The rate of successful ingestion is too low. Look for errors in the logs of the perf-ingest process.
Android Clustering Rate is too low. Look to see if PubSub events are being sent: http://go/android-perf-ingest-stall
Also confirm that files are being sent with actual data in them (sometimes they can be corrupted with a bad config on the sending side). Look in:
gs://skia-perf/android-master-ingest/tx_log/
Perf Clustering Rate is too low. Look to see if PubSub events are being sent:
Also confirm that files are being sent with actual data in them (sometimes they can be corrupted with a bad config on the sending side). Look in:
gs://skia-perf/nano-json-v1/
The perf instance has not detected any regressions in an hour, which is unlikely because of the large amount of traces these instances ingest.
Check that data is arriving to the instances that do event driven regression:
Check that PubSub messages are being processed: http://go/android-perf-ingest-stall
And determine when regression detection stopped: