2 new buckets are created, one for each event type (as the timings will differs). As of today queue events are likely to take less time than system events hence the more "aggressive" buckets. Another major difference between the 2 buckets is that system_events gather the processing time (time elapsed between the beginning and the end of processing) while the queue bucket gather the whole lead time (from enqueue to end of processing). The diffrence is due to the "wait randomization" introduced in system_events that delay the start of processing to avoid contentions on clock time due to old cron. In this situation, measuring the lead time on system event would actually mean measuring the randomization. It's useless because people already knows that the random number is 4. Part of request #13891 Gather more metrics about Tuleap backend Change-Id: I6b169d7f742a2adb3e6190ce04d50791e65477de

+324 −197

Author

Manuel Vacelet (vaceletm) 2019-09-19 17:49

Committer

Manuel Vacelet (vaceletm) 2019-09-26 17:33

Hash f3a7ac38d70846a23e092d88c4915cc40a74862c

Parent 5f4dde7b9d8382c8c02c59e44b8c862c33122809

Reference git #tuleap/stable/f3a7ac38d70846a23e092d88c4915cc40a74862c