Skip to content

Mount checks for scheduled backups#

Before each scheduled backup run, MKE 4 verifies that schedule.path resolves to the expected mount. The check prevents MKE 4 from writing archives to the underlying directory after the mount drops. MKE 4 runs the check regardless of the storage type at schedule.path.

How MKE 4 identifies the expected mount#

MKE 4 expects schedule.path to be on a mounted filesystem when either of the following sources exists on the controller:

  • An /etc/fstab entry for the mount point that contains schedule.path, which can be schedule.path itself or an ancestor directory
  • The status.mountPoint value of the controller's last Completed scheduled backup at the same schedule.path

MKE 4 has no backup history to check in the following cases:

  • Initial installation
  • The first run after an upgrade from an MKE 4 version that does not record a mount point
  • The first run on a controller with no prior Completed backup recording a mount point for this schedule.path
  • The first run after you change schedule.path

In these cases, MKE 4 detects a missing mount only if an /etc/fstab entry declares it.

Declare the mount in /etc/fstab#

Add an /etc/fstab entry for the mount at schedule.path to /etc/fstab on every controller that can run the scheduled backup. With replication: LeaderOnly, add the entry on every controller that can become leader, not only the current leader.

MKE 4 accepts any entry at schedule.path, regardless of its mount options, including noauto and _netdev.

Outcomes of a missing mount#

Condition Result
The expected mount is missing. MKE 4 records the attempt as Skipped, sets the label mke.mirantis.com/backup-result=Skipped, and emits a BackupSkipped Warning event.
MKE 4 cannot determine whether the mount is present; for example, it cannot read /etc/fstab or query the controller's backup history. MKE 4 records the attempt as Failed.

In both cases, MKE 4 writes no archive to schedule.path. To inspect these records, refer to Check the backup history.

To recover, resolve the condition that stopped the run:

  • For a Skipped attempt, restore the mount.
  • For a Failed attempt, resolve the inspection error that MKE 4 reports. Restoring the mount does not fix an unreadable /etc/fstab or a backup history API error.

The next scheduled attempt retries automatically, without a service restart or file cleanup.

Check which mount a backup used#

Scheduled attempts that successfully inspect the backup filesystem report it in status.filesystem:

status.filesystem: nfs4 10.0.0.5:/exports/mke on /mnt/backups

The status.filesystem field is absent in the following cases:

  • MKE 4 skips the attempt while it waits for configuration synchronization, before mount inspection.
  • MKE 4 fails the attempt before it probes the current mount, for example, because of a raw inspection error.

If the attempt fails after MKE 4 probes the mount, status.filesystem still reports the probed mount, even though the attempt status is Failed. For example, this occurs when /etc/fstab is unreadable or when the backup-history API returns an error.

Completed scheduled records also report status.mountPoint. Later runs on the same controller and schedule.path use this value as the expected mount.

Scope of this check#

  • The check confirms that the expected mount is present. It does not confirm that the server behind it is reachable or that reads and writes against it complete in a timely manner. A network share can remain mounted while its server becomes unreachable, and filesystem operations against it can block.
  • The check does not verify which storage backs the mount. MKE 4 accepts a different filesystem mounted at schedule.path, and it accepts unmounting and remounting the same storage between runs, as long as a mount is present when the run starts.