455755 - [tracking bug] automate running talos on mobile linux builds

Status: VERIFIED FIXED

(Release Engineering :: General, defect, P2)

Description

[alice nodelman [:alice] [:anode]]

Tracking bug for talos code improvements for mobile testing.

Comment 1

[John O'Duinn [:joduinn] (please use "needinfo?" flag)]

Per meeting with dougt, alice, removing some blockers that are nice-to-have, but not actually blocking.

Comment 3

[Aki Sasaki (not active)]

Update:

 * I'm changing the branch from fennec to mobile, since talos looks for the browser process name and the 'fennec' branch for the 'fennec' process was causing issues.  If I should avoid the 'mobile' name as well let me know.

 * The date was off on n810-talos05; now fixed.

 * The browser_wait seems to need to be different for each test, and if I set it to too long it seems to break the test, as does too short a setting.  This seems to be an underlying issue but for now I'm investigating what the best browser_wait and timeout is for each test.  If anyone has any ideas what numbers work best, please let me know.

 * Adding the "Unknown platform" bug to the dependency list.

 * Three of the n810s are running tp3 at one cycle, the rest at 3 cycles, which makes for two separate groupings of results on graphs-stage.  I'll probably set them all at 3 once I can reliably run tp3 without either browser_wait'ing too long/short and not crash mid-stream.

I'll be out of the office next week.  John knows how to reboot them if they hang and he will check them daily.

Comment 4

[Aki Sasaki (not active)]

re: browser_wait times: still custom per test on n810-talos{03,04,05}.  However, they look a lot healthier now that I've reduced the display brightness, which is helping with heat.

Comment 5

[John O'Duinn [:joduinn] (please use "needinfo?" flag)]

All talos suites (except Tp) have been running fine on graph-stage.m.o for a while now. We'll start posting results for all suites except Tp to production graphs.m.o as soon as bug#469721 is fixed in production graphs.m.o (later today/monday).

The crash in Tp is being tracked in bug#471585.

Comment 6

[Aki Sasaki (not active)]

This has been running long enough that I'm going to call it fixed.  The JIT crash/hang and the config merge are tracked in separate bugs.

Comment 7

[Joel Maher ( :jmaher ) (UTC -8)]

verified with beta3