EngineeringProductivity/Projects/Debugging UX improvements

From MozillaWiki
Jump to: navigation, search

This project is composed of multiple projects which will be improving the overall experience when debugging issues on try.

Q4's 2016 focus is mainly improving end to end times on Try (Thunder Try project).

Context

The original discussions are discussed in here. The projects defined below are the ones we're considering tackling at this point.

P1 means that we're commiting to complete a bug within the current quarter. P2 means that it's nice to fix and we will tackle if we have time.

All possible projects - bugzilla tree dependency

Meeting

We are trying to meet every four weeks. We will be posting details in the tools mailing list. Etherpad: https://public.etherpad-mozilla.org/p/debugging_ux_improvements

Dial-in information

 # This meeting will start at 10:00-10:50AM PDT 
 # Guest:  https://v.mozilla.com/flex.html?roomdirect.html&key=43kzfyP8BDSG
 # Vidyo:  A-Team Vidyo Room
 # Phone:  650-903-0800 or 650-215-1282 x92 Conf# 98416  (US/INTL)
 #         1-800-707-2533 (pin 369) Conf# 98416 (free for anyone to join)
 # IRC:    irc://irc.mozilla.org:6697/#ateam

Project #1 - Thunder Try - Improve end to end times

Current main focus:

  • Artifact builds on try
  • Automation metrics
  • Better sscache for try
  • Hyper chunking
  • Web platform tests from source checkout

Docs:

Tracked bugs (P1s are the ones we commit to for this quarter)

ID Priority Summary Whiteboard Assigned to
1286336 -- [meta] Improve interaction of Firefox automation with hg.mozilla.org Gregory Szorc [:gps]
1286900 -- Run wpt tests from source checkout Gregory Szorc [:gps]
1280633 -- Step timings (Mozharness) to determine where to optimize our end to end times [metrics]
1301567 -- Display end-to-end times of jobs in Treeherder
1315030 -- Determine how many try pushes could benefit from using artifact builds
1286856 -- Create a dashboard for tracking and analyzing try end-to-end times Kyle Lahnakoski [:ekyle]
1291445 -- Record the time of ingestion for completed jobs
1294706 -- "Mochitest other" chunk on win7 debug slow and often last to complete
1305768 -- Periodic spikes in response time for requests from Treeherder for task logs through cloud-mirror
1291940 P1 TaskCluster slows down considerably when using AUFS Greg Arndt [:garndt]
1284882 P1 [tracker] Artifact builds on automation Chris Manchester (:chmanchester)
1280641 P1 Implement a two-tier S3 cache for sccache [build-run-speed-up] Ted Mielczarek [:ted.mielczarek]
1240644 P2 Allow skipping build and reusing existing build result for try push

13 Total; 13 Open (100%); 0 Resolved (0%); 0 Verified (0%);


Maintenance

Debugging tests on interactive workers

Developers can go from requesting a Linux loaner on TaskCluster to having a debugging environment ready to run tests in a reasonable time with minimal interaction with external documentation or need a minimum amount of help.

Tracked bugs (P1s are the ones we commit to for this quarter)

ID Priority Summary Whiteboard Assigned to
1302520 -- Support running jittest from an interactive worker
1301382 -- Cloning from interactive loaners should pin hg.mozilla.org fingerprint
1302519 -- Support running gtest from an interactive worker
1303031 -- Support running Android xpcshell tests on interactive loaners Andrew Halberstadt [:ahal] [PTO until late May]
1293259 -- Ability to resolve tests outside of a build context (i.e srcdir or test package)
1302518 -- Support running cppunittests from an interactive worker
1262605 P1 [Tracking] Make test harness mach commands available on the $PATH for interactive jobs Andrew Halberstadt [:ahal] [PTO until late May]
1278540 P2 Increase the time window to grab an interactive loaner
1290126 P2 Task Creator should re-direct to /one-click-loaner/connect if interactive=true
1271677 P2 Interactive tasks should be scheduled with high priority
1278313 P2 Interactive shell needs better scroll functionality
1284998 P2 Create tests for interactive debugging workflows
1273158 P2 Can't run desktop_unittest.py with --no-run-tests followed by --run-tests afterwards

13 Total; 13 Open (100%); 0 Resolved (0%); 0 Verified (0%);