Open a (new) stream to an existing peer?

[ralexstokes]

What is wrong?

Not sure exactly what is going on but had a failure during beacon chain sync.

A MplexUnavailable exception was raised in b/t two "batches" of the initial chain sync.

How can it be fixed

Needs further debugging. Can drop the MAX_REQUEST_BLOCKS to raise the bug faster.

[ralexstokes]

Not able to replicate w/ smaller batch sizes... could have been a transient issue w/ machine hiberation?

could replicate... needs further debugging...

after some more operating the node while benchmarking sync, it seems that this may be a race condition in mplex? sometimes we get the exception and sometimes not...

[ralexstokes]

relevant: this happens when syncing from a lighthouse peer.

and after some more investigation, it may be that the lighthouse peer is closing our connection (need to verify if/why) and then py-libp2p is not detecting the underlying closed connection