[HADOOP-18060] RPCMetrics increases the number of handlers in processing

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.9.2
• Fix Version/s: 3.4.0
• Component/s: rpc-server
• Labels: pull-request-available
• Assignee: JiangHua Zhu
• Reporter: JiangHua Zhu
• Created: 22/Dec/21 04:04
• Updated: 31/Dec/21 08:41
• Resolved: 31/Dec/21 08:41

Description

When using RPC, we recorded a lot of useful information, such as Queue time, Processing time. These are very helpful.
But we can't know how many handlers are actually working now (only those that handle Call), especially when the Call Queue is very high. This is also not conducive to us optimizing the cluster.
It would be very helpful if we can see the number of handlers being processed in RPCMetrics.

Issue Links

• links to: GitHub Pull Request #3822