Closed Bug 562023 Opened 10 years ago Closed 10 years ago

add full acceptance testing for MIPS to deep phase


(Tamarin Graveyard :: Build Config, defect)

Not set


(Not tracked)

Q3 11 - Serrano


(Reporter: brbaker, Assigned: brbaker)


(Whiteboard: buildbot has-patch)


(1 file, 2 obsolete files)

Currently MIPS testing is too slow to have in the phase 1 "test" section so we are currently only running smokes (to ensure that the shell is working). We need to add a second MIPS machine into the build system that we can run a full acceptance pass release and debug in the deep phase of the build system.

This will require 
1) configuration updates to the build system 
2) setting up a host machine to run the buildbot slave and to remotely access the MIPS box (testing on mips is done remotely from another machine, see bugs #524129, #559038)
3) modify/confirm entries in testconfig.txt for current "arm-*" skips/failures are the same for "mips-*"
Flags: flashplayer-triage+
Flags: flashplayer-qrb?
Summary: [buildbot] add full acceptance testing for MIPS to deep phase → add full acceptance testing for MIPS to deep phase
Whiteboard: buildbot
full acceptance pass said to take ~ 1h40+ minutes to run
Ideally we could cross-compile python so that a host machine isn't necessary, but I've been unsuccessful thus far.

This is the most promising info I've found:

I'm unable to successfully compile, mostly due to missing dependencies which I've not been able to fully resolve.  I've tried ubuntu 9.10 and 10.4.  I'll try with openSUSE next.
Assignee: nobody → brbaker
Flags: flashplayer-qrb? → flashplayer-qrb+
Target Milestone: --- → flash10.2
Attached patch add mips-deep (obsolete) — Splinter Review
- adds a generic way to run acceptance via the all/ script
- runs acceptance release and debug currently, may add before this is pushed
- TODO: still need to update testconfig.txt to mark/match known failures and to skip tests that will not run properly (will start with arm-* as the base)
Attachment #453404 - Flags: review?(jsudduth)
Comment on attachment 453404 [details] [diff] [review]
add mips-deep

- one thing that I wasn't sure on and would like feedback is if the run-acceptance-generic-ssh script should be hardcoded to --threads=1 in the runtests call, or should that just be in the buildbot config. Currently for the way that it is working it MUST be single threaded otherwise the client that is being ssh'd into will have necessary files deleted by one thread while the other is trying to run
Attachment #453404 - Flags: review?(cpeyer)
Whiteboard: buildbot → buildbot has-patch
Attachment #453404 - Flags: review?(jsudduth) → review+
Comment on attachment 453404 [details] [diff] [review]
add mips-deep

+ Pending an answer to the question about --threads=1.
Attached patch v2 add mips deep (obsolete) — Splinter Review
- move the --threads=1 into the buildbot config, no sense hard coding in the generic script since future uses of the script may handle multiple threads
- fix an issue in the script where it would deploy the wrong shell, was deploying the $shell_release instead of what was passed to the shell and downloaded
Attachment #453796 - Flags: review?(cpeyer)
Attachment #453404 - Attachment is obsolete: true
Attachment #453404 - Flags: review?(cpeyer)
Attached patch v3 add mips deepSplinter Review
version that will be pushed

- make sure to remove EXITCODE=# from the stdout before returning it to, this makes sure that tests that have .out file matching work correctly

- make sure that the script writes to a local file instead of a common /tmp/stdout, the previous way obviously causes problems if you try and run the script from multiple slaves on one machine
Attachment #453796 - Attachment is obsolete: true
Attachment #453796 - Flags: review?(cpeyer)
Comment on attachment 454029 [details] [diff] [review]
v3 add mips deep

patch pushed as 4881:04811339be3b
Closed: 10 years ago
Resolution: --- → FIXED
You need to log in before you can comment on or make changes to this bug.