Linux – Netcat fails to start in listening mode

The -k option should do the trick.

From the manpage of nc(1):

 -k      Forces nc to stay listening for another connection after its
         current connection is completed.  It is an error to use this
         option without the -l option.

I've noticed the netcat-traditional package on Debian/Ubuntu does not keep listening as it should. In that case use the netcat-openbsd package instead and try again!

Alternatively, use socat, which is more targeted to your usecase of a proxy server. A random TCP-forwarder example from the manpage of socat which needs some modifications of course.

   socat -d -d -lmlocal2 \
   TCP4-LISTEN:80,bind=myaddr1,reuseaddr,fork,su=nobody,range=10.0.0.0/8 \
   TCP4:www.domain.org:80,bind=myaddr2

          TCP  port  forwarder,  each  side  bound to another local IP
          address (bind). This example  handles  an  almost  arbitrary
          number  of parallel or consecutive connections by fork'ing a
          new process after each accept() . It provides a little secu‐
          rity by su'ing to user nobody after forking; it only permits
          connections from the private  10  network  (range);  due  to
          reuseaddr,   it   allows   immediate  restart  after  master
          process's termination, even if some child  sockets  are  not
          completely  shut down.  With -lmlocal2, socat logs to stderr
          until successfully reaching the accept loop. Further logging
          is directed to syslog with facility local2.

Try

ssh -n 10.10.10.11 "nc -l 8023 | mbuffer -s 128k -m 1G | zfs receive $zfsPath &"
zfs send $newestSnap | mbuffer -s 128k -m 1G | nc -w 60 10.10.10.11 8023

This runs the command and puts it in the background allowing control to pass on to the next line.

I don't have a zfs system to hand so tested it with

ssh -n remote.tld  "nc -l 8023 | wc -l  >/tmp/test &"
ls | nc -w 60 remote.tld 8023

and the correct output appeared in the remote /tmp/test file.