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
OpenTSDB maintains internal statistics on various parameters which are retrievable with /api/stats REST call. Querying stats via REST is useful for monitoring in some scenarios, but it is not helpful for post-factum problem analysis for sporadic, hard to reproduce performance issues. It seems to be useful to provide a feature that would optionally enable periodic logging of the same statistics into log file. It could be common OpenTSDB log file or dedicated log file for stats. Logging interval could also be configurable, e.g. default of 1 min could be reasonable.
The text was updated successfully, but these errors were encountered:
Typically users will use a tool to periodically scrape the stats back into TSDB or another monitoring system. E.g. The Tcollector code can scrape and store multiple tsds.
In 3.0 we have a publishing mechanism like Dropwizard that can emit to TSDB or Influx and other systems in the future.
But I'd be ok with a PR for 2.x that had a feature flag folks could turn on to log the stats periodically.
OpenTSDB maintains internal statistics on various parameters which are retrievable with /api/stats REST call. Querying stats via REST is useful for monitoring in some scenarios, but it is not helpful for post-factum problem analysis for sporadic, hard to reproduce performance issues. It seems to be useful to provide a feature that would optionally enable periodic logging of the same statistics into log file. It could be common OpenTSDB log file or dedicated log file for stats. Logging interval could also be configurable, e.g. default of 1 min could be reasonable.
The text was updated successfully, but these errors were encountered: