Space Management

This cycle is activated when the disk usage of one or more file systems exceeds the percentage full defined by the high-water value. When reached, LOSPACE policies are executed in an attempt to reach the low-water mark on each affected file system.

By default, the policies are executed once in this order on all affected file systems:

• Relocation policy
• Truncation policy

The high-water and low-water values are displayed by the GUI File System Monitor and can be modified via the StorNext GUI. By default, these values are set to 85% and 75%, respectively.

In contrast to the Emergency policies described in the next section, what's different in the behavior of the LOSPACE policies is that MINTRUNCTIME and MINRELOCTIME are not ignored. Only files that can be truly relocated and truncated are affected.

First, the relocation policy is executed and it continues until there are no more relocation candidates available at which time it terminates.

The call made to perform the LOSPACE relocation is:

If the file system usage still exceeds the high-water mark, the truncation policy is executed and it truncates all candidates until no further candidates are available, at which time it terminates.

The call made to perform the LOSPACE truncation is:

At this time the LOSPACE Space Cycle is complete for this file system. All other affected file systems are then processed in the same manner, first by running the relocation policy and then the truncation policy, if needed.

After all file systems have been processed, if any of them still exceed the high-water mark, a new LOSPACE cycle is started after a one-minute wait.

Thus, the low-water percentage may or may not be reached on any given file system. It depends solely on whether there are enough candidates available for relocation and/or truncation for that file system.

Emergency policies are executed when either of the following conditions is met for a file system:

1. When a file system encounters the NOSPACE event, i.e. a file write has failed because of lack of space.
2. When the file system usage is greater than 99%.

By default, the policies are executed once in this order:

1. Emergency truncation policy.
2. Emergency relocation policy.
3. Emergency store policy.

The emergency truncation policy finds up to the 3000 largest files that can be truncated, ignoring MINTRUNCTIME, and performs the truncation. This is executed once each time the NOSPACE condition is reached.

The call made to perform this emergency truncation is:

If the file system usage has not dropped below 100% after the emergency truncation, the emergency relocation policy is now run.

When the emergency relocation policy is run, it finds all files that can be relocated, ignoring MINRELOCTIME, and performs the relocation. As with the emergency truncation policy, this is executed once each time the EMERGENCY condition is reached.

The call made to perform the emergency relocation is:

If the file system usage is still not below 100% after the emergency relocation, an emergency store policy on the file system is performed.

An emergency store means that the request is placed first in the queue, and that any files in the file system which can be stored will be stored regardless of policy. As with the other emergency policies, it is run only once.

The call made to perform the emergency store is:

At this point the Emergency Space Cycle is complete.

Beginning with StorNext 6.2, you can use a CLI-only program (sn_offline_duration) to evaluate the effectiveness of truncation-policy settings.

• The program allows you to scan TAC log messages for truncation and retrieval operations and correlates them to glean information about the elapsed days between these operations.

• The program outputs data for a histogram report in terms of the number of days between truncation and retrieval and the number of occurrences for each day value. For example, when a large number of retrieve occurrences are shown in the report, you can delay a truncation policy by the day value correlating to those occurrences, in order to reduce the need for retrieves.

  Note: This increases the need for disk space, so you might have to perform additional tuning to configure the optimal settings for your workload.

• You can have a time window for the report by specifying the earliest truncation log message and the latest retrieval log message to be considered.

  Note: The report does not distinguish between policies or file systems because truncation and retrieval log messages do not provide this information.

For complete details about the sn_offline_duration program, see the StorNext 6 Man Pages Reference Guide.