This message was deleted.
# generals
Slackbot
10/06/2023, 2:44 PMThis message was deleted.
k
Kyle Hoondert
10/06/2023, 2:50 PMthe segment cache needs to have enough space to hold the segments. If you have 30GB of segments, you will need 30GB of segment cache for 1 replica, 60GB for 2 replicas (30GB on each of 2 servers)
Kyle Hoondert
10/06/2023, 2:50 PMall the active segments are copied locally
a
Adith Reddy
10/06/2023, 2:52 PMOkay, got it. Thanks! Would the inactive data be removed from cache regularly?
a
Anil Gupta
10/06/2023, 3:00 PMYou can automatically remove older data from the historicals via retention rules (https://druid.apache.org/docs/latest/operations/rule-configuration/#set-retention-rules)
Anil Gupta
10/06/2023, 3:01 PMHowever, if a segment is to be queryable using the native query engine that gives you fast query results, then any segment that is in a time frame that is a candidate for querying must stay in the segment cache.
a
Adith Reddy
10/06/2023, 3:03 PMGot it! Will explore the retention rules.
Adith Reddy
10/06/2023, 3:03 PMIn the case that there are two replicas of historical will the data be split between the replicas or is it another copy of the same data?
a
Anil Gupta
10/06/2023, 3:06 PMBy default, Druid stores 2 replicas of every segment file. (This can be increased or decreased to 1.). Segment replicas are never stored on the same historical node. The idea is that, if a single node goes down, you still have full availability of your data. In order to be even more resilient on the data tier, you can store more than 2 replicas.
3 Views