When running some external compactor tests, it was noticed that zk resource usage seemed to increase in relation to the number of idle compactors. This did not seem to occur if the compactors were busy.
This should be investigated to ensure we aren't creating additional ZK load as part of the job polling operations.
When running some external compactor tests, it was noticed that zk resource usage seemed to increase in relation to the number of idle compactors. This did not seem to occur if the compactors were busy.
This should be investigated to ensure we aren't creating additional ZK load as part of the job polling operations.
General Investigation Steps: