When I use sftp to make a connection via IPv6, I get the message: "Received message too long 1230009910". After that, sftp terminates. Why? To answer my own question: when I use ssh to connect, I get the message: "IPv6 addr/mask: ::1/128" (the connection is local). Via IPv4 no message is given. So Where does the message "IPv6 addr/mask: ::1/128" originate? After a little searching I found that /usr/lib/libwrap.so.0.7.6 was the culprit. When I use "ALL: [::1/128]" in /etc/hosts.allow, the message appears, and when I use "ALL: LOCAL", I don't reveive the message. I found out about this problem because Nautilus (aftp) timed out after a very long time (and no connection was made). This problem was introduced in Fedora 19. In Fedora 18 I used "sftp://root@localhost/" as a bookmark, and I couldn't connect after the update to Fedora 19. But why is libwrap issuing a message in IPv6 anyway? I see no reason for it (since it doesn't for IPv4). openssh-clients-6.2p2-3.fc19.x86_64 tcp_wrappers-libs-7.6-73.fc19.i686
Forgot to mention that sftp doesn't seem to like messages issued during connection setup (according to http://www.snailbook.com/faq/sftp-corruption.auto.html). The number in the message "Received message too long 1230009910" is decimal and means: 1230009910 (decimal) = 49507636 (hexadecimal) = "IPv6" (ASCII). This text is the first four characters in the string "IPv6 addr/mask: ::1/128".
This is most probably as issue as bug #977995. I'm working on it and I'll push an update asap. Thanks for the report.
I believe that this issue is fixed by https://admin.fedoraproject.org/updates/FEDORA-2013-12660/tcp_wrappers-7.6-74.fc19 update. Feel free to reopen this bug if the problem still persists.