[WPB-5687] more legalhold tests

[MangoIV]

https://wearezeta.atlassian.net/browse/WPB-5687
it appears I had overlooked to port some of the legalhold tests that were still failing

• legalhold status in user profile
• "Cannot create conversation with both LH activated and non-consenting users" testCannotCreateGroupWithUsersInConflict,
• 

Checklist

• Add a new entry in an appropriate subdirectory of changelog.d
• Read and follow the PR guidelines

[MangoIV]

I have completely removed testGroupConvInvitationHandlesLHConflicts as it was a dejavu, this is tested multiple times across the file, I will port it if I missed something and this tests otherwise untested behaviour but I think this is the same as in other tests.

[MangoIV]

I don't think we need a further changelog entry as this is a follow up PR.

[akshaymankar]

Why do all these tests need a dynamic backend?

[akshaymankar]

I also see that you left some of the LH tests in the old test suite, why so?

[MangoIV]

Why do all these tests need a dynamic backend?

because otherwise the rest of the services don't see the mock LH service; perhaps they don't need it though, might just be a skill issue on my behalf :3 do you have a pointer at some place where this is done without dynamic backend?

I also see that you left some of the LH tests in the old test suite, why so?

they don't flake.

[akshaymankar]

because otherwise the rest of the services don't see the mock LH service; perhaps they don't need it though, might just be a skill issue on my behalf :3 do you have a pointer at some place where this is done without dynamic backend?

If you try to access wire-server-integration-integration.<namespace>.pod.cluster.local from brig/galley it should get to any port listening in the integration test pod.

they don't flake.

Given these flakes are for galley trying to talk to the dynamic service that we spin up, I think they might flake anyway. So, instead of going through this cycle of moving them in batches, IMO it is best if we move all of them.

[MangoIV]

it's still 1400 lines of code to port, this is work that gets tedious after a while and it's harder for concentration, I think this work might benefit from doing in batches ordered by urgency.

[akshaymankar]

it's still 1400 lines of code to port, this is work that gets tedious after a while and it's harder for concentration, I think this work might benefit from doing in batches ordered by urgency.

Sure, IMO the urgency is the same, which tests fail is not consistent across runs. If you find it easier to do it in batches, that's fine too. I think we shouldn't close the ticket until all of these are migrated away.

[MangoIV]

That’s fine, let’s keep it open. These tests should be the ones that are in the hundreds of failures, not the ones that only failed recently.

[fisx]

looks good! please consider my nit-picks before you merge.

[MangoIV]

I have applied your suggestions @fisx with some slight deviations from your doc proposals.