Manage swap space

On cache-node-1, add a 2 GiB temporary swap file, enable it immediately, and make no persistent changes. Verify swap is active and report total swap and free swap afterwards. I will remove it after the maintenance window ends.

Before I change swap on prod-search-2, prepare a clear summary: current RAM usage, existing swap size and utilization, recent OOM events, and how adding swap would change observed memory pressure. Recommend the minimal swap size and whether a temporary swap file or partition is safer, plus any performance impact to warn stakeholders about.

During a nightly batch on analytics-01 the OOM killer started killing workers. Swap is currently zero and I don't know if adding swap would mask a memory leak. I fear adding swap will hide the root cause, but leaving none risks further OOMs. What's the most pragmatic next step to stabilize the job while preserving forensic data for debugging?

Our ops habit is to add swap whenever memory issues appear, which reduces urgency to fix memory leaks and bloats instance sizes. This defers real fixes and increases costs. What policy and practice should we adopt to decide when swap is acceptable, and how to pair temporary swap with follow-up actions to ensure leaks are tracked and resolved?