Test failover logic for MSC3083 #172
Conversation
tests/msc3083_test.go
Outdated
There was a problem hiding this comment.
This should probably be synced with alice.SyncUntilTimelineHas as well right?
There was a problem hiding this comment.
Is there any reason this is just always synced? Seems like it would be an improvement.
tests/msc3083_test.go
Outdated
There was a problem hiding this comment.
Why don't you use setupRestrictedRoom in the previous test?
There was a problem hiding this comment.
The joiners need to be on the same server while the creator needs to be on a separate server. Unfortunately BlueprintFederationTwoLocalOneRemote has alice and bob on one server and charlie on another.
tests/msc3083_test.go
Outdated
There was a problem hiding this comment.
Shouldn't we wait for these to come down alice sync stream?
There was a problem hiding this comment.
It doesn't matter if alice sees them.
tests/msc3083_test.go
Outdated
There was a problem hiding this comment.
Fairly sure there won't be any guarantee that hs2 would've seen the join from hs3 because the join was between hs3 and hs1. Therefore, doing this will be racey.
clokep
force-pushed
the
clokep/restricted-auth-err
branch
from
0209430 to
231d6be
Compare
clokep
force-pushed
the
clokep/restricted-auth-err
branch
from
231d6be to
11ddf9c
Compare
|
I rebased this, sorry about that. Wanted to remove the commits from the previous PR and forgot there had been a review on this already. 🤦 |
anoadragon453
left a comment
anoadragon453
left a comment
There was a problem hiding this comment.
This looks pretty sensible to me now from a test coverage perspective. Just one minor comment on a previous discussion.
This tests what happens if a server is unable to authorize a room join (due to not being able to invite users or not being able to check the proper allowed rooms).
Corresponds to matrix-org/synapse#10447
Based on #145