[SERVER-64005] Add metric to track the 'fullness' of the TCP accept queue Created: 25/Feb/22 Updated: 29/Oct/23 Resolved: 27/Sep/22 |
|
| Status: | Closed |
| Project: | Core Server |
| Component/s: | None |
| Affects Version/s: | None |
| Fix Version/s: | 6.2.0-rc0 |
| Type: | Improvement | Priority: | Major - P3 |
| Reporter: | Blake Oler | Assignee: | Blake Oler |
| Resolution: | Fixed | Votes: | 0 |
| Labels: | diagnostics | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Issue Links: |
|
||||||||
| Backwards Compatibility: | Fully Compatible | ||||||||
| Sprint: | Service Arch 2022-07-25, Service Arch 2022-08-08, Service Arch 2022-08-22, Service Arch 2022-09-05, Service Arch 2022-09-19, Service Arch 2022-10-03 | ||||||||
| Participants: | |||||||||
| Story Points: | 3 | ||||||||
| Description |
|
This ticket concerns connections incoming to a mongos or mongod – we call the incoming connection path the ingress layer. Before the ingress layer accepts a connection, this connection sits in the accept queue on the TCP layer. The accept queue has a fixed size, and when the queue is full, the OS will start dropping incoming connections at the pre-mongo level. An overflowing queue means that either a mongo process is not accepting connections fast enough, or that the kernel is throttling connection establishment. Of particular note are generic TCP metrics TcpExtListenOverflows and TcpExtListenDrops. |
| Comments |
| Comment by Githook User [ 21/Sep/22 ] |
|
Author: {'name': 'Blake Oler', 'email': 'blake.oler@mongodb.com', 'username': 'BlakeIsBlake'}Message: |
| Comment by Bruce Lucas (Inactive) [ 01/Mar/22 ] |
|
This would be great - is there a way to track the accept queue length? Regarding TcpExtListenOverflows and TcpExtListenDrops - those are already captured in FTDC in the "netstat" section, but of course this only gives a signal once the queue fills up. |