This message was deleted.
# troubleshootings
Slackbot
10/17/2022, 6:22 PMThis message was deleted.
g
Gian Merlino
10/17/2022, 7:08 PMwhat do you mean by "segment merge wait time"?
r
Rohit
10/18/2022, 5:51 AMwe have data in S3. When we query in druid, segments gets loaded from there to disk and from disk to memory right, if i am not wrong? We want to reduce the time to fetch while querying the data in terms of that. we are using i3.4x nodes(30). wanted to check if there are few druid config properties that can be tweaked to achieve better performance.
k
Karan Kumar
10/18/2022, 5:55 AM@Rohit Segments are fetched from s3 when the historicals are coming up not at query time
Karan Kumar
10/18/2022, 5:55 AMSo are you trying to optimize on query time or starting the historical faster
r
Rohit
10/18/2022, 5:56 AMi want to optmize on query time
k
Karan Kumar
10/18/2022, 5:58 AMPlease share the query.
The segment shard spec
and the number of segments
Karan Kumar
10/18/2022, 5:59 AMWhat do you think is getting bottlenecked
r
Rohit
10/28/2022, 12:01 PM@Karan Kumar sorry for late response, as i was out for few days. single segment for an hour, 3 replicas, total segment for a datasource - ~10k . these are select queries with zeroFilledTopN or groupby
k
Karan Kumar
10/28/2022, 2:53 PMCan you please share the metrics underneath the graph ?
https://druid.apache.org/docs/latest/operations/metrics.html#historical
https://druid.apache.org/docs/latest/operations/metrics.html#historical-1
It would be present from the above list