Possible mitigation for Python smoke flakes

[itowlson]

No description provided.

[rylev]

@itowlson what is the status of this PR? I'm not sure why it's marked as a draft.

[itowlson]

@rylev It's having the tests run on a regular schedule of "when I remember" to see if the flake still happens. I don't have a lot of faith that this will really fix it, but the flake is intermittent enough that it's hard to know if it's made a difference. Shall I close it? I guess there is not much to do except close it for lack of evidence, or merge it and risk the flake coming back - open to people's preferences.

[itowlson]

I got it to flake... but it flaked on Vault not on Python. Fie!

[vdice]

@itowlson is this change still something we'd like to get in?

[itowlson]

We've not seen Python smoke flakes for a while, I think, so I'm gonna treat this as "creating the PR scared the tests into behaving" and close it.

No it may not be an accurate mental model of computers but it works surprisingly well...