Re: PyPI JSON API redirect loop for all unpublished packages

[Dustin Ingram]

We didn't really experience this at the Bloomberg sprints this fall, but
the sprint size was quite a bit smaller than the PyCon sprints.

On Fri, Dec 21, 2018 at 3:07 PM Sumana Harihareswara 
wrote:

> Donald and Dustin: have we been running into these kinds of Travis
> problems in the past few months or does it seem to have settled down?
>
> --
> Sumana Harihareswara
> Changeset Consulting
> https://changeset.nyc
>
> On 5/19/18 4:35 PM, Brett Cannon wrote:
> > On Fri, May 18, 2018, 06:08 Dustin Ingram, 
> wrote:
> >
> >> I did reach out to the one contact we had there from when GCP/Fastly
> >> were having issues that affected Travis/PyPI (Emma) on Monday, but got
> >> no response.
> >>
> >
> > If Travis doesn't work out then let Steve Dower and me know and we can
> see
> > if we can get you extra credits on VSTS (teammate of mine was already
> > working with Jason at the PyCon sprints to get setuptools up on
> > PyPA.visualstudio.com).
> >
> > -Brett
> >
> >
> >> D.
> >>
> >> On Fri, May 18, 2018 at 8:54 AM, Sumana Harihareswara  >
> >> wrote:
> >>> In my opinion, this kind of bottleneck is likely to happen more
> >> frequently as we increase PyPA development activity, so it'd be worth
> >> asking Travis to bump up our account's oomph. (I am very tired and on a
> >> train so please forgive handwavy wording.) Do we have a contact there?
> >>>
> >>> --
> >>> Sumana Harihareswara
> >>> Changeset Consulting
> >>> https://changeset.nyc
> >>>
> >>> On 05/16/2018 04:36 PM, Dustin Ingram wrote:
>  Thanks for the report. I reverted the commit that caused this in
>  https://github.com/pypa/warehouse/pull/4015, however it might take a
>  bit for this to get deployed to PyPI because there's currently a
>  pretty long backlog in Travis due to all the PyPA development
>  happening during the sprints.
> 
>  D.
>

Re: PyPI JSON API redirect loop for all unpublished packages

[Sumana Harihareswara]

Donald and Dustin: have we been running into these kinds of Travis problems in 
the past few months or does it seem to have settled down?

-- 
Sumana Harihareswara
Changeset Consulting
https://changeset.nyc

On 5/19/18 4:35 PM, Brett Cannon wrote:
> On Fri, May 18, 2018, 06:08 Dustin Ingram,  wrote:
> 
>> I did reach out to the one contact we had there from when GCP/Fastly
>> were having issues that affected Travis/PyPI (Emma) on Monday, but got
>> no response.
>>
> 
> If Travis doesn't work out then let Steve Dower and me know and we can see
> if we can get you extra credits on VSTS (teammate of mine was already
> working with Jason at the PyCon sprints to get setuptools up on
> PyPA.visualstudio.com).
> 
> -Brett
> 
> 
>> D.
>>
>> On Fri, May 18, 2018 at 8:54 AM, Sumana Harihareswara 
>> wrote:
>>> In my opinion, this kind of bottleneck is likely to happen more
>> frequently as we increase PyPA development activity, so it'd be worth
>> asking Travis to bump up our account's oomph. (I am very tired and on a
>> train so please forgive handwavy wording.) Do we have a contact there?
>>>
>>> --
>>> Sumana Harihareswara
>>> Changeset Consulting
>>> https://changeset.nyc
>>>
>>> On 05/16/2018 04:36 PM, Dustin Ingram wrote:
 Thanks for the report. I reverted the commit that caused this in
 https://github.com/pypa/warehouse/pull/4015, however it might take a
 bit for this to get deployed to PyPI because there's currently a
 pretty long backlog in Travis due to all the PyPA development
 happening during the sprints.

 D.

Re: PyPI JSON API redirect loop for all unpublished packages

[Doug Hellmann]

Excerpts from Donald Stufft's message of 2018-05-16 17:09:12 -0400:
> 
> > On May 16, 2018, at 5:06 PM, Doug Hellmann  wrote:
> > 
> > 
> > 
> > We had an issue testing redirect rules in the OpenStack documentation
> > build, so we built a tool to test the .htaccess files. I don't know
> > if whereto would be directly useful (maybe you're not using Apache?),
> > but I thought I'd mention it in case someone thinks it is. I would also
> > happily take patches to make it read some other input format, if needed.
> > 
> > https://docs.openstack.org/whereto/latest/
> > 
> > 
> > 
> 
> 
> Unfortunately, it was an issue with the app itself rather than the web server 
> (we don’t actually use a web server to handle any logic besides generic 
> buffering and passing on requests to our app server).

Ah, well.

Doug

Re: PyPI JSON API redirect loop for all unpublished packages

[Donald Stufft]

> On May 16, 2018, at 5:06 PM, Doug Hellmann  wrote:
> 
> 
> 
> We had an issue testing redirect rules in the OpenStack documentation
> build, so we built a tool to test the .htaccess files. I don't know
> if whereto would be directly useful (maybe you're not using Apache?),
> but I thought I'd mention it in case someone thinks it is. I would also
> happily take patches to make it read some other input format, if needed.
> 
> https://docs.openstack.org/whereto/latest/
> 
> 
> 


Unfortunately, it was an issue with the app itself rather than the web server 
(we don’t actually use a web server to handle any logic besides generic 
buffering and passing on requests to our app server).

Re: PyPI JSON API redirect loop for all unpublished packages

[Doug Hellmann]

We had an issue testing redirect rules in the OpenStack documentation
build, so we built a tool to test the .htaccess files. I don't know
if whereto would be directly useful (maybe you're not using Apache?),
but I thought I'd mention it in case someone thinks it is. I would also
happily take patches to make it read some other input format, if needed.

https://docs.openstack.org/whereto/latest/



Excerpts from Dustin Ingram's message of 2018-05-16 16:36:48 -0400:
> Thanks for the report. I reverted the commit that caused this in
> https://github.com/pypa/warehouse/pull/4015, however it might take a
> bit for this to get deployed to PyPI because there's currently a
> pretty long backlog in Travis due to all the PyPA development
> happening during the sprints.
> 
> D.
> 
> On Wed, May 16, 2018 at 4:28 PM, Alex Becker  wrote:
> > In the last hour I started observing the following loop:
> >
> > https://pypi.org/pypi/nonexistant-package/json ->
> > https://pypi.org/pypi/nonexistant-package/json/ ->
> > https://pypi.org/pypi/nonexistant-package/json
> >
> > This is breaking pipenv package resolution for us, since we use 2 different
> > indices and it checks for a 404 from PyPI before going on to the second one.
> > I assume any pipenv or pip user relying on multiple indices is currently
> > unable to resolve packages.

Re: PyPI JSON API redirect loop for all unpublished packages

[Dustin Ingram]

Thanks for the report. I reverted the commit that caused this in
https://github.com/pypa/warehouse/pull/4015, however it might take a
bit for this to get deployed to PyPI because there's currently a
pretty long backlog in Travis due to all the PyPA development
happening during the sprints.

D.

On Wed, May 16, 2018 at 4:28 PM, Alex Becker  wrote:
> In the last hour I started observing the following loop:
>
> https://pypi.org/pypi/nonexistant-package/json ->
> https://pypi.org/pypi/nonexistant-package/json/ ->
> https://pypi.org/pypi/nonexistant-package/json
>
> This is breaking pipenv package resolution for us, since we use 2 different
> indices and it checks for a 404 from PyPI before going on to the second one.
> I assume any pipenv or pip user relying on multiple indices is currently
> unable to resolve packages.

Re: PyPI JSON API redirect loop for all unpublished packages

[Alex Becker]

Looks like a fix was just merged to 
Warehouse: https://github.com/pypa/warehouse/pull/4015

I assume that will go out shortly, so this can probably be ignored.

On Wednesday, May 16, 2018 at 1:28:23 PM UTC-7, Alex Becker wrote:
>
> In the last hour I started observing the following loop:
>
> https://pypi.org/pypi/nonexistant-package/json -> 
> https://pypi.org/pypi/nonexistant-package/json/ -> 
> https://pypi.org/pypi/nonexistant-package/json
>
> This is breaking pipenv package resolution for us, since we use 2 
> different indices and it checks for a 404 from PyPI before going on to the 
> second one. I assume any pipenv or pip user relying on multiple indices is 
> currently unable to resolve packages.
>