Based on an awesome presentation by @mikecote , there has been significant improvements and changes in Kibana task manager regarding performance, poll_interval, and scaling recommendations. The improvements were released in 8.18 and I guess all the migrated docs in the new system for 9.x were based on 8.17 information.

The purpose of this issue is to review the following docs in order to determine needed changes:

• Kibana task management

• Performance and scaling guide

Some notes from that meeting:

• How to scale efficiently:
  • Recommend to upgrade to 8.18
  • Fine tune xpack.task_manager.capacity.
  • Change xpack.alerting.rules.maxScheduledPerMinute if you want more than 32k rules per minute.
  • Isolate background work using node.roles on-prem or by scaling Kibana to 8G+ in ECH.

Based on our current docs:

• I see we still say An Elasticsearch task index is polled for overdue tasks at 3-second intervals --> Hasn't this been changed to 0.5 secs @mikecote ?
• The Rough throughput estimation section of the scaling guide mentions A default Kibana instance can support up to 200/tpm, and maybe that has changed.

It would be great if @mikecote , or any @elastic/kibana-pm could review our 2 main docs about this area and suggest the necessary changes.

cc: @shainaraskas