OOM Killer Report

The Out-of-Memory (OOM) killer is a default component of the Linux kernel. In case of necessity (i.e. no free memory), it suspends the currently unimportant processes to release RAM for more prioritized services. The selection of the process to be killed is based on principles of losing the minimal amount of work done and simultaneously recovering the most considerable amount of memory.

The platform runs OOM killer on all containers by default, allowing them to continue functioning when the RAM limit is reached. The Reports > OOM Killer section of the JCA provides full information on each out-of-memory issue occurred on the platform:

  • Domain - a name of the affected environment (click to locate it in the Cluster > Environments section automatically)
  • Host - a hardware server with the problematic container
  • User ID - unique identifier of the appropriate environment owner (click to locate this client in the Users section automatically)
  • Container ID - a unique identifier of the container on the host
  • Node ID - unique identifier of the container in the dashboard
  • Node Type - a software stack type
  • Process Name - a killed process name (refer to the linked documentation for detailed information on processes that may be affected by OOM killer)
  • Process ID - unique identifier of the process within the container
  • Time - time and date of the OOM killer execution

jca oom killer reports

If needed, you can set a period of search (Start and End Date) and specify the following criteria to filter the list of killed processes:

  • Domain or Alias of the affected environment
  • User ID of the environment owner
  • Process Name terminated by OOM killer

When ready, click Search to update a list and see the results. Use the Clear button to drop the applied search parameters and see the complete list of killed processes.

What’s next?