Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [platform-releng-dev] Heads up on test failures ... a "real" unit test failure

I released a fix for this on Friday (thanks to Sergey Prigogin for the quick patch).

John



From:        David M Williams <david_williams@xxxxxxxxxx>
To:        "Eclipse platform release engineering list." <platform-releng-dev@xxxxxxxxxxx>,
Date:        08/22/2014 06:31 PM
Subject:        Re: [platform-releng-dev] Heads up on test failures ... a "real" unit test failure
Sent by:        platform-releng-dev-bounces@xxxxxxxxxxx




I know its late, I know its Friday, I know we just had two major check points ... but ... in case anyone is curious about status of unit tests ...

Windows has been all cleaned up and I re-ran the tests from last night's nightly, and the same test fails on it, and Linux, so I think it is "real" (
testProjectLocationValidation). Sure hope it's not because we no longer sign those inner jars :/ (bug 442119).

On the Mac, after a couple of days of chasing the wrong thing (bug
442322) I finally tracked down the real issue and those tests (from last night) are re-running now. (Lesson learned: use bash scripts, not "easy to use" tools :)





From:        
David M Williams/Raleigh/IBM@IBMUS
To:        
"Eclipse platform release engineering list." <platform-releng-dev@xxxxxxxxxxx>,
Date:        
08/22/2014 02:20 AM
Subject:        
Re: [platform-releng-dev] Heads up on test failures ... and on        Linux ...
Sent by:        
platform-releng-dev-bounces@xxxxxxxxxxx




I'd noticed Linux "ran fine" ... wasn't surprised ... and wasn't until just now I noticed there were only about 12,000 tests that ran in 17 minutes!

So, I'll point out another of my mistakes ... but this time bragging a little too :) ... as part of the refactoring, I've structured the scripts so that, from Hudson, it once again works to run just select suites ... not necessarily all ... and I had been experimenting with running only the ones we call "platform tests" and I had forgotten and left it 'set' that way, when I should have unset it, or set it to "all". [Long ways to go, still before all this is documented, consistent, in it's final form, etc., ... but that ability will be helpful in a number of use-cases (most notably, for me, is running subsets of "performance tests").

Thanks,







From:        
David M Williams/Raleigh/IBM@IBMUS
To:        
"Eclipse platform release engineering list." <platform-releng-dev@xxxxxxxxxxx>,
Date:        
08/22/2014 01:37 AM
Subject:        
Re: [platform-releng-dev] Heads up on test failures ...
Sent by:        
platform-releng-dev-bounces@xxxxxxxxxxx




Just to keep everyone informed, the problem with the windows tests was primarily due to having "java.io.tmp" incorrectly defined, so some file paths were ending up being hundreds and hundreds of characters long ... and we all know Windows doesn't like that.

AND THEN, to try and re-run the tests, I could not "wipe out" the Hudson workspace from the previous run. The tests from 0821 Nightly might mostly be ok, since each of them primarily run under their own "buildId space", but I'm sure there is some opportunity for "interference" from previous, failed run.

Bottom line, take any failures of what's currently running (for 0821 build) with a grain of salt, and we'll hope the IT Staff can get the machine cleaned up on Friday. (
Bug 442328).

Thanks, and apologies for the delays.





From:        
David M Williams/Raleigh/IBM@IBMUS
To:        
platform-releng-dev@xxxxxxxxxxx,
Date:        
08/21/2014 08:00 PM
Subject:        
[platform-releng-dev] Heads up on test failures ...
Sent by:        
platform-releng-dev-bounces@xxxxxxxxxxx




I don't know if anyone noticed, but last night we did not get any "Windows Tests" (with 0820 N-build) due to Hudson "going down". (bug
442229).

Rather than re-run them today, I have been focusing on getting the refactored "production test scripts" in place. And judging from the Windows results I got, apparently there are some path problems, or properties I'm not setting, or something. See Windows results at


http://download.eclipse.org/eclipse/downloads/drops4/N20140820-2000/testResults.php

So if the problem is obvious to anyone (such as on "what's required") let me know. Otherwise I'll continue looking for clues, and we'll see how a "top to bottom" run turns out.

Plus, second heads up, I can't get the Mac tests to run at all, at the moment ... (bug
442322) ... not sure if I can work around that one on my own, or not, or if I need to wait for webmasters to upgrade Ant. (the right way).

I haven't tried the Linux tests on eclipse.org, but that's where I tested the refactoring (on my Linux machines) so for now, will hope they are more "normal".

Just wanted you to know what I know ...

Thanks,

_______________________________________________
platform-releng-dev mailing list
platform-releng-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit

https://dev.eclipse.org/mailman/listinfo/platform-releng-dev _______________________________________________
platform-releng-dev mailing list
platform-releng-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit

https://dev.eclipse.org/mailman/listinfo/platform-releng-dev _______________________________________________
platform-releng-dev mailing list
platform-releng-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit

https://dev.eclipse.org/mailman/listinfo/platform-releng-dev _______________________________________________
platform-releng-dev mailing list
platform-releng-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/platform-releng-dev

Back to the top