New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
mon$memory_usage: Incorrect database memory reported on CS and SC [CORE2478] #2891
Comments
Modified by: @dyemanovassignee: Dmitry Yemanov [ dimitr ] |
Modified by: @dyemanovstatus: Open [ 1 ] => In Progress [ 3 ] |
Commented by: @dyemanov The only way to report the more or less correct memory counters would be the following: 1) The record linked to MON$DATABASE always reports zeroes (it means that Classic has no shared cache) It somewhat breaks the idea behind aggregated counters (one might expect that a database pool counter would contain a sum of all its child pools), but this idea doesn't work in Classic anyway. Another alternative would be to report proper memory statistics for attachments but random statistics for a database (instead of zeroes). This is what can be done in v2.5. If you are not happy with that, a better solution will be deferred till v3.0. |
Commented by: Douglas Tosi (douglasht) I believe zeroed stats is less confusing than random. |
Commented by: @dyemanov Would superclassic behave this way too? Would total memory used by the server be the sum of the memory used by all attachments? |
Modified by: @dyemanovstatus: In Progress [ 3 ] => In Progress [ 3 ] Fix Version: 2.5 RC1 [ 10362 ] |
Modified by: @dyemanovstatus: In Progress [ 3 ] => Open [ 1 ] |
Modified by: @pcisarstatus: Resolved [ 5 ] => Closed [ 6 ] |
Modified by: @pavel-zotovQA Status: No test |
Submitted by: Douglas Tosi (douglasht)
When using Classic or SuperClassic:
- The reported attachment memory does not include the attachment cache. Because the cache was allocated specifically for that attachment, it should be reported as owned by the attachment.
- The database memory (stat_group=0) starts at 35MB and stays like that for the entire load-test. Which seems incorrect given that task manager shows peak of 200MB. (I'm not trying to directly compare mon$memory and task manager. It's just to show that memory usage *has* changed during the load-test.)
Commits: bcebd6f
The text was updated successfully, but these errors were encountered: