Repo Gets into a bad state when Automerge URLs are being requested simultaneously

[justin0mcateer]

We are seeing a reproducible issue where the Repo seems to get into a bad state where all 'find' requests for remote documents fail. We only see this when we have two or more nodes which are creating documents and requesting documents at the same time. We are creating documents in quick succession on each node, when they are in the ready state, we are exchanging the Automerge Urls (out of band). When the remote node tries to request the document it times out after 60 seconds. Once the Repo gets into this state, it seems to never respond to any find request. The problem persists until the node which is the source of the document is restarted.

In our environment, this isn't hard to reproduce, but you need a way to exchange the Automerge Urls quickly.

[pvh]

Thanks for this report! This should, naturally, work. Can you produce a failing test case for it? I think if you look at Repo.test.ts it think it shouldn't be too hard to find an inspirational example.