Appendix D. Ceph File System mirrors configuration reference
This section lists configuration options for Ceph File System (CephFS) mirrors.
cephfs_mirror_max_concurrent_directory_syncs
- Description
-
Maximum number of directory snapshots that can be synchronized concurrently by
cephfs-mirror
daemon. Controls the number of synchronization threads. - Type
- Integer
- Default
-
3
- Min
-
1
cephfs_mirror_action_update_interval
- Description
- Interval in seconds to process pending mirror update actions.
- Type
-
secs
- Default
-
2
- Min
-
1
cephfs_mirror_restart_mirror_on_blocklist_interval
- Description
-
Interval in seconds to restart blocklisted mirror instances. Setting to zero (
0
) disables restarting blocklisted instances. - Type
-
secs
- Default
-
30
- Min
-
0
cephfs_mirror_max_snapshot_sync_per_cycle
- Description
- Maximum number of snapshots to mirror when a directory is picked up for mirroring by worker threads.
- Type
- Integer
- Default
-
3
- Min
-
1
cephfs_mirror_directory_scan_interval
- Description
- Interval in seconds to scan configured directories for snapshot mirroring.
- Type
- Integer
- Default
-
10
- Min
-
1
cephfs_mirror_max_consecutive_failures_per_directory
- Description
- Number of consecutive snapshot synchronization failures to mark a directory as “failed”. Failed directories are retried for synchronization less frequently.
- Type
- Integer
- Default
-
10
- Min
-
0
cephfs_mirror_retry_failed_directories_interval
- Description
- Interval in seconds to retry synchronization for failed directories.
- Type
- Integer
- Default
-
60
- Min
-
1
cephfs_mirror_restart_mirror_on_failure_interval
- Description
-
Interval in seconds to restart failed mirror instances. Setting to zero (
0
) disables restarting failed mirror instances. - Type
-
secs
- Default
-
20
- Min
-
0
cephfs_mirror_mount_timeout
- Description
-
Timeout in seconds for mounting primary or secondary CephFS by the
cephfs-mirror
daemon. Setting this to a higher value could result in the mirror daemon getting stalled when mounting a file system if the cluster is not reachable. This option is used to override the usualclient_mount_timeout
. - Type
-
secs
- Default
-
10
- Min
-
0