2da390414e
With this change pre hourly processor which does the hourly aggregation (second stage) and writes the final aggregated metrics to metris topic in kafka now accounts for any early arriving metrics. This change along with two previous changes to pre hourly processor that added 1.) configurable late metrics slack time (https://review.openstack.org/#/c/394497/), and 2.) batch filtering (https://review.openstack.org/#/c/363100/) will make sure all late arriving or early arriving metrics for an hour are aggregated appropriately. Also made improvement in MySQL offset to call delete excess revisions only once. Change-Id: I919cddf343821fe52ad6a1d4170362311f84c0e4 |
||
|---|---|---|
| .. | ||
| config | ||
| cpu_kafka_data | ||
| cpu_kafka_data_second_stage | ||
| fetch_quantity_data | ||
| fetch_quantity_data_second_stage | ||
| fetch_quantity_util_second_stage | ||
| kafka_data | ||
| kafka_data_second_stage | ||
| mem_total_all | ||
| metrics_pre_hourly_data | ||
| __init__.py | ||
| mock_component_manager.py | ||
| mock_data_driven_specs_repo.py | ||
| test_read_offsets_on_start.json | ||