Open ralexstokes opened 4 years ago
~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...
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
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.