Set CONDA_REMOTE_BACKOFF to 2.

[mbargull]

Description

Same as gh-351 but using "backoff" instead of more retries:

This might help with the random connection errors when running the tests.

Checklist - did you ...

• Add a file to the news directory (using the template) for the next release's release notes?
• Add / update necessary tests?
• Add / update outdated documentation?

[mbargull]

Thanks for letting this run twice, @jezdez!

Apparently this one doesn't seem to help, too... :/

Looking at the recent CI runs (from other branches), the connection error seems to predominantly occur in job macos (conda-forge, 3.9, unit, 1, 2).

For this branch, in the first CI run attempt tests/cli/test_subcommands.py::test_env_create failed for jobs

• windows (conda-forge, 3.11, unit, 1)
• macos (conda-forge, 3.9, unit, 1, 2)

In the second attempt only for

• macos (conda-forge, 3.9, unit, 1, 2)

---

Unrelated to the connection issues, the first run also had linux (conda-forge, 3.11, unit, 2) fail with:

2023-11-04T18:21:50.3060042Z   File "/opt/conda-src/tests/core/test_solve.py", line 77, in test_solve_1
2023-11-04T18:21:50.3060722Z     assert convert_to_dist_str(final_state) == order
2023-11-04T18:21:50.3064988Z AssertionError: assert ('channel-1/linux-64::openssl-1.0.1c-0', 'channel-1/linux-64::readline-6.2-0', 'channel-1/linux-64::sqlite-3.7.13-0', 'channel-1/linux-64::system-5.8-1', 'channel-1/linux-64::tk-8.5.13-0', 'channel-1/linux-64::zlib-1.2.7-0', 'channel-1/linux-64::python-2.6.8-6', 'channel-1/linux-64::numpy-1.7.1-py26_0') == ('channel-1/linux-64::openssl-1.0.1c-0', 'channel-1/linux-64::readline-6.2-0', 'channel-1/linux-64::sqlite-3.7.13-0', 'channel-1/linux-64::system-5.8-1', 'channel-1/linux-64::tk-8.5.13-0', 'channel-1/linux-64::zlib-1.2.7-0', 'channel-1/linux-64::python-2.7.5-0', 'channel-1/linux-64::numpy-1.7.1-py27_0')
2023-11-04T18:21:50.3068741Z   At index 6 diff: 'channel-1/linux-64::python-2.6.8-6' != 'channel-1/linux-64::python-2.7.5-0'
2023-11-04T18:21:50.3069422Z   Full diff:
2023-11-04T18:21:50.3069669Z     (
2023-11-04T18:21:50.3070104Z      'channel-1/linux-64::openssl-1.0.1c-0',
2023-11-04T18:21:50.3070657Z      'channel-1/linux-64::readline-6.2-0',
2023-11-04T18:21:50.3071199Z      'channel-1/linux-64::sqlite-3.7.13-0',
2023-11-04T18:21:50.3071715Z      'channel-1/linux-64::system-5.8-1',
2023-11-04T18:21:50.3072216Z      'channel-1/linux-64::tk-8.5.13-0',
2023-11-04T18:21:50.3072712Z      'channel-1/linux-64::zlib-1.2.7-0',
2023-11-04T18:21:50.3073208Z   -  'channel-1/linux-64::python-2.7.5-0',
2023-11-04T18:21:50.3073633Z   ?                                ^ ^ ^
2023-11-04T18:21:50.3074123Z   +  'channel-1/linux-64::python-2.6.8-6',
2023-11-04T18:21:50.3074520Z   ?                                ^ ^ ^
2023-11-04T18:21:50.3075040Z   -  'channel-1/linux-64::numpy-1.7.1-py27_0',
2023-11-04T18:21:50.3075483Z   ?                                      ^
2023-11-04T18:21:50.3075987Z   +  'channel-1/linux-64::numpy-1.7.1-py26_0',
2023-11-04T18:21:50.3076413Z   ?                                      ^
2023-11-04T18:21:50.3076770Z     )

^-- is that one known/has it been investigated yet.

[mbargull]

Oh, I just read the tests code...
Well, no matter how many/how long the retries were, we'd always fail the test if a retry happened:
https://github.com/conda/conda/blob/16de74fd6844809b45eea0bf61a70b04d2c96715/tests/cli/test_subcommands.py#L221
We get the warning log from https://github.com/urllib3/urllib3/blob/2.0.7/src/urllib3/connectionpool.py#L872-L874 output to err.
We should filter that log message. Not sure on what level though...

[mbargull]

We should filter that log message. Not sure on what level though...

Addressing upstream in conda/conda#13284 .