Finally I had to do a Plesk restore. I think I may have developed an ulcer trying to prepare for this event, but I have survived it.
I had a full Plesk dump following the backup method mentioned here.
The restore was quite straight forward:
1. I had to create an ip map file. My file contained the following information, since I was restoring Plesk to the same server:
10.0.0.24 -> eth0 : 10.0.0.24 255.255.255.255 (Only this single line). The file was called: ipmap
2. The restore required a shells map file. My file contained the following information, since I was restoring Plesk to the same server:
/bin/sh => /bin/sh
/usr/local/bin/rbash => /usr/local/bin/rbash
/usr/bin/false => /usr/bin/false
/bin/csh => /bin/csh
/bin/bash => /bin/bash
The file was called: shellsmap
I then ran a test restore, using the following command:
cat <backup_file_base>.* | /usr/local/psa/bin/psarestore -t --force --restore-admin --restore-server -m ipmap -s shellsmap -f -
Finally I ran a full restore, using the following command:
cat <backup_file_base>.* | /usr/local/psa/bin/psarestore --force --restore-admin --restore-server -m ipmap -s
shellsmap -f -
During the restore I had a large number of errors, about:
sh: – : invalid option
Usage: sh [GNU long option] [option] …
sh [GNU long option] [option] script-file …
It would however appear as if these errors had very little impact on the over all restore.
The biggest problem I had after the restore was that no FTP account was working any more. Luckily I wrote a little script to extract the FTP usernames and passwords from the database and then used the information to reset the password on the system account. This fixed the problem.