It seems like the problem was the high concurrency of the checks. I've
reduced the concurrency and got it to pass:
https://github.com/apache/beam/pull/7989

Thanks everyone for your thoughts : )
Best
-P.

On Mon, Mar 4, 2019 at 3:56 PM Thomas Weise <[email protected]> wrote:

> Perhaps exclude the JIRA links from link checking?
>
>
> On Mon, Mar 4, 2019 at 9:01 AM Pablo Estrada <[email protected]> wrote:
>
>> Fair enough. Thanks for your feedback. I'll try to look into that.
>> Best
>> -P.
>>
>> On Mon, Mar 4, 2019 at 2:30 AM Maximilian Michels <[email protected]> wrote:
>>
>>> JIRA is quite flaky and drops connections occasionally. Since we check
>>> hundreds of links in a row, the failures are not surprising.
>>>
>>> Let's check if we can introduce more retries. I'll try to talk with
>>> INFRA to see if they can improve JIRA performance.
>>>
>>> Thanks,
>>> Max
>>>
>>> On 02.03.19 00:54, Ruoyun Huang wrote:
>>> > The log says running on 880 external links, but only showing 100ish
>>> (and
>>> > the number varies across different run) failure messages.  Likely it
>>> is
>>> > just flakey due to connection not stable on JIRA's site?
>>> >
>>> > Not sure how our tests are organized, but maybe re-try http request
>>> helps?
>>> >
>>> > On Fri, Mar 1, 2019 at 12:46 PM Pablo Estrada <[email protected]
>>> > <mailto:[email protected]>> wrote:
>>> >
>>> >     Hello all,
>>> >     the website tests are broken. I've filed BEAM-6760
>>> >     <https://issues.apache.org/jira/browse/BEAM-6760> to track fixing
>>> >     them, but I wanted to see if anyone has any idea about why it may
>>> be
>>> >     failing.
>>> >
>>> >     It's been broken for a few days:
>>> >     https://builds.apache.org/job/beam_PreCommit_Website_Cron/
>>> >
>>> >     And looking at the failures, it seems that they represent broken
>>> links:
>>> >
>>> https://builds.apache.org/job/beam_PreCommit_Website_Cron/725/console
>>> >
>>> >     But looking at each of the links opens their website without
>>> problems.
>>> >
>>> >     It may be some environmental temporary issue, but why would it fail
>>> >     consistently for the last few days then?
>>> >     Thoughts?
>>> >     Thanks
>>> >     -P.
>>> >
>>> >
>>> >
>>> > --
>>> > ================
>>> > Ruoyun  Huang
>>> >
>>>
>>

Reply via email to