<div dir="ltr"><br><div class="gmail_extra"><br><div class="gmail_quote">On Wed, Jun 8, 2016 at 7:55 PM, john hood <span dir="ltr">&lt;<a href="mailto:cgull@glup.org" target="_blank">cgull@glup.org</a>&gt;</span> wrote:<br><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;border-left-color:rgb(204,204,204);padding-left:1ex">Yeah, would you open a GitHub issue?</blockquote><div><br></div><div>Will do. </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;border-left-color:rgb(204,204,204);padding-left:1ex">
<br>
I tried my two Macs, and it breaks for me, but differently:  they get<br>
link-local IPv6 addresses from mDNS, requiring an interface identifier,<br>
so mosh-client doesn&#39;t know where to send packets.<br>
<br>
I&#39;m wondering how you&#39;re getting global IPv6 addresses from mDNS, even<br>
though both your client and server are on the same local network.<br>
Judging by the hostnames you&#39;re using and the IPv6 address, this is a<br>
corporate OS X install on $yourverylargecompany&#39;s network.<br>
<br></blockquote><div>Yes, good guess.  :-) I think it&#39;s relatively common to have a global v6 address allocated to a system which is RFC1918 in v4 these days.</div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;border-left-color:rgb(204,204,204);padding-left:1ex">
If the server involved has regular DNS A/AAAA records, does that work<br>
with Mosh?<br></blockquote><div><br></div><div>I verified I see the same behavior moshing into systems with both global A/AAAA records, running an older version of mosh server (mosh 1.2.4a).  So it seems it isn&#39;t dependent on mDNS, just an old server running on a dual stack server.  I&#39;ll write up the bug.</div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;border-left-color:rgb(204,204,204);padding-left:1ex">
<br>
regards,<br>
<br>
  --jh<br>
<span class=""><br>
On 6/8/16 9:37 PM, John Hood wrote:<br>
&gt; Er, that&#39;s an interesting bug, it should work for either v4 or v6.  I&#39;ll<br>
&gt; check mDNS tonight.<br>
&gt;<br>
&gt; regards,<br>
&gt;<br>
&gt;   --jh<br>
&gt;<br>
&gt; On 06/08/2016 19:54, Bryce Bockman wrote:<br>
&gt;&gt; I installed the OS X binary version (gotten from<br>
</span>&gt;&gt; &lt;<a href="https://github.com/mobile-shell/mosh/releases/download/mosh-1.2.5.95rc1/mosh-1.2.5.95rc1.pkg" rel="noreferrer" target="_blank">https://github.com/mobile-shell/mosh/releases/download/mosh-1.2.5.95rc1/mosh-1.2.5.95rc1.pkg</a>&gt;<a href="https://github.com/mobile-shell/mosh/releases/download/mosh-1.2.5.95rc1/mosh-1.2.5.95rc1.pkg" rel="noreferrer" target="_blank">https://github.com/mobile-shell/mosh/releases/download/mosh-1.2.5.95rc1/mosh-1.2.5.95rc1.pkg</a>).<br>
<span class="">&gt;&gt; The install went fine, and I was able to mosh into an older server<br>
&gt;&gt; running on stock Ubuntu 14.04.<br>
&gt;&gt;<br>
&gt;&gt; Found an interesting bug/feature, likely due to ipv6 related changes.<br>
&gt;&gt; I often use mDNS name resolution. When moshing from the release<br>
&gt;&gt; candidate to the older server mentioned above using mDNS target I get<br>
&gt;&gt; the following:<br>
&gt;&gt;<br>
&gt;&gt; bb@bb-mac:~$ mosh beb-dev01.local<br>
&gt;&gt;<br>
&gt;&gt; mosh-server: Bad IP address (2606:b400:605:d0a0:dacb:8aff:fe69:7a71)<br>
&gt;&gt;<br>
&gt;&gt; Usage: mosh-server new [-s] [-v] [-i LOCALADDR] [-p PORT[:PORT2]] [-c<br>
&gt;&gt; COLORS] [-l NAME=VALUE] [-- COMMAND...]<br>
&gt;&gt;<br>
&gt;&gt; Connection to beb-dev01.local closed.<br>
&gt;&gt;<br>
&gt;&gt; /usr/local/bin/mosh: Did not find mosh server startup message.<br>
&gt;&gt;<br>
&gt;&gt; bb@bb-mac:~$<br>
&gt;&gt;<br>
&gt;&gt;<br>
&gt;&gt; Note that mDNS resolves either v4 or v6 addresses:<br>
&gt;&gt;<br>
&gt;&gt;<br>
&gt;&gt; bbockman@bbockman-mac:~$ ping6 beb-dev01.local<br>
&gt;&gt;<br>
&gt;&gt; PING6(56=40+8+8 bytes) 2606:b400:605:d0a0:16b1:c8ff:fe00:5da0 --&gt;<br>
&gt;&gt; 2606:b400:605:d0a0:dacb:8aff:fe69:7a71<br>
&gt;&gt;<br>
&gt;&gt; 16 bytes from 2606:b400:605:d0a0:dacb:8aff:fe69:7a71, icmp_seq=0<br>
&gt;&gt; hlim=64 time=0.549 ms<br>
&gt;&gt;<br>
&gt;&gt; bbockman@bbockman-mac:~$ ping beb-dev01.local<br>
&gt;&gt;<br>
&gt;&gt; PING beb-dev01.local (10.145.137.58): 56 data bytes<br>
&gt;&gt;<br>
</span>&gt;&gt; 64 bytes from 10.145.137.58 &lt;<a href="http://10.145.137.58" rel="noreferrer" target="_blank">http://10.145.137.58</a>&gt;: icmp_seq=0 ttl=64<br>
<span class="">&gt;&gt; time=0.359 ms<br>
&gt;&gt;<br>
&gt;&gt;<br>
&gt;&gt; If I explicitely mosh using the v4 address things work (mosh<br>
&gt;&gt; 10.145.137.58).<br>
&gt;&gt;<br>
&gt;&gt; Not sure the extent of the behavior. e.g. Would the same behavior be<br>
&gt;&gt; seen with an older server that has a AAAA record with newer client or<br>
&gt;&gt; not.  Should open an issue on github?<br>
&gt;&gt;<br>
&gt;&gt;<br>
&gt;&gt;<br>
&gt;&gt; On Wed, Jun 8, 2016 at 12:44 PM, Keith Winstein &lt;<a href="mailto:keithw@mit.edu">keithw@mit.edu</a><br>
</span><span class="">&gt;&gt; &lt;mailto:<a href="mailto:keithw@mit.edu">keithw@mit.edu</a>&gt;&gt; wrote:<br>
&gt;&gt;<br>
&gt;&gt;     Thanks all who have tested the mosh 1.2.6 release candidate so far.<br>
&gt;&gt;<br>
&gt;&gt;     We&#39;re still looking for a postive ACK that somebody has<br>
&gt;&gt;     successfully tested the release candidate on:<br>
&gt;&gt;<br>
&gt;&gt;     - Fedora<br>
&gt;&gt;     - OS X binary .pkg<br>
&gt;&gt;     (<a href="https://github.com/mobile-shell/mosh/releases/download/untagged-0d43a0f73825e9952677/mosh-1.2.5.95rc1.pkg" rel="noreferrer" target="_blank">https://github.com/mobile-shell/mosh/releases/download/untagged-0d43a0f73825e9952677/mosh-1.2.5.95rc1.pkg</a>)<br>
&gt;&gt;     - OS X compiling from source<br>
&gt;&gt;<br>
&gt;&gt;     Thanks all,<br>
&gt;&gt;     Keith<br>
&gt;&gt;<br>
&gt;&gt;     On Sun, Jun 5, 2016 at 8:39 PM, Richard Woodbury<br>
</span><span class="">&gt;&gt;     &lt;&lt;mailto:<a href="mailto:rpwoodbu@mybox.org">rpwoodbu@mybox.org</a>&gt;<a href="mailto:rpwoodbu@mybox.org">rpwoodbu@mybox.org</a>&gt; wrote:<br>
&gt;&gt;<br>
&gt;&gt;         I have confirmed that the RC builds and runs OK in Mosh for<br>
&gt;&gt;         Chrome.<br>
&gt;&gt;<br>
&gt;&gt;         On Thu, Jun 2, 2016 at 11:47 PM Richard Woodbury<br>
</span><span class="">&gt;&gt;         &lt;&lt;mailto:<a href="mailto:rpwoodbu@mybox.org">rpwoodbu@mybox.org</a>&gt;<a href="mailto:rpwoodbu@mybox.org">rpwoodbu@mybox.org</a>&gt; wrote:<br>
&gt;&gt;<br>
&gt;&gt;             Hi, Keith. I don&#39;t have predictable time right now, but as<br>
&gt;&gt;             I can, I&#39;ll look into this for Mosh for Chrome. I did<br>
&gt;&gt;             manage to do a quick &quot;smoke test&quot; build, but I&#39;m getting a<br>
&gt;&gt;             bunch of protobuf link errors. I&#39;ll need to do more<br>
&gt;&gt;             investigation, which may include bringing in a newer NaCl<br>
&gt;&gt;             SDK so I can also get newer libraries from naclports.<br>
&gt;&gt;             Hopefully I&#39;ll find time in the next day or two.<br>
&gt;&gt;<br>
&gt;&gt;             I can add that it builds OK on the Raspberry Pi (ARM,<br>
&gt;&gt;             Raspbian), and the performance improvement makes a<br>
</span>&gt;&gt;             /huge/ difference! mosh-server used to take most of the<br>
<span class="">&gt;&gt;             CPU for busy output or large window size and was notably<br>
&gt;&gt;             sluggish, depending on local echo to be usable. Now the<br>
&gt;&gt;             CPU usage is in the noise, and it responds wonderfully.<br>
&gt;&gt;<br>
&gt;&gt;<br>
&gt;&gt;             On Mon, May 30, 2016 at 3:34 AM Keith Winstein<br>
</span><span class="">&gt;&gt;             &lt;&lt;mailto:<a href="mailto:keithw@mit.edu">keithw@mit.edu</a>&gt;<a href="mailto:keithw@mit.edu">keithw@mit.edu</a>&gt; wrote:<br>
&gt;&gt;<br>
&gt;&gt;                 Hi folks,<br>
&gt;&gt;<br>
&gt;&gt;                 Could you please send in some positive testing reports<br>
&gt;&gt;                 on the Mosh 1.2.6 release candidate?<br>
&gt;&gt;<br>
&gt;&gt;                 It would be great to have an independent &quot;looks good&quot;<br>
&gt;&gt;                 from at least the following platforms before we cut<br>
&gt;&gt;                 the release:<br>
&gt;&gt;<br>
&gt;&gt;                 - Fedora<br>
&gt;&gt;                 - the OS X binary .pkg<br>
&gt;&gt;                 - OS X compiling from source<br>
&gt;&gt;                 - OpenBSD<br>
&gt;&gt;                 - FreeBSD<br>
&gt;&gt;                 - NetBSD<br>
&gt;&gt;                 - Chrome<br>
&gt;&gt;<br>
&gt;&gt;                 Thanks,<br>
&gt;&gt;                 Keith<br>
&gt;&gt;<br>
&gt;&gt;                 On Wed, May 25, 2016 at 8:56 PM, john hood<br>
</span><div><div class="h5">&gt;&gt;                 &lt;&lt;mailto:<a href="mailto:cgull@glup.org">cgull@glup.org</a>&gt;<a href="mailto:cgull@glup.org">cgull@glup.org</a>&gt; wrote:<br>
&gt;&gt;<br>
&gt;&gt;                     Hi all,<br>
&gt;&gt;<br>
&gt;&gt;                     We&#39;re happy to announce the upcoming release of<br>
&gt;&gt;                     Mosh 1.2.6, and are<br>
&gt;&gt;                     calling for testing on Mosh 1.2.5.95rc1.  The<br>
&gt;&gt;                     release has picked up<br>
&gt;&gt;                     some minor new features in the year since the last<br>
&gt;&gt;                     release such as<br>
&gt;&gt;                     better IPv6 support and tools to handle orphaned<br>
&gt;&gt;                     sessions.  However,<br>
&gt;&gt;                     it&#39;s also seen significant improvements in<br>
&gt;&gt;                     performance, testing, and<br>
&gt;&gt;                     portability.<br>
&gt;&gt;<br>
&gt;&gt;                     The Changelog for this release:<br>
&gt;&gt;<br>
&gt;&gt;                       * New features:<br>
&gt;&gt;                         * Add Travis CI builds for Linux and Mac.<br>
&gt;&gt;                     (Anders Kaseorg, others)<br>
&gt;&gt;                         * Add a --local option to run without ssh.<br>
&gt;&gt;                     (John Hood)<br>
&gt;&gt;                         * Mosh now returns exitstatus reflecting<br>
&gt;&gt;                     connection success.<br>
&gt;&gt;                           (John Hood)<br>
&gt;&gt;                         * Add a end-to-end test suite and many tests.<br>
&gt;&gt;                     (John Hood)<br>
&gt;&gt;                         * Implement timeouts and signals to help<br>
&gt;&gt;                     address orphaned sessions.<br>
&gt;&gt;                           (John Hood)<br>
&gt;&gt;                         * Major rework of Mosh&#39;s display<br>
&gt;&gt;                     differencing/rendering<br>
&gt;&gt;                           code with much improved performance for slow<br>
&gt;&gt;                     machines.  (John Hood)<br>
&gt;&gt;                         * Implement ANSI back/forward tab (CSI CBT,<br>
&gt;&gt;                     CSI CHT).<br>
&gt;&gt;                           (John Hood)<br>
&gt;&gt;                         * Do not start user shell until network<br>
&gt;&gt;                     session starts.<br>
&gt;&gt;                           (John Hood)<br>
&gt;&gt;                         * Add options for more flexible specification<br>
&gt;&gt;                     of IPv4/IPv6<br>
&gt;&gt;                           hostname resolution.  (John Hood)<br>
&gt;&gt;                         * Improved bash completion.  (Steve Dignam,<br>
&gt;&gt;                     HIGUCHI Yuta)<br>
&gt;&gt;                         * Add options for different methods of<br>
&gt;&gt;                     resolving the remote host<br>
&gt;&gt;                           address, allowing operation without<br>
&gt;&gt;                     SshProxyCommand.  (John Hood)<br>
&gt;&gt;<br>
&gt;&gt;                       * Platform support:<br>
&gt;&gt;                         * Add configurable support for Apple Common<br>
&gt;&gt;                     Crypto and<br>
&gt;&gt;                           Nettle, in place of OpenSSL.  Implement<br>
&gt;&gt;                     base64 locally.<br>
&gt;&gt;                           (John Hood)<br>
&gt;&gt;                         * Workaround Cygwin select() bug.  (John Hood)<br>
&gt;&gt;                         * Updates to Debian packaging.  (Anders<br>
&gt;&gt;                     Kaseorg, Keith Winstein)<br>
&gt;&gt;                         * Workaround a glibc-2.22 issue causing<br>
&gt;&gt;                     segfaults on Debian Sid.<br>
&gt;&gt;                           (John Hood with help from many others)<br>
&gt;&gt;                         * Prefer c++ to g++, for systems like FreeBSD<br>
&gt;&gt;                     where g++ is not usable.<br>
&gt;&gt;                           (John Hood)<br>
&gt;&gt;                         * Fixes for Illumos Hipster 20151003.  (John Hood)<br>
&gt;&gt;                         * Disable -Werror for protobuf code, to<br>
&gt;&gt;                     resolve a new gcc6 warning.<br>
&gt;&gt;                           (John Hood)<br>
&gt;&gt;                         * Link test for -fstack-protector-all on an<br>
&gt;&gt;                     embedded platform.<br>
&gt;&gt;                           (Baruch Siach)<br>
&gt;&gt;                         * Resolve issue with bswap64() on<br>
&gt;&gt;                     FreeBSD-CURRENT with libc++-3.8.0.<br>
&gt;&gt;                           (John Hood)<br>
&gt;&gt;                         * Fix issue with RECVTOS error message on<br>
&gt;&gt;                     client on FreeBSD.<br>
&gt;&gt;                           (John Hood)<br>
&gt;&gt;<br>
&gt;&gt;                       * Bug fixes:<br>
&gt;&gt;                         * Remove an assertion causing aborts on<br>
&gt;&gt;                     Unicode fallback found by<br>
&gt;&gt;                           fuzzing with afl.  (Keith Winstein)<br>
&gt;&gt;                         * Fix a server hang with XON/XOFF on BSD<br>
&gt;&gt;                     systems.  (John Hood)<br>
&gt;&gt;                         * Fix a typeahead-prediction bug that caused<br>
&gt;&gt;                     display corruption on<br>
&gt;&gt;                           urxvt.  (John Hood)<br>
&gt;&gt;<br>
&gt;&gt;                     Source code is available as<br>
</div></div>&gt;&gt;                     &lt;&lt;<a href="https://github.com/mobile-shell/mosh/releases/download/untagged-0d43a0f73825e9952677/mosh-1.2.5.95rc1.tar.gz" rel="noreferrer" target="_blank">https://github.com/mobile-shell/mosh/releases/download/untagged-0d43a0f73825e9952677/mosh-1.2.5.95rc1.tar.gz</a>&gt;<a href="https://github.com/mobile-shell/mosh/releases/download/untagged-0d43a0f73825e9952677/mosh-1.2.5.95rc1.tar.gz" rel="noreferrer" target="_blank">https://github.com/mobile-shell/mosh/releases/download/untagged-0d43a0f73825e9952677/mosh-1.2.5.95rc1.tar.gz</a>&gt;.<br>
<span class="">&gt;&gt;                      The SHA256 sum for this file is<br>
&gt;&gt;                     a2697c41cfc8c92dc7a743dd101849a7a508c6986b24d6f44711d8533d18fcf5<br>
&gt;&gt;<br>
&gt;&gt;                     One standalone OS X package is available:<br>
&gt;&gt;<br>
&gt;&gt;                     *<br>
</span>&gt;&gt;                     &lt;&lt;<a href="https://github.com/mobile-shell/mosh/releases/download/untagged-0d43a0f73825e9952677/mosh-1.2.5.95rc1.pkg" rel="noreferrer" target="_blank">https://github.com/mobile-shell/mosh/releases/download/untagged-0d43a0f73825e9952677/mosh-1.2.5.95rc1.pkg</a>&gt;<a href="https://github.com/mobile-shell/mosh/releases/download/untagged-0d43a0f73825e9952677/mosh-1.2.5.95rc1.pkg" rel="noreferrer" target="_blank">https://github.com/mobile-shell/mosh/releases/download/untagged-0d43a0f73825e9952677/mosh-1.2.5.95rc1.pkg</a>&gt;<br>
<div><div class="h5">&gt;&gt;                     is an i386/x86_64 build for OS X 10.9 and higher.<br>
&gt;&gt;                     The SHA256 sum for<br>
&gt;&gt;                     this file is<br>
&gt;&gt;                     48a56d83d0ef655d38e0ea596fd9cac98c0dc433cb5356205d26748350d47e6c<br>
&gt;&gt;<br>
&gt;&gt;                     (If you are using a package system such as<br>
&gt;&gt;                     MacPorts or Homebrew, I<br>
&gt;&gt;                     recommend using that, though.)<br>
&gt;&gt;<br>
&gt;&gt;                     As always, Ubuntu PPA builds of the latest source<br>
&gt;&gt;                     are available at<br>
&gt;&gt;                     ppa:keithw/mosh.<br>
&gt;&gt;<br>
&gt;&gt;                     Packagers, please note that Mosh has some minor<br>
&gt;&gt;                     dependency changes: Perl<br>
&gt;&gt;                     is now required to be &gt;= 5.14, but IO::Socket<br>
&gt;&gt;                     modules are no longer<br>
&gt;&gt;                     required.  If anybody needs to package for older<br>
&gt;&gt;                     versions of Perl, talk<br>
&gt;&gt;                     to me and I&#39;ll probably bring something into the<br>
&gt;&gt;                     release.  Also, if<br>
&gt;&gt;                     anyone needs an OS X package for 10.8 or lower,<br>
&gt;&gt;                     please contact me.<br>
&gt;&gt;<br>
&gt;&gt;                     Your testing is very unlikely to prove Mosh to be<br>
&gt;&gt;                     free of bugs, but your<br>
&gt;&gt;                     testing will help us make 1.2.6 a better release.<br>
&gt;&gt;                     Please report any<br>
&gt;&gt;                     issues you find on Github, and we can be found on<br>
&gt;&gt;                     IRC at<br>
&gt;&gt;                     &lt;irc://<a href="http://irc.freenode.net/mosh" rel="noreferrer" target="_blank">irc.freenode.net/mosh</a><br>
</div></div>&gt;&gt;                     &lt;<a href="http://irc.freenode.net/mosh" rel="noreferrer" target="_blank">http://irc.freenode.net/mosh</a>&gt;&gt;.<br>
<span class="">&gt;&gt;<br>
&gt;&gt;                     Looking ahead, we expect Mosh 1.3 to be a feature<br>
&gt;&gt;                     release, bringing<br>
&gt;&gt;                     significant new functionality.  SSH agent<br>
&gt;&gt;                     forwarding is high on the list.<br>
&gt;&gt;<br>
&gt;&gt;                     The Mosh team thanks you for your help.<br>
&gt;&gt;<br>
&gt;&gt;                       --John Hood<br>
&gt;&gt;<br>
&gt;&gt;<br>
&gt;&gt;                     _______________________________________________<br>
&gt;&gt;<br>
&gt;&gt;                     mosh-devel mailing list<br>
</span>&gt;&gt;                     <a href="mailto:mosh-devel@mit.edu">mosh-devel@mit.edu</a> &lt;mailto:<a href="mailto:mosh-devel@mit.edu">mosh-devel@mit.edu</a>&gt;<br>
<span class="">&gt;&gt;                     <a href="http://mailman.mit.edu/mailman/listinfo/mosh-devel" rel="noreferrer" target="_blank">http://mailman.mit.edu/mailman/listinfo/mosh-devel</a><br>
&gt;&gt;<br>
&gt;&gt;<br>
&gt;&gt;<br>
&gt;&gt;<br>
&gt;&gt;     _______________________________________________<br>
&gt;&gt;     mosh-devel mailing list<br>
</span>&gt;&gt;     <a href="mailto:mosh-devel@mit.edu">mosh-devel@mit.edu</a> &lt;mailto:<a href="mailto:mosh-devel@mit.edu">mosh-devel@mit.edu</a>&gt;<br>
<div class=""><div class="h5">&gt;&gt;     <a href="http://mailman.mit.edu/mailman/listinfo/mosh-devel" rel="noreferrer" target="_blank">http://mailman.mit.edu/mailman/listinfo/mosh-devel</a><br>
&gt;&gt;<br>
&gt;&gt;<br>
&gt;&gt;<br>
&gt;&gt;<br>
&gt;&gt; _______________________________________________<br>
&gt;&gt; mosh-devel mailing list<br>
&gt;&gt; <a href="mailto:mosh-devel@mit.edu">mosh-devel@mit.edu</a><br>
&gt;&gt; <a href="http://mailman.mit.edu/mailman/listinfo/mosh-devel" rel="noreferrer" target="_blank">http://mailman.mit.edu/mailman/listinfo/mosh-devel</a><br>
&gt;<br>
<br>
</div></div></blockquote></div><br></div></div>