New issue
Advanced search Search tips

Issue 670608 link

Starred by 2 users

Issue metadata

Status: Fixed
Owner:
Closed: Dec 2017
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 1
Type: Feature

Blocking:
issue 620331
issue v8:5680



Sign in to add a comment

Split V8's gatekeeper_ng into a tree-closing and a non-tree-closing one

Project Member Reported by machenb...@chromium.org, Dec 2 2016

Issue description

It is not possible with gatekeeper_ng to mix notifications that close and others that don't close the tree for a builder on the same tree.

Originally there was a sections-feature in gatekeeper to deal with this, but it was removed.

This proposes to overcome this problem by running independent gatekeeper instances with independent configs. In particular we'll split off V8's gatekeeper from the others.

We'll then split V8's gatekeeper config into several:
- One for all tree closers
- One for each type of breakage that should cause an email only to the blamelist when the breakage occurs for the first time.

A possible negative impact might be a higher request load on the v8 masters as:
- before, there was a natural but undesired delay when v8 gatekeeper waited for all others like chromium.fyi to be processed,
- several gatekeeper instances for the same master poll more frequently as they scale up

To mitigate possible overloading, we could make the polling frequency a bit lower for those instances.
 
Some thoughts that make things easier:
- We don't need to split the bot, it'd be enough to have dedicated v8 json files for tree-closers and non-tree-closers.
- We need to make those overrideable in the gatekeeper_trees.json file.
- We'd then have the different v8 notifications as different steps on:
https://build.chromium.org/p/chromium.gatekeeper/builders/Chromium%20Gatekeeper/builds/732261

Optionally we could still move those to a different bot (and also splitting the gatekeeper_trees.json) to not be delayed by the slow chromium trees, but that's orthogonal.
Project Member

Comment 2 by bugdroid1@chromium.org, Dec 2 2016

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/tools/build.git/+/6c2a0017a5f8279ae7c6f64204b1fc92b10822ef

commit 6c2a0017a5f8279ae7c6f64204b1fc92b10822ef
Author: Michael Achenbach <machenbach@chromium.org>
Date: Fri Dec 02 13:07:50 2016

Split off V8 from gatekeeper.json

Also adds a test case using production data for the trees file for better coverage.

BUG= 670608 

Change-Id: I5328dec1a7cb168ae0245991378a9a9c8a21b2e0
Reviewed-on: https://chromium-review.googlesource.com/415733
Commit-Queue: Michael Achenbach <machenbach@chromium.org>
Reviewed-by: Andrii Shyshkalov <tandrii@chromium.org>

[modify] https://crrev.com/6c2a0017a5f8279ae7c6f64204b1fc92b10822ef/scripts/slave/gatekeeper.json
[modify] https://crrev.com/6c2a0017a5f8279ae7c6f64204b1fc92b10822ef/scripts/slave/gatekeeper_ng_config.py
[modify] https://crrev.com/6c2a0017a5f8279ae7c6f64204b1fc92b10822ef/scripts/slave/gatekeeper_trees.json
[modify] https://crrev.com/6c2a0017a5f8279ae7c6f64204b1fc92b10822ef/scripts/slave/recipe_modules/gatekeeper/api.py
[modify] https://crrev.com/6c2a0017a5f8279ae7c6f64204b1fc92b10822ef/scripts/slave/recipe_modules/gatekeeper/test_api.py
[add] https://crrev.com/6c2a0017a5f8279ae7c6f64204b1fc92b10822ef/scripts/slave/recipes/gatekeeper.expected/production_data.json
[modify] https://crrev.com/6c2a0017a5f8279ae7c6f64204b1fc92b10822ef/scripts/slave/recipes/gatekeeper.py
[add] https://crrev.com/6c2a0017a5f8279ae7c6f64204b1fc92b10822ef/scripts/slave/v8/OWNERS
[add] https://crrev.com/6c2a0017a5f8279ae7c6f64204b1fc92b10822ef/scripts/slave/v8/gatekeeper_v8_tree_closers.json

Blocking: 620331
Ping - please provide an update to your high priority bug. This bug is stale. Is it really P-1?

Comment 7 by stip@chromium.org, Feb 10 2017

Cc: -stip@chromium.org
Status: Fixed (was: Assigned)
Long done.

Sign in to add a comment