You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Add support for cataloging distinct aggregation pipelines 'shapes' and aggregating count/min/max/avg runtime, just as for find() and update() operations.
The text was updated successfully, but these errors were encountered:
This would be great! I found while digging through some recent logs that our aggregate queries were the highest total time and had great variability. Having visibility via mloginfo would be fantastic.
Aggregation commands log a pipeline rather than a query pattern, so it isn't obvious how to create concise output that would be analogous to find or update queries with specific critieria. Pipelines can also be rewritten as a result of server Pipeline Optimization.
However, mloginfo --queries does output stats for aggregate queries (count, min, max, etc) without trying to infer a pattern.
To investigate aggregation issues I'd suggest:
use mloginfo --queries to identify potentially problematic aggregate commands
use mlogfilter to extract log lines of interest (by namespace, duration, command, ...)
use mplotqueries to explore the outliers in your filtered output
After reviewing available log information I'm going to close this as not viable for now, but please feel free to comment if you have suggestions on how to extract a meaningful summary.
Add support for cataloging distinct aggregation pipelines 'shapes' and aggregating count/min/max/avg runtime, just as for find() and update() operations.
The text was updated successfully, but these errors were encountered: