Immediately restart the cluster to quickly resolve the performance issues without any investigation, as this is the fastest way to restore functionality.

Investigate the cause of the memory leak by reviewing the application logs and code, apply necessary fixes to the code, and then restart the cluster to ensure the fixes are effective and to clear any residual issues.

Continue running the cluster as is, monitoring the situation closely, but take no immediate action to avoid any disruption to ongoing processes.

Restart the cluster and immediately rerun the same jobs without any changes, assuming the restart will permanently resolve the memory leak issue.