[jira] [Updated] (BEAM-8194) virtualSetupEnv task is flaky

["Ahmet Altay (Jira)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/BEAM-8194?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Ahmet Altay updated BEAM-8194:
------------------------------
    Status: Open  (was: Triage Needed)

> virtualSetupEnv task is flaky
> -----------------------------
>
>                 Key: BEAM-8194
>                 URL: https://issues.apache.org/jira/browse/BEAM-8194
>             Project: Beam
>          Issue Type: Bug
>          Components: sdk-py-core
>            Reporter: Ahmet Altay
>            Assignee: Valentyn Tymofieiev
>            Priority: Critical
>
> PyPI sometimes fails when it cannot find dependencies. This happens even when dependencies exist.
> Example: https://builds.apache.org/job/beam_PostCommit_Python2/433/consoleFull
> Log: 
> 05:00:54 Collecting filelock<4,>=3.0.0 (from tox==3.11.1)
> 05:00:54   ERROR: Could not find a version that satisfies the requirement filelock<4,>=3.0.0 (from tox==3.11.1) (from versions: none)
> 05:00:54 ERROR: No matching distribution found for filelock<4,>=3.0.0 (from tox==3.11.1)
> 05:00:57 
> 05:00:57 > Task :sdks:python:setupVirtualenv FAILED
> There was a previous attempt to fix this (https://github.com/apache/beam/pull/7519/files) by adding `--retries 10` to the pip command. However, this only works with connections (https://pip.pypa.io/en/stable/reference/pip/#cmdoption-retries) and I am guessing that when pip decides that a dependency does not exist it requires more than a network retry.
> We could try a gradle task level retry, or wrap the virtualenv creation in a shell script and try do a retry on that.



--
This message was sent by Atlassian Jira
(v8.3.2#803003)