rsync-sudo.sh

#!/bin/bash

# this is a little recipe for using `sudo rsync` on
# old machines that require a tty to call `sudo`.
#
# This prevents errors like:
#
#   sudo: sorry, you must have a tty to run sudo
#
# when using rsync with sudo.
#
# For additional info, see:
# https://bugzilla.redhat.com/show_bug.cgi?id=1020147


# shell quote the given arguments.
function quote {
  # quote each argument, interposing a space each time
  local quoted="$(printf "%q " "$@")"
  # nuke the trailing space
  quoted="${quoted%?}"
  printf "%s" "$quoted"
}

# use `script` to run the given command under a new tty.
function faketty {
  if [[ "$(uname)" == "Darwin" ]]; then
    script -q /dev/null "$@"
  else
    script -qefc "$(quote "$@")" /dev/null
  fi
}

# the bash command to run in place of rsync on the remote host that
# proceeds as follows:
#
# 1) includes a copy of the `quote` and `faketty` definitions
# 2) leverages `script` to run `sudo rsync` under a pty
# 3) passes along the rsync server arguments ($@)
# 4) sets "rsync" as $0
SUDO_RSYNC_CMD="$(
  quote "bash" "-c" "$(declare -f quote); $(declare -f faketty); faketty sudo rsync \"\$@\"" "rsync"
)"

rsync \
  -e "ssh -qi $HOME/.ssh/some.key" \
  --rsync-path "$SUDO_RSYNC_CMD" \
  file.txt user@host:/dest/dir

The faketty stuff could possibly be avoided by passing -tt to ssh, so that it'll create a tty for us. The problem here is that many servers are configured to spew a bunch of noise from interactive shells, so by delaying the tty allocation until we're ready to call sudo, we avoid that trap.

Really, servers just shouldn't set Defaults requiretty and shouldn't spew a bunch of noise, but what can you do...