Issues uploading to Anaconda

[jakirkham]

It appears there are some issues uploading to Anaconda.org on all CIs. Here is a log from one such failed AppVeyor upload failure. Though similar ones can be seen for all CIs. Snippet of traceback shown below. Restarting seems to sometimes fix it.

python ci_support\upload_or_check_non_existence.py .\recipe conda-forge --channel=main
Using Anaconda API: https://api.anaconda.org
Traceback (most recent call last):
  File "ci_support\upload_or_check_non_existence.py", line 124, in <module>
    main()
  File "ci_support\upload_or_check_non_existence.py", line 106, in main
    on_channel = distribution_exists_on_channel(cli, meta, owner, channel)
  File "ci_support\upload_or_check_non_existence.py", line 69, in distribution_exists_on_channel
    binstar_cli.show_channel(owner=owner, channel=channel)['files']]
  File "C:\conda\lib\site-packages\binstar_client\mixins\channels.py", line 29, in show_channel
    self._check_response(res, [200])
  File "C:\conda\lib\site-packages\binstar_client\__init__.py", line 200, in _check_response
    raise ErrCls(msg, res.status_code)
binstar_client.errors.NotFound: ('Not Found: Nothing matches the given URI ([GET] https://api.anaconda.org/channels/conda-forge/main -> 404)', 404)
Command exited with code 1

cc @msarahan @mcg1969

[jakirkham]

Raising for awareness of affected teams/feedstocks that have tried to deploy recently and are now having deployment issues. Feel free to track an issue locally on your feedstock until all missing packages are uploaded.

cc @conda-forge/bqplot @conda-forge/requests-futures @conda-forge/hdbscan @conda-forge/versioneer

[jakirkham]

Restarted builds for the apparently affected feedstocks. Seems to be localized to very recent updates or new feedstocks.

[jakirkham]

All of the restarted builds appear to have passed and uploaded packages FWICT. If others can confirm, that would be great. Going to go ahead and close for now assuming this is resolved.

[lmcinnes]

The Circle CI builds were having the same problem, but the AppVeyor build looks good for me now. Presumably restarting the Circle CI builds will get those working too. Thanks.

[jakirkham]

Yeah, was just noticing this with a different CircleCI build just now. That build was from a few days ago. Going to elevate this issue.

[jakirkham]

Raised issue ( https://github.com/Anaconda-Platform/support/issues/60 ) about this general problem.

[SylvainCorlay]

I am also missing a linux-64/py35 file on my recent build.

Using Anaconda API: https://api.anaconda.org
Traceback (most recent call last):
  File "/feedstock_root/ci_support/upload_or_check_non_existence.py", line 124, in <module>
    main()
  File "/feedstock_root/ci_support/upload_or_check_non_existence.py", line 106, in main
    on_channel = distribution_exists_on_channel(cli, meta, owner, channel)
  File "/feedstock_root/ci_support/upload_or_check_non_existence.py", line 69, in distribution_exists_on_channel
    binstar_cli.show_channel(owner=owner, channel=channel)['files']]
  File "/opt/conda/lib/python3.5/site-packages/binstar_client/mixins/channels.py", line 29, in show_channel
    self._check_response(res, [200])
  File "/opt/conda/lib/python3.5/site-packages/binstar_client/__init__.py", line 200, in _check_response
    raise ErrCls(msg, res.status_code)
binstar_client.errors.NotFound: ('Not Found: Nothing matches the given URI ([GET] https://api.anaconda.org/channels/conda-forge/main -> 404)', 404)

Just restarted the build.

[jakirkham]

Try restarting the build and see if it fixes it. This has also worked for me with CircleCI builds that failed to upload.

[jakirkham]

In some cases, am also seeing a new failure when rerunning an existing build that had already uploaded. This rerunning is being done in order to resolve some of the failed uploads. This new failure suggests the problem is not strictly limited to uploading. See the traceback below for the error and this build for more details.

python ci_support\upload_or_check_non_existence.py .\recipe conda-forge --channel=main
Using Anaconda API: https://api.anaconda.org
Traceback (most recent call last):
  File "ci_support\upload_or_check_non_existence.py", line 124, in <module>
    main()
  File "ci_support\upload_or_check_non_existence.py", line 106, in main
    on_channel = distribution_exists_on_channel(cli, meta, owner, channel)
  File "ci_support\upload_or_check_non_existence.py", line 69, in distribution_exists_on_channel
    binstar_cli.show_channel(owner=owner, channel=channel)['files']]
  File "C:\conda\lib\site-packages\binstar_client\mixins\channels.py", line 28, in show_channel
    res = self.session.get(url)
  File "C:\conda\lib\site-packages\requests\sessions.py", line 488, in get
    return self.request('GET', url, **kwargs)
  File "C:\conda\lib\site-packages\requests\sessions.py", line 475, in request
    resp = self.send(prep, **send_kwargs)
  File "C:\conda\lib\site-packages\requests\sessions.py", line 628, in send
    r.content
  File "C:\conda\lib\site-packages\requests\models.py", line 755, in content
    self._content = bytes().join(self.iter_content(CONTENT_CHUNK_SIZE)) or bytes()
  File "C:\conda\lib\site-packages\requests\models.py", line 679, in generate
    raise ChunkedEncodingError(e)
requests.exceptions.ChunkedEncodingError: ('Connection broken: IncompleteRead(0 bytes read)', IncompleteRead(0 bytes read))
Command exited with code 1

[jakirkham]

Appears your feedstock was affected as well, @conda-forge/cssutils. Have restarted the CircleCI build.

[jakirkham]

Was starting to look like things were stabilizing as I hadn't seen this issue for awhile. However, saw this build from gdal that is struggling with a related problem.

Edit: Restarted the build.

cc @conda-forge/gdal

[jakirkham]

Also happened with some curl builds recently.

Edit: Restarted those too.

cc @conda-forge/curl

[jakirkham]

Appears pysocks suffered from this in yesterday's builds. Have restarted them.

cc @conda-forge/pysocks

[jakirkham]

Affecting sphinx on both CircleCI and AppVeyor. Have restarted both.

cc @conda-forge/sphinx

[jakirkham]

Can confirm this is happening with all CIs now thanks to being able to track down this issue with Travis CI. ( https://github.com/conda-forge/line_profiler-feedstock/issues/6 ) Have already triggered a rebuild to get the missing packages.

cc @conda-forge/line_profiler

[jakirkham]

It looks like we may need some improvements to how we are doing uploads. This is documented briefly in issue ( https://github.com/conda-forge/conda-smithy/issues/270 ). It relies on a new Anaconda API that is going through some testing before it rolls out in the next day or two.

[jakirkham]

Turns out this issue goes back several weeks as demonstrated by this failure.

[jakirkham]

FWIW I haven't seen this issue really at all today. So maybe things are already improving.

[jakirkham]

This is still persisting. Here is a recent failure.

[jakirkham]

Seeing an uptick in this issue of late.

[vamega]

I think I'm seeing another instance of this issue at: https://github.com/conda-forge/yapf-feedstock/pull/3

[jakirkham]

That's a different issue. It's fixed by re-rendering with a newer version of conda-smithy. Please try that and see if it fixes it.

[jakirkham]

On another note, we have released a new copy of conda-smithy that should address the upload issue that was causing us difficulties before. If you encounter this issue, also please try re-rendering. Am going to close this as we have hopefully fixed it. If not please ping us, and we can open a fresh issue. Thanks. 😄