sometimes a borg backup gets interrupted with "connection closed by remote".

it is unclear why this happens or whether this is a problem in borg, in the ssh configuration or on the network layer.

Suspicions:

• ssh / tcp keepalive configuration (this was pointed out on the atticmatic site, but it did not help in a experiment a borg user did), so it either does not help or is sometimes not sufficient (from IRC: Azelphur$ FYI, TCPKeepAlive yes, ServerAliveInterval 60, ServerAliveCountMax 5 didn't work :( )
  See also: http://www.cyberciti.biz/tips/open-ssh-server-connection-drops-out-after-few-or-n-minutes-of-inactivity.html
• borg: EINTR and in general exception handling maybe could be improved in remote.py (though it is unclear what would cause EINTR in our case and would not terminate the process anyway). See there for details: http://250bpm.com/blog:12
• there might be longer times with no or only little data transfer (e.g. if borg is reading over a lot of zeros in a sparse file or if all the chunks are already stored in the repo - e.g. when backing up VMs)
• a router might be configured in its security/DoS settings to do something special in case a lot of packets arrive - could this be triggered by borg?
• also (trivial), it might be just a unstable network connection that breaks down