KB-1039 "Unable to rename writing_X.kdb: "domain"" error thrown during checkpoint

Symptoms

The checkengine.bat|sh script shows intermittent checkpoint failures for one or more engines. In the gw_*.log file, the following similar lines are generated when an engine is trying to checkpoint:

2015-09-10 08:00:02 [PX021] INFO .a.gw.ssa "Standard save initiated"
2015-09-10 08:01:58 [PX021] WARN .a.gw.ssa "Database save took more than 60 sec - Please review hardware and software configuration."
2015-09-10 08:01:58 [PX021] INFO .a.gw.ssa "Notify database of switch (fast track)"
2015-09-10 08:01:58 [PX021] INFO .a.gw.ssa "Initiating switch 4444 4216843"
2015-09-10 08:01:58 [PX021] WARN .a.gw.ssa "Server failed to switch to 4444 4216843: Unable to rename writing_pe4444.kdb: "domain""

Cause

There are insufficient resources on the server to complete the checkpoint. In order for the checkpoint to complete there must be sufficient memory and disk space on the server(s) hosting the engines.

Action

There are multiple options to resolve this:

• Increase the hardware on the server. Make sure there is enough free memory and disk space available for the new engines created during checkpointing.
• Attempt to lower the size of the engines.
• Stagger the checkpoints so not all engines are checkpointed at once. This will reduce the memory requirements during checkpointing, but is only a temporary fix. Pass in the name of an engine as a parameter to the checkpoint-suite.bat|sh script:
• • forums
  • notify
  • notify-email
  • channels
  • content
  • collaboration-statistics
  • personalization
  • portal
  • process-design
  • process-analyticsX (where X is 0-15)
  • process-executionX (where X is 0-15)

When selecting a process-analytics or process-execution engine, you must specify the engine number (such as process-analytics0). Refer to the Configuring Application Checkpointing documentation for more information.

Affected Versions

This article applies to all versions of Appian.