122077

 

[gijsbert]

Hi Igor,

I checked the article, but /dev/shm is mounted as a tmpfs filesystem and psaadm is inside the group lock-manager. Maybe it's just normal behaviour that, once you delete the container_locks, it will be re-created with uid psaadm? I upgraded about a 100 servers and all have root as uid on the container_locks except 3 servers and I just don't understand why, there must be something wrong or corrupt...

 

[IgorG]

Do you have 12.0 or 12.5 version?
I can't reproduce it on 12.0

 

[gijsbert]

Hi Igor, version 12.0.18

 

[IgorG]

Have you tried upgrade to latest 12.5 version?

 

[gijsbert]

Beside the problems I noticed I also got some warnings running phpMyAdmin. I re-installed all patches and the problem seem to be fixed.

 

[gijsbert]

unfortenately the problem is still there. I have no idea what happens. I re-installed all patches and things seem to be ok for a couple of days, but today the container_locks dit have problems again when a customer removed a database and created a new one. So what I did was:

ls -la /var/lock/lmlib/
drwxrwx--- 3 root lock-manager 4096 Dec 16 17:55 container_locks

/etc/init.d/sw-engine stop; rm -rf /var/lock/lmlib/container_locks; rm -f /usr/local/psa/var/cache/* ; /etc/init.d/sw-engine start
Stopping sw-engine-fpm: [ OK ]
Starting sw-engine-fpm: [ OK ]

ls -la /var/lock/lmlib/
drwxrwx--- 3 psaadm lock-manager 4096 Dec 16 21:13 container_locks

So the owner is now psaadm and then it seems to work again (for now). But my guess is that the problem will re-occur within a few days.

 

[gijsbert]

Here is the attached panel.log. The error refers to the following article: https://kb.odin.com/en/125173. But that's exactly what I did (see post above).

 

[IgorG]

I would suggest you contact Plesk Support Team for investigation and fixing this issue directly on your server.