A scheduled backup in Plesk fails but files remain in the file system Print

This article involves working on the server over SSH with root privileges. If you do not have root access, or would rather not do this yourself, open a ticket from your client area.

Symptoms

• A scheduled backup does not appear in Plesk under Subscriptions → example.com → Backup Manager.
• The tar/tgz archives are nevertheless present in the file system, but the backup_info_yymmddxxxx.xml file is missing:

# ls -l /var/lib/psa/dumps/clients/johndoe/domains/example.com/
-rw-r----- 1 root root      45 backup_apache-files_1808081427.tgz
-rw-r----- 1 root root    2713 backup_conf_1808081427.tgz
-rw-r----- 1 root root 2241374 backup_domainmail_1808081427.tgz
-rw-r----- 1 root root 9997985 backup_user-data_1808081427.tgz
drwxr-xr-x 4 root root      46 databases

• The log /var/log/plesk/PMM/backup-yyyy-mm-dd-hh-mm-ss-xxx/backup.log contains errors such as:

INFO: TransportError: Transport error: Unable to get directory listing:
No such file or directory: "/dumps/clients/johndoe/domains/example.com/databases/db_name"

ERROR Runtime error: The dump has content errors!
at /usr/local/psa/admin/bin/plesk_agent_manager line 1297.

Cause

A backup that failed the content validation check was not removed correctly. Its files remain in the file system without the index file, so Plesk does not show the backup in Backup Manager but the space is still occupied.

What to check first

Before anything else, check the free space on the partition holding the dumps:

# df -h /var/lib/psa/dumps

A full disk is by far the most common reason for a backup to fail partway through, and leftover incomplete backups make the situation worse over time. If the disk is full, freeing space may be all that is required.

Removing the invalid backup

1. Connect to the server over SSH.
2. Check the dump storage path:

   # grep DUMP_D /etc/psa/psa.conf
   DUMP_D /var/lib/psa/dumps

3. Using that path and the date and time of the backup in yymmddHHMM format, remove the remaining files:

   # /usr/local/psa/admin/bin/pmm-ras --delete-dump-files \
       --dump-storage=/var/lib/psa/dumps \
       --dump-file-specification=backup_info_1808081427.xml

Use this command rather than deleting the files with rm: it removes the backup the way Plesk expects, leaving no inconsistent state behind.

Avoiding a repeat

Keep an eye on the number of backups retained in the schedule and on the free space on the server: incomplete backups accumulate precisely when there is not enough room to complete them. Storing backups off the server rather than locally removes the problem entirely — see Backing up your account and websites in Plesk.

If the problem recurs after cleaning up, and the disk has free space, open a ticket from your client area and attach the backup log — the exact error in it determines what to do next.