[CDRIVER-3429] Turn off libmongoc shared memory counters by default? Created: 11/Nov/19 Updated: 05/Feb/24 |
|
| Status: | Backlog |
| Project: | C Driver |
| Component/s: | cmake, libmongoc, Performance |
| Affects Version/s: | None |
| Fix Version/s: | 2.0.0 |
| Type: | Improvement | Priority: | Minor - P4 |
| Reporter: | Kevin Albertson | Assignee: | Unassigned |
| Resolution: | Unresolved | Votes: | 0 |
| Labels: | None | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Issue Links: |
|
||||||||
| Epic Link: | CDRIVER-4577 | ||||||||
| Description |
|
By default, if shared memory is supported, libmongoc gathers statistics about client/cursor/socket usage. See the Basic Troubleshooting page for more info. They are currently enabled by default on Linux if shared memory is supported. They may be opted-out at configuration time with -DENABLE_SHM_COUNTERS=OFF or at runtime with the MONGOC_DISABLE_SHM. The shared memory segment is unlinked in mongoc_cleanup, but if a user does not call that function, or a crash occurs before that function is called, then this may result in shared memory segments being left around. Let's consider whether this should be an opt-in feature rather than opt-out. This would be a bit of a behavior change, but it's not clear to me how disruptive changing this default would be. |
| Comments |
| Comment by Kevin Albertson [ 06/Jan/20 ] |
|
In |
| Comment by Roberto Sanchez [ 02/Dec/19 ] |
|
kevin.albertson I see no reason why it would be disruptive. The only wrinkle that would affect packaging would be if this change resulted in a SONAME bump. That seems unlikely to be the case. |
| Comment by Kevin Albertson [ 02/Dec/19 ] |
|
roberto.sanchez from a packaging perspective, do you think disabling shared memory counters by default would be disruptive? |