[zorg] r300573 - Remove the ppc64le-clang-lnt buildbot because it continues to be flaky making it hard to listen to/use buildbots effectively due to the noise.

[Matthias Braun via llvm-commits]

> On Apr 19, 2017, at 7:31 AM, Davide Italiano via llvm-commits <llvm-commits at lists.llvm.org> wrote:
> 
> On Wed, Apr 19, 2017 at 7:23 AM, David Blaikie <dblaikie at gmail.com <mailto:dblaikie at gmail.com>> wrote:
>> On Wed, Apr 19, 2017 at 6:24 AM Bill Seurer <seurer at linux.vnet.ibm.com>
>> wrote:
>>> 
>>> What test or tests were being flaky?  Was it by chance in thread
>>> sanitizer tests?
>> 
>> 
>> I believe it might've been, yes - though I don't remember off-hand.
>> 
>>> 
>>> I had previously spent a good chunk of time tracing
>>> back to the source of other thread-related problems and can adapt the
>>> scripts I used then to investigate whatever you were seeing.  I suspect
>>> this might be the same problem as I don't think the previous issues were
>>> ever fixed.
>> 
>> 
>> Likely was never fixed, by the sounds of it.
>> 
>>> 
>>> Which IRC channel?
>> 
>> 
>> #llvm on irc.oftc.net ( http://llvm.org/docs/#irc )
>> 
>>> 
>>>  And can you send me a summary of the IRC discussion?
>> 
>> 
>> I've since rebooted my machine & the channel isn't logged.
>> 
>> Though perhaps some of the others in the conversation recall which test
>> case. (+eugenis, +dannyb, +matzeb, +d0k, +davide - I /think/ they were all
>> involved in the conversation)
>> 
> 
> Yes, I contacted Galina in the past about this (and other flakey bots
> over the years).
> Sorry, I don't think what David did is unreasonable. I understand
> losing the history is inconvenient, but it's also not ideal sending
> mails with spurious failures. The bot failed fairly frequently for me,
> to a point where when I've seen `ppc64-lnt` as subject in my mail I
> didn't even bother clicking on the link contained in the mail.
> To the best of my knowledge this is the only bot in this state, FWIW.

I remember starting 1 or 2 threads about sanitizer tests failing as well. The effect is that the bot is sending out hundreds of failure messages that the actual commiters did not cause and can do nothing about. I agree with both David(e)s that this decreases the value of the bot testing because it decreases the communities trust into the reporter problems. It's also one of those cases where you unnecessarily take up a small amount of time out of every persons days but you are doing that to hundreds of people...

If there is no way to get the tests fixed, at least find ways to disable them or not send failure mails about the known flaky tests. What good are those tests or processes anyway if they can break for months and not get fixed...

- Matthias
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20170419/ce35c9a8/attachment.html>