Hi Team, I want to know if the query service has some flag , something like a

latencyOffset

to ensure clickhouse does not return recently written samples. This is to ensure that we do not receive inconsistent results due to the fact that only a part of the values are scraped in the last scrape interval.

@Srikanth Chekuri Any ideas on this?

It does by default exclude up to one minute

1. Is it configurable? The reason being this will inhibit applications from aggressive scaling in a very short intervals when scaling using external metrics. 2. Is it applicable for traces as well ? When running Clickhouse in cluster mode , for ex: 3 shards & 2 replicas, I have observed clicking on traces gives a 404 which might be due to replication delay as ClickHouse uses asynchronous replication by default. // @Srikanth Chekuri