relativitydev/relativity-trace-documentation

Document cold start scenario

krzysztof-kukla1974 opened this issue · 1 comments

krzysztof-kukla1974 commented

We should have documented special considerations related to Cold Start scenario.

  1. How to tweak Collect Data Source parameters?
  2. How to avoid throttling issues?
  3. How to monitor the system and adjust?
  4. How to use "merge data batches during cold start" parameter?
  5. How to configure Trace -> Setup -> Data Retrieval -> Run Interval - to not produce too many Data Batches on start and to not kill the 10 collection jobs in parallel queue.

Example:

  1. Stop Teams connector.
  2. Configure Trace -> Setup -> Data Retrieval -> Run Interval to 1 hour - so, only 1 Data Batch (each type) will be created per 1 hour.
  3. Abandon all Data Batches which were previously run (for instance with incorrect setup) - that would cancel all underlying Collect jobs.
  4. Reconfigure Data Source.
  5. Set Start Date to XYX.
  6. Set Frequency in Minutes to 1440.
  7. Set Merge Batches During Cold Start to Yes.
  8. Set Max Number of Batches To Merge to 30 <<< this means we'll create Data Batches each 30 day long.
  9. Reset Data Source.
  10. Enable Data Source.
krzysztof-kukla1974 commented

@peterhallerREL Maciej Kleczar would have good insight to this.