Using the marker file interface for universal share operations
A VPFS scheduler daemon provides a marker file interface to perform various universal share operations.
To configure the VPFS scheduler daemon
- Enable the VPFS scheduler daemon.
The VPFS scheduler is enabled automatically after one of its following features is enabled.
The universal share concurrent dump control
See the topic To configure the universal share concurrent dump control for more information.
The automated universal share backup.
See the topic To configure the automated universal share backup for more information.
- The VPFS scheduler daemon starts automatically after at least one feature is enabled. To start VPFS scheduler manually, run the following command:
/usr/openv/pdde/vpfs/bin/vpfs_sched start
- Configure the following parameters of the VPFS scheduler daemon.
VpfsScheduler.vpfsSchedInterval
The interval time in seconds between each loop of operations.
VpfsScheduler.vpfsSchedLogLevel
The minimum logging level.
Allowed values: error, warning, information, and debug.
To configure the universal share concurrent dump control
- Enable the universal share concurrent dump control.
/usr/openv/pdde/vpfs/bin/vpfs_actions --action setVpfsConfigValue --key enableShareDumpThrottling --value true
- Configure the following parameters of universal share concurrent dump control.
enableShareDumpThrottling
Specify whether the share dump throttling feature is enabled or not. If false, no share concurrency is enforced.
VpfsScheduler.maxConcurrent ActiveClientHostSharePairs
The maximum "Client Host-Share Pairs" that can be active at once.
When this limit is reached, any requested dumps that belong to the new "Client Host-Share Pair" are put into a queue. They are activated when other "Client Host-Share Pairs" are set to inactive.
A higher value means that more dumps happen concurrently. The higher concurrent dumps may not yield a higher aggregate throughput and the concurrency needs to be tuned for a given system.
VpfsScheduler.shareDumpsInactive Threshold
The minimum time in seconds since the last IO operation in a share that the VPFS scheduler daemon waits. After the specified time, "Client Host-Share Pair" is inactive.
For example, if this value is set to 300, and no IO operations have been detected in a share in at least 300 seconds, the "Client Host-Share Pair" is removed from the active list.
VpfsScheduler.enableShareDump ThrottlingReporting
Specify whether to log the history of movement of "Client Host-Share Pairs" within the throttle table.
Movement includes "Client Host-Share Pairs" being queued, activated, and deactivated.
- Copy dump scripts to the existing shares that are created before feature is enabled.
/usr/openv/pdde/vpfs/bin/vpfs_actions --action syncShareScript
- Add the dump starts script to the start of the workload and use the dump_end script to the end of the workload.
For example:
Windows batch dump scripts
\\hostname\ushare-smb1\.share-builtin-scripts\dump_start.bat --dump-id <dump_id> Existing workload dumping data to universal share \\hostname\ushare-smb1\.share-builtin-scripts\dump_start.bat --dump-id <dump_id>
Linux bash dump scripts
/mnt/ushare_smb1/.share-builtin-scripts/dump_start.sh --dump-id <dump_id> Existing workload dumping data to universal share /mnt/ushare_smb1/.share-builtin-scripts/dump_end.sh --dump-id <dump_id>
Note:
You must provide the dump_id. A randomly generated ID is recommended but it must be different from other dump scripts that are used in the same ClientHost-Share Pair. For SQL server, it is recommended to use the Job ID as the dump_id.
View the throttling table for existing universal dumps.
/usr/openv/pdde/vpfs/bin/vpfs_actions -a dumpThrottleTable
To configure the automated universal share backup
- Enable the automated universal share backup.
/usr/openv/pdde/vpfs/bin/vpfs_actions --action setVpfsConfigValue --key VpfsScheduler.autoUShareBackupEnabled --value true
- Configure the following parameters for automated universal share backup:
VpfsScheduler.autoUShareBackup Enabled
Enable the automated universal share backup feature.
VpfsScheduler.autoUShareBackup CheckInterval
Interval for checking touch files on universal share.
VpfsScheduler.autoUShareBackup JobCheckInterval
Interval for checking the status of running universal share backup job that is triggered by this feature.
VpfsScheduler.autoUShareBackup PolicyFile
The location of policy file that is configured in the file system.
VpfsScheduler.autoUShareBackup OutputFile
The location of an output file in the file system. The completed universal share backup job detail is written into the file.
- Configure the universal share backup policy to be triggered by automated universal share backup.
Configure a policy file that consists of a list of universal share policies that need to be triggered. The policy file location is a configurable path VpfsScheduler.autoUShareBackupPolicyFile, which can be configured using the vpfs_actions command.
Each entry of the policy configuration consists of the policy name, client name, and the marker file location.
For example,
pl-ushare-nfs-1,host.domain.com,/mnt/vpfs_shares/ usha/ushare-nfs-1/backup pl-ushare-nfs-2,host.domain.com,/mnt/vpfs_shares/usha/ushare -nfs-2/backup
When the policy file is modified, the policy configuration is automatically reloaded.
- Configure the backup schedule rotation.
The backup schedule is configured with a key-value pair where the key is the weekday (Monday - Sunday) and the value is the list of universal share backup type. The default configuration is FULL, INCR, CINC every day from Monday to Sunday.
The following are the supported backup types:
FULL: Full backup
INCR: Differential incremental backup
CINC: Cumulative incremental backup
- Create a marker file in the configured location in the universal share. After you configure the market file path in the policy file VpfsScheduler.autoUShareBackupPolicyFile, create marker file in designated location to trigger the backup.
The marker must be named with the following format:
BACKUP_SUCCESS_<XXXX>_touch_<Date> or BACKUP_SUCCESS_<XXXX>_touch_<Date>_<Timestamp>
Where,<XXXX> is optional and can be any string with no space.<Date> must be in the YYYYMMDD format.<Timestamp> should be a unique timestamp each time in a day such as HHmmSS.
The data and the timestamp are converted to a weekday. Then it is used to determine the backup schedule.
A universal share backup is triggered when it finds a new marker file that has a creation timestamp older than 5 minutes.
- Verify that the backup is successful.
To check the status of the backup job, see NetBackup web UI. Once the backup is successful, the job information is also saved to the file that can be configured with VpfsScheduler.autoUShareBackupOutputFile.
The format of the output:
<job_id>,<schedule>,<state>,<status>,<total_size_kb>KB,<start_time>, <end_time>,<client>,<policy>,<ushare_mount>