Closed Bug 704859 Opened 14 years ago Closed 14 years ago

Refactoring of automation scripts revealed underlying issues in testrun.py

Categories

(Mozilla QA Graveyard :: Mozmill Automation, defect)

defect
Not set
normal

Tracking

(Not tracked)

RESOLVED FIXED

People

(Reporter: whimboo, Assigned: whimboo)

References

Details

Attachments

(2 files)

With the change on bug 693306 we now no longer execute the individual testruns via subprocess.call but directly create an instance of the testrun class. Looks like we carry over settings from previous testruns. 1. If you are running the daily tests for two builds, the second build will get the update channel from the first build. That's why our Aurora builds on qa-horus have been updated to Nightly builds: http://mozmill-release.brasstacks.mozilla.com/#/update/report/d41f7e0eb42da3cd5270e990a21ad631 2. Also it looks like we cache results and send a combined report of multiple testruns in a single report. That behavior only happened with yesterdays daily testruns across platforms but not with the todays ones: http://mozmill-release.brasstacks.mozilla.com/#/endurance/report/d41f7e0eb42da3cd5270e990a2494e08 I will have to investigate that further and check the remaining results from todays testrun later.
Yeah, somehow we are setting the 'nightly' update channel for successive builds. In this case 5.0.1 should still be on the release channel. * 11.0a1 => 11.0a1, minor, en-US, complete, nightly, 2011-11-23, '''FAIL''' * 5.0.1 => 5.0.1, minor, en-US, complete, nightly, 2011-11-23, '''FAIL'''
I found the cause for the update issue. It's simply that we modify self.options.channel and set it to a new value for the first build. Successive builds in this testrun will get the same channel assigned to.
If anyone else comes first and want to review, I would kinda appreciate it. This has to be checked-in ASAP.
Attachment #576546 - Flags: review?(jhammel)
Attachment #576546 - Flags: review?(jhammel) → review?(dburns)
Comment on attachment 576546 [details] [diff] [review] Patch v1 (update channel) [checked-in] looks good to me
Attachment #576546 - Flags: review?(dburns) → review+
Comment on attachment 576546 [details] [diff] [review] Patch v1 (update channel) [checked-in] Landed as: http://hg.mozilla.org/qa/mozmill-automation/rev/3f1221d87fa6 I will now investigate the issue we have with Endurance tests, after I have updated all boxes with the correct builds again. :S I kinda would like to have the CI system.
Attachment #576546 - Attachment description: Patch v1 (update channel) → Patch v1 (update channel) [checked-in]
This fixes the endurance testrun and does not blow up the result array for each additional binary we are testing.
Attachment #576782 - Flags: review?(dave.hunt)
Attachment #576782 - Flags: review?(dave.hunt) → review+
Attachment #576782 - Attachment description: Patch v1 (Endurance tests) → Patch v1 (Endurance tests) [checked-in]
Ok, looks like that any regression has been fixed now. If not feel free to reopen. For now I will close this bug.
Status: ASSIGNED → RESOLVED
Closed: 14 years ago
Resolution: --- → FIXED
Product: Mozilla QA → Mozilla QA Graveyard
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: