Swarming: use one TaskToRun per task enqueueing |
||
Issue descriptionTaskToRun are reused for task retries on internal_failure. That's an error in hindsight, as the reuse means we are loosing information about when the task was re-enqueued. As part of issue 781021 , using separate TaskToRun per TaskSlice will make the whole thing much more manageable and auditable as task triggering information won't be lost anymore.
,
Mar 24 2018
The following revision refers to this bug: https://chromium.googlesource.com/infra/luci/luci-py.git/+/c5db5135a63f3db4452308b1535ec47f189f668b commit c5db5135a63f3db4452308b1535ec47f189f668b Author: Marc-Antoine Ruel <maruel@chromium.org> Date: Sat Mar 24 03:54:44 2018 Swarming: one TaskToRun per attempt. - Change TaskToRun key id to the try number. It will contain more information once issue 781021 is implemented. - Do not store TaskToRun for deduped task. - Do not reuse TaskToRun for try #2. - Add TaskToRun.created_ts to know when a task was enqueued. It will be needed for issue 781021 . - Rename a few 'task' local variable to 'to_run' make it things clearer. This change temporarily breaks task cancelation and dead bot handling for the tasks that are on-going. Bug: 817831 , 781021 Change-Id: If252001ba8c8bcb499134b1d21ab2f10cd06203b Reviewed-on: https://chromium-review.googlesource.com/969760 Commit-Queue: Marc-Antoine Ruel <maruel@chromium.org> Reviewed-by: Vadim Shtayura <vadimsh@chromium.org> [modify] https://crrev.com/c5db5135a63f3db4452308b1535ec47f189f668b/appengine/swarming/doc/Schemas.md [modify] https://crrev.com/c5db5135a63f3db4452308b1535ec47f189f668b/appengine/swarming/server/task_result_test.py [modify] https://crrev.com/c5db5135a63f3db4452308b1535ec47f189f668b/appengine/swarming/server/task_scheduler.py [modify] https://crrev.com/c5db5135a63f3db4452308b1535ec47f189f668b/appengine/swarming/server/task_scheduler_test.py [modify] https://crrev.com/c5db5135a63f3db4452308b1535ec47f189f668b/appengine/swarming/server/task_to_run.py [modify] https://crrev.com/c5db5135a63f3db4452308b1535ec47f189f668b/appengine/swarming/server/task_to_run_test.py
,
Mar 24 2018
Deployed to prod.
,
May 3 2018
The following revision refers to this bug: https://chromium.googlesource.com/infra/luci/luci-py.git/+/7c1f40f87fec05a05bb8285cceb40d52ecea304a commit 7c1f40f87fec05a05bb8285cceb40d52ecea304a Author: Marc-Antoine Ruel <maruel@chromium.org> Date: Thu May 03 21:01:18 2018 [swarming] remove negative cache in dead bot handling This is not necessary anymore since https://crbug.com/817831 was fixed. It was removing the entry for try_number 1 but the new TaskToRun entity for try_number 2 has a different negative cache entry. Bug: 817831 Change-Id: I7c8e06776b8a1f316d14ea2a36b5aba561d8acac Reviewed-on: https://chromium-review.googlesource.com/1042886 Commit-Queue: Marc-Antoine Ruel <maruel@chromium.org> Reviewed-by: Quinten Yearsley <qyearsley@chromium.org> [modify] https://crrev.com/7c1f40f87fec05a05bb8285cceb40d52ecea304a/appengine/swarming/server/task_scheduler.py [modify] https://crrev.com/7c1f40f87fec05a05bb8285cceb40d52ecea304a/appengine/swarming/server/task_scheduler_test.py |
||
►
Sign in to add a comment |
||
Comment 1 by bugdroid1@chromium.org
, Mar 22 2018