Open grafana and set HR to custom and increase number to atleast 15
Connect mongo to PMM
Open Mongo Instance Summary and Mongo Cluster Summary
See that browser (chrome) is stucked
Also pmm-agent stucked on QAN profiler
Problems and limitations: Output with more than 10K DBs on /metrics endpoint has around 9MB per everyscrape, so it will take a long time to parse it. Also now we have limit 10MB size in nginx inside pmm server. So limit can be reached very easily.
Are we going to support so much amount of DBs for QAN?
Actual Result: pmm-agent is stucked, grafana is stucked on mongo dashboards
Both. For me it was lagged exporter (opened in chrome and it was stucked for a few seconds then ok and then again stucked). In QAN I didnt have any data at all (because output has around 9MB and it was not possible to scrape it in default HR interval for me).
Roma Novikov October 1, 2021 at 3:34 PM
, is this QAN or Exporter problem?
I see a linked issue where we discussed the impact of the number of Collections / DB's for Exporter ..
Medium
Dashboards for mongo are stucked and they fails on timeout. QAN doesnt show data related to mongo.
For example requests http://127.0.0.1/graph/api/datasources/proxy/1/api/v1/series, http://127.0.0.1/graph/api/datasources/proxy/1/api/v1/query stucked in pending state.
Steps to reproduce:
Run mongo
Create more than 10K dbs or more, use this loop:
Open grafana and set HR to custom and increase number to atleast 15
Connect mongo to PMM
Open Mongo Instance Summary and Mongo Cluster Summary
See that browser (chrome) is stucked
Also pmm-agent stucked on QAN profiler
Problems and limitations: Output with more than 10K DBs on /metrics endpoint has around 9MB per everyscrape, so it will take a long time to parse it. Also now we have limit 10MB size in nginx inside pmm server. So limit can be reached very easily.
Are we going to support so much amount of DBs for QAN?
Actual Result:
pmm-agent is stucked, grafana is stucked on mongo dashboards
Expected Result:
Nothing is stucked and working.