coordinator DELETE segments moved from hdfs had delay

Hi all,

I have a problem with coordinator DELETE.

19:46:15 I run the command as follows.

curl -v -X “DELETE” “…/druid/coordinator/v1/datasource/m2_001?kill=true&interval=2016-11-01T17:00:00Z/2016-11-01T18:00:00Z”

19:46:23 The kill-task SUCCESS, and all segments moved from HDFS. I got from the task log.

19:54:53 The historical node began to moved the segments. I got “new request [Drop …]” from the historical log.

I want to know the detail follow about how to run a Kill task for a given interval and datasource. Especially the time when segments moved from HDFS, and the time moved from historical node. And is it configurable?



Hi Xinxin,

KillTask removes the segments from HDFS and removes the corresponding entries from metadata storage.

The coordinator periodically polls the metadata storage for any additions/deletion of the segments and assigns/drop segments from historical nodes.

you can configure druid.manager.segments.pollDuration and druid.coordinator.period to configure the poll durations.
above configs are documented here -

Hi Nishant,

My coordinator uses default configuration except the follows:



And I want to know the exact interval between the segment deleted from HDFS and deleted from Historical node.

Thanks a lot,


在 2016年11月15日星期二 UTC+8下午6:13:05,Nishant Bangarwa写道:

THe kill task only works for segments that are disabled. You need to disable these segments through rules first in the coordinator console.