jonasplaum opened a new issue, #5446: URL: https://github.com/apache/couchdb/issues/5446
## Description The quantiles are in Seconds and the sum is in milliseconds, but both are supposed to use the same unit. Analysis by rnewson in Slack: https://couchdb.slack.com/archives/C49LEE7NW/p1740575676084599?thread_ts=1740575450.920029&cid=C49LEE7NW "looks like a bug to me; ``` T0 = os:timestamp(), {ok, Info} = fabric:get_db_info(Db), Suffix = mem3:shard_suffix(Db), Etag = chttpd:make_etag({Info, Suffix}), DeltaT = timer:now_diff(os:timestamp(), T0) / 1000, couch_stats:update_histogram([couchdb, dbinfo], DeltaT), ``` os:timestamp is microseconds, so dividing by 1000 gives milliseconds, but the prometheuse metric says "seconds" and then here ``` to_prom_summary(Path, Info) -> Used 3 times Metric = path_to_name(Path ++ ["seconds"]), {value, Value} = lists:keyfind(value, 1, Info), {arithmetic_mean, Mean} = lists:keyfind(arithmetic_mean, 1, Value), {percentile, Percentiles} = lists:keyfind(percentile, 1, Value), {n, Count} = lists:keyfind(n, 1, Value), Quantiles = lists:map( fun({Perc, Val0}) -> % Prometheus uses seconds, so we need to convert milliseconds to seconds Val = Val0 / 1000, ``` we convert to seconds for the quantiles but not the mean average." ## Steps to Reproduce Look at a metric that has a sum, like couchdb_dbinfo_seconds_sum. The average is 1000 times higher than expected. ## Your Environment * CouchDB version used: 3.4.2 -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: notifications-unsubscr...@couchdb.apache.org.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
