Plesk 8.0 causes server crash once every 1-3 days.

[fishnyc22]

Hey...

Its running FedoraCore 4 running with Plesk 8.
Do you know what backupmng does?

 

[jwdick]

Do you know what backupmng does?

backupmng is the cron job that runs your scheduled backups. Disabling backupmng will disable scheduled backups.

 

[fishnyc22]

I'm running the server with 3 of my own sites on it. (no other clients accounts or anything like that)

Nightly I run this script (see below) to back everything up. Should I be running anything more than this?(forgive my ignorance)

The Schedule backups don't allow FTP so I found this option on this forum which I thought worked great. Also, what scheduled backups run every 15 min? The domain specific Scheduled Backups offer daily, weekly, monthly.

What does the domain specific scheduled backups do different.

Thx

----------------------Backup Script----------------------------
#!/bin/sh

# Setting variables

# Setting date to name the backupfile with todays date
date=`/bin/date "+%Y%m%d"`

# Your username on the ftp server
username="myUSERNAME"

# Your password on the ftp server
passwd="myPASSWORD"

# FTP servers adress
ftpserver="mySERVER"

# Local directory where the backupfiles will be temporarily placed

backupdir="/usr/local/src"

# You do not need to change anything below this line

# Running Plesk backupscript. Use line with the command nostop is not recommended and are not disabling webservices
# as long the pleskbackup is running. If preferedd comment in an # before that line an uncomment the next line and
# the backup will run with apache, psa and qmail stopped on the server. That is the recommended mode.

/usr/local/psa/bin/pleskbackup --all $backupdir/dump$date.svr

# Changing to dump directory

cd $backupdir

# Transfering files to your backupserver

ftp -n $ftpserver <<!EOF!
user $username $passwd
binary
cd /fif_backups
prompt
mput dump*
quit
!EOF!

cd /$backupdir
rm -rf dump*

echo "DONE"
#Done

 

[renegadeavenger]

fish, try to run a fedora update and see if it helps resolve this. I am trying to find a solution for my suse servers as well.

 

[jwdick]

The script you are using is to replace the "scheduled backups" that are run with backupmng until SWSOFT fixes the ftp backup bugs.

If you are happy with the success of the script (I use that also) then you can safely disable backupmng in the crontab.

Also, what scheduled backups run every 15 min? The domain specific Scheduled Backups offer daily, weekly, monthly.

Backups run as often as you schedule them for (i.e. daily, weekly, monthly, etc.) The "backupmng" job runs every 15 mintues, to see if you have a backup scheduled. If you have 500 domains on this server, you would not want to backup them up all at the same time so you would schedule them 15 to 20 minutes apart throughout the day.

Just remember, IF you delete the cron job "backupmng" from the crontab, then later decide that you want to use the scheduled backups from the plesk interface, you will have to re-enter the cron job. So you might want to save the settings so you can put it back.

 

[fishnyc22]

Hey Jack, Thanks for the detailed response. I really apprecaite it. I'm going to disable it and see if the crashes stop.

I may however want to split the backups into separate tasks down the road a bit (and when FTP is enabled again)

Since you use this script as well, quick question. Say I did have a catatrophic crash...what would be my plan of action to use the backup to restore the system. I should probably know that.

thanks for the help.

 

[jwdick]

You would run pleskrestore first to create a map file, then again the second time to restore the domains. You can choose to restore only one domain or all of them. See detail restore instrucitons here:

http://download1.swsoft.com/Plesk/Plesk8.0/Doc/plesk-8-backup-restore-users-guide/index.htm

 

[fishnyc22]

Awesome. I hope I never have to use it.

Thanks again.

 

[renegadeavenger]

one of my servers crashed again, this is becoming very ridiculous.

 

[pavel]

renegadeavenger, do you have physical access to the server? If so, what do you see on the system console after the crash? Symptoms, outlined in this thread sounds very similar to possible problems with hard drives and/or disk controller. If this is the case you won't be able to see anything in logs but there is a good chance that kernel was able to print something to the system console

 

[renegadeavenger]

Pavel, if this was the case, this problem would not exist on 5 other servers that have Plesk 8.0 installed on them.

 

[pavel]

renegadeavenger, this might be everything from buggy software to faulty hardware or, say, problems with hw drivers in the version of the system kernel you have installed on this machine. This problem cannot be duplicated in the lab environment and it appears that it is affecting only limited number of Plesk 8 customers. It is quite obvious that Plesk 8 is currently installed on a bit larger number of servers than 5-10 mentioned in this thread.
It is not that easy to fix something that cannot be duplicated. As such in order to understand the problem better some additional information is required. The very first step in troubleshooting of this problem is to take a look at the system console for any errors that was reported by the kernel right before machine died.
Also according to this thread you've disabled plesk backup utils and still have machine(s) crashing periodically, so this leads me to believe that initial assumption that backup utils are related to these crashes is not correct. Don't get me wrong, I'm not trying to say that this is something 100% non-Plesk related. All that I'm trying to do is to collect more information to get a bigger picture of the problem you're having.
What kind of disk controller do you have and what version of kernel is running on this machine? Do you have similar hw / kernel configuration on all affected servers?

 

[renegadeavenger]

All the current servers have suse 10.0 running on stock kernel. There are no issues in the console log. This was my step for all setups with suse 10.0.

Installed with minimal install ( without graphics ) and installed with Apache/web services and C++ compilers.

Yast updated everything to latest version

Installed Plesk 8.0 or 8.01 not installing acronis, tomcat, or any of the game server modules.

In terms of HD controller, I am unsure what is currently on the board.

 

[renegadeavenger]

also forgot to mention, I installed postgresql through Yast before installing Plesk.

 

[renegadeavenger]

pavel, please communicate in ticket: 170117

thanks

 

[renegadeavenger]

by the way, I had also replaced hardware, ran mem checks, stressed cpu.. all was fine. I really think it is something to do with the crons that are running, it doesn't have to be exactly backupmng but it is definitely something that is causing the server to randomly crash without any error messages. I am 100% sure this is really not hardware based as identical hardware running suse 10 and 7.5 reloaded it is stable. Just all of my new servers with suse 10.0 and 8.0 are all having these random crashing issues.

 

[Hultenius]

I have the very same problem on Plesk 7.5.4. The scheduled backups seems to kill one of my servers sometimes (~ 30-50% chance on each run) but only when it does backups for one specific domain. I tried to delete the scheduled backup settings for this domain and guess what? The problem stopped.

Plesk 7.5.4 on CentOS 4.2

 

[renegadeavenger]

hmm I am just curious guys.. what version of the kernel are you guys running on?

I am on 2.6.13-15

 

[PE-Steve]

Originally posted by renegadeavenger
hmm I am just curious guys.. what version of the kernel are you guys running on?

I am on 2.6.13-15

After updating to 2.6.17.1, and Plesk 8.0.1, I haven't seen the server crash yet, although I'm definitely still waiting. It's actually been up now for 20 days.

 

[fishnyc22]

I'm scared to upgrade to 8.0.1