What happens if I make my log too small?

Points we need to consider when estimating the minimum size of the log.

If you make your log too small:

  • Long running transactions will get backed out.
  • The next checkpoint wants to start before the previous one has ended.

Important: No matter how inaccurately you estimate the size of our log, data integrity is maintained.

See Use checkpointing to ensure complete recovery for an explanation of checkpoints. If the amount of log space left in the active log extents is becoming short, the queue manager schedules checkpoints more frequently.

A checkpoint takes some amount of time; it is not instantaneous. The more data that needs to be recorded in the checkpoint, the longer the checkpoint takes. If the log is small checkpoints can overlap, meaning that the next checkpoint is requested before the previous checkpoint has ended. If this happens error messages are written.

If long running transactions get backed out, or checkpoints overlap, the queue manager continues processing the workload. Short-lived transactions continue running as normal.

However, the queue manager is not running optimally and performance might be degraded. We should restart the queue manager with sufficient log space.

Parent topic: Calculating the size of the log