Nondeterministic crashes in extracted program

[palmskog]

Seemingly at random, a node crashes due to failing to send or receive a message. Typical log messages:

in recv_len - recv(): Connection reset by peer

in recv_len - recv(): Resource temporarily unavailable
Fatal error: exception Failure("Marshal.data_size: bad object")

[palmskog]

Hypothesis: may be due to the shim doing repeated reads/sends on a socket in short succession.