Plesk Scheduled FTP Backup

[PleskUser3]

Backup

I have a problem with Scheduled backups, and this has been going on for a few months now. I have tried alot of suggestions that I have found on forums and such, many just haven't worked.

I would like to start by clarifying that I have updated the FTP details since the 8.4 installation, and I know that FTP works as if I run a manual backup to the FTP Repository through Plesk, it works fine everytime. This problem began after I upgraded from the default installation of 8.1.

I have around 10 domains that I am backing up. They are scheduled between 1am and 3.30am, and one backup occurs every 15 minutes. The server is set to only allow 1 backup to run at any time.

Yesterday morning, I woke up, and checked my local FTP site, and every backup had completed successfully at the time given (give or take 10 minutes). This morning I have woken up, and checked the local FTP site, and only 1 backup has completed.

I have the email option set on each backup, so that it will email me if there is a problem - no emails in my inbox from Plesk this morning - except this one just came through:

Domain: mydomain.com
Plesk entry point: https://myplesk.installation:8443/

Following error is occured during scheduled backup process:

Unable to create UNIX socket /usr/local/psa/tmp/agent-runner.12121: Connection refused

If i list the /var/lib/psa/dumps/tmp folder, I can see:

drwxr-x--- 8 root psaadm 1024 Jul 5 01:26 2008-07-05-01.25.22.262844

So i know that it started the backup, but has got stuck on something. If I look at running processes on the unix box, I can see:

7299 psaadm 15 0 22812 2116 1516 S 0 0.0 0:00.00 backupmng
7344 psaadm 18 0 22812 892 292 S 0 0.0 0:00.00 backupmng
7345 psaadm 15 0 77468 26m 5636 S 0 0.1 0:05.15 scheduled_backu
7419 root 16 0 73220 11m 1944 S 0 0.0 0:17.40 agent_runner

So the process is still sitting there, but what its doing is anyone's guess. I have also tail -20 the last 20 lines of the agent-runner.log file:

[root@root tmp]# tail -20 2008-07-05-01.25.22.262844/agent-runner.log
<preparing></preparing>
</working>
</action-response>

2008-07-05 09:26:47,305 INFO Packet succesfully validated.
2008-07-05 09:26:50,307 INFO Incoming packet:
<?xml version='1.0' encoding='UTF-8'?>
<action>
<get-status/>
</action>

2008-07-05 09:26:50,308 INFO Packet succesfully validated.
2008-07-05 09:26:50,308 INFO Outgoing packet:
<?xml version="1.0" encoding="UTF-8"?><action-response>
<working>
<preparing></preparing>
</working>
</action-response>

2008-07-05 09:26:50,309 INFO Packet succesfully validated.

You'll notice that the time on these log file lines is the time now - so the backup is doing something, but what??

I have contacted my host about this, and they are baffled and every time I ask that they look into it or check something else, it takes them a few days to respond whilst they negotiate with Parallels. I just need this fixing - I have had to resort to creating a cron job that tar's up the httpdocs folder and FTPs it to me - this cron job I created has worked 100% every night over FTP - so again to reiterate, I think the FTP part is working fine!

 

[poldito]

I've seen you have posted this issue several times and nobody has answered. I know this is a bug (BIG BUG) that use to happened also in some older versions of Plesk because I found some older posts in this forum.

I have also those kind of problems lately, since version 8 so far. My conclusion is that it happened when a new cron job for backup manager is fired and the former one (even from other domain) is still active (slow or lost connection?); then processes start sitting there and they never finish.

My way to solve this issue till now was to kill every process for backup manager, agent runner, etc., go and refresh the backup schedule and it goes again working normally.

Now, after upgrading to 8.6 it is very bad... I am at the point that backups never finish and every time backup manager is fired (every 15 minutes) it starts again and again every job from every domain in the web server! I had to stop cron task for backups because I had a permanent 24 hours trafic between web server and ftp server with a long queue of jobs.

The problem now is:
- backups are written to ftp server, then it is working fine and permissions are OK
- backup manager is not able to READ from ftp server the files it puts, so next time it starts again putting a new backup file

I have tried to work without firewalls and NAT's to avoid any interference, I have tested other systems and FTP server software (like Filezilla and Faststream) at the FTP Server side and the problem is always the same, so it has to be backup manager having trouble to list files at ftp server. I use separated dedicated machines for web servers and ftp servers.

If you found any solution, please post it here; I have no clue at this point of how to solve this issue.