Segments 0.0% available


I deploy apache druid (imply) to gke kubernetes. First time it worked, I add two sources (bucket), and it was ok but when I begin to stream a new data with kafka (one data every second), then I have multiples errors:

imply-data (data-tier…) → Pods have warnings

[GC pause (G1 Evacuation Pause) (young) (initial-mark) 255M->255M(256M), 0.0028872 secs]

Jul 21 16:43:34 INFO [main] org.apache.druid.server.coordination.BatchDataSegmentAnnouncer - Announcing segment XXXX_2021-06-21T21:00:00.000Z_2021-06-21T22:00:00.000Z_2021-07-21T10:22:45.490Z] at path[/imply/druid_ec5aXXX/segments/IP/IP_historical__default_tier_2021-07-21T16:43:…]

And I have always the error “waiting for all segments to become available druid” in cluster manager (imply)

Actions that I make:
Increase memory and cpu.

It is OK. I checked ressources. I had to increase cpu and memory.