[SERVER-25733] Do clean shutdown in mongos under asan builds Created: 22/Aug/16  Updated: 06/Dec/22

Status: Open
Project: Core Server
Component/s: Internal Code
Affects Version/s: 3.3.11
Fix Version/s: None

Type: Improvement Priority: Minor - P4
Reporter: Mira Carey Assignee: Backlog - Service Architecture
Resolution: Unresolved Votes: 0
Labels: platforms-re-triaged
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Assigned Teams:
Service Arch
Backwards Compatibility: Fully Compatible
Participants:

 Description   

For ASAN, mongod makes sure to shut down the listener, close all connections and wait for worker threads to exit before shutting down. We don't do this in mongos and doing so will probably make ASAN builds more robust for mongos.

We may also need to investigate the performance cost of doing this in mongos (how much longer the sharding jscore passthrough suite takes)



 Comments   
Comment by Lauren Lewis (Inactive) [ 25/Apr/22 ]

Moving to open and within PM-774 for prioritization.
cc: blake.oler@mongodb.com

Comment by Daniel Morilha (Inactive) [ 19/Apr/22 ]

After speaking about this with Sam, there are other areas that should be assessed specifically for mongos.

Comment by Daniel Morilha (Inactive) [ 18/Apr/22 ]

This ticket doesn't clarify how mongod and mongos differed back in 2016, making it difficult to assess if the problem still exists.

The quick ten minutes assessment reveled that in what regards ServiceEntryPointImpl , both mongod and mongos act similarly when built with address sanitizer support.

I ran a simplified version of noPassthrough/exit_logging.js "grepping" for a message which is only shown when ServiceEntryPointImpl runs its tear down to its full extension : "shutting down listeners, closing all connections and waiting for worker threads to exit before shutting down", and in both cases the message shows.

$ time { ninja -j400 -v -d keeprsp &&  buildscripts/resmoke.py run --repeatTests=1 jstests/noPassthrough/exit_logging.js | grep --color '6367401'; }
 
ninja: no work to do.
[js_test:exit_logging] d20020| {"t":{"$date":"2022-04-18T15:50:34.387+00:00"},"s":"I",  "c":"NETWORK",  "id":6367401, "ctx":"thread25","msg":"Shutting down service entry point and waiting for sessions to join"}
[js_test:exit_logging] s20026| {"t":{"$date":"2022-04-18T15:51:14.742+00:00"},"s":"I",  "c":"NETWORK",  "id":6367401, "ctx":"thread27","msg":"Shutting down service entry point and waiting for sessions to join"}
[js_test:exit_logging] s20025| {"t":{"$date":"2022-04-18T15:51:16.506+00:00"},"s":"I",  "c":"NETWORK",  "id":6367401, "ctx":"SignalHandler","msg":"Shutting down service entry point and waiting for sessions to join"}
[js_test:exit_logging] d20021| {"t":{"$date":"2022-04-18T15:51:19.661+00:00"},"s":"I",  "c":"NETWORK",  "id":6367401, "ctx":"SignalHandler","msg":"Shutting down service entry point and waiting for sessions to join"}
[js_test:exit_logging] c20023| {"t":{"$date":"2022-04-18T15:51:26.658+00:00"},"s":"I",  "c":"NETWORK",  "id":6367401, "ctx":"SignalHandler","msg":"Shutting down service entry point and waiting for sessions to join"}
[js_test:exit_logging] c20024| {"t":{"$date":"2022-04-18T15:51:26.659+00:00"},"s":"I",  "c":"NETWORK",  "id":6367401, "ctx":"SignalHandler","msg":"Shutting down service entry point and waiting for sessions to join"}
[js_test:exit_logging] c20022| {"t":{"$date":"2022-04-18T15:51:27.217+00:00"},"s":"I",  "c":"NETWORK",  "id":6367401, "ctx":"SignalHandler","msg":"Shutting down service entry point and waiting for sessions to join"}
[j0] {"t":{"$date":"2022-04-18T15:51:30.156+00:00"},"s":"I",  "c":"NETWORK",  "id":6367401, "ctx":"SignalHandler","msg":"Shutting down service entry point and waiting for sessions to join"}

For a better assessment, I would need to understand what other places mongos does what the ticket describes, and if it acts accordingly. Also, I am lowering the priority of this ticket.

Comment by Lauren Lewis (Inactive) [ 21/Dec/21 ]

We haven’t heard back from you in at least 1 year, so I'm going to close this ticket. If this is still an issue for you, please provide additional information and we will reopen the ticket.

Generated at Thu Feb 08 04:10:03 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.