Support for ipv6 addresses (with or without port specified) #536
Conversation
Codecov ReportBase: 71.21% // Head: 70.91% // Decreases project coverage by
Additional details and impacted files@@ Coverage Diff @@
## master #536 +/- ##
==========================================
- Coverage 71.21% 70.91% -0.30%
==========================================
Files 49 49
Lines 3071 3074 +3
==========================================
- Hits 2187 2180 -7
- Misses 884 894 +10
Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here. ☔ View full report at Codecov. |
|
This PR is in great condition. Can you modify the connect_to_initial_version circleci test to use ipv6 when connecting to the initial version? Don't modify the address going the other way because version 6.0.0 won't have your commit. |
|
Sure! Note that this does not support ipv6 zero reduction (i.e. 0:0:0:0:0:0:0:1 -> ::1. I'll file a task to implement that. |
jshort
force-pushed
the
master
branch
3 times, most recently
from
26c9b06 to
e77190d
Compare
|
Pushed a temp debug config for circle ci to see if ListenAddress :: is in the sshd config(s). EDIT: Need to add inet6 to this:
EDIT2: That didn't work. Uncommented both ListenAddress 0.0.0.0 and ListenAddress :: to no avail: Output of Looks like ssh isn't running for ipv6. |
jshort
force-pushed
the
master
branch
17 times, most recently
from
69ee277 to
160fec9
Compare
|
Ok I got it working. Sorry for those 16 force pushes! I was running every tool I could think of to see why sshd was rejecting the ipv6 connection: netstat (gone in this version of Ubuntu), ss, ufw, ip(6)tables, lsof, ip addr, etc. Turns out that sshd was running by default so the |
Currently et does not support passing an ipv6 address as the host positional argument is assumed to be an ipv4 address or hostname with a single colon with the port after it and the rest is ignored, eg: ``` $ et 2620:10d:c083:1418:14ac:6bbc:4737:56ee 15:19:52 10.07.22 100% █ Could not reach the ET server: 2620:10 ``` This change will support ipv4 address or hostname with port specified or ipv6 address with or without port specified. Currently, if a port is specified in the positional arg, it overwrites what was specified (explicitly or by default) with the -p,--port option. I think this is backwards and may offer a PR to invert this logic.
No-op changes: * TerminalMain: change binary name/description * TerminalMain: get rid of Daemon log file (does nothing now) * SubprocessToString: reorganize forking logic and remove irrelevant comments and cleanup trailing whitespace * PsuedoUserTerminal: breaks in forking switch statement
The prefix option to the terminal client is misleading as it isn't a prefix (or search path) and is treated as an absolute path to the etterminal executable. This renames it as such and removes the superfluous positional arg in the command string building for the ssh command (i.e. /usr/local/bin/etterminal etterminal -v 0)
|
Great! Thanks so much for this high quality PR. |
Currently et does not support passing an ipv6 address as the host positional argument is assumed to be an ipv4 address or hostname with a single colon with the port after it and the rest is ignored, eg:
This change will support ipv4 address or hostname with port specified or ipv6 address with or without port specified. Currently, if a port is specified in the positional arg, it overwrites what was specified (explicitly or by default) with the -p,--port option. I think this is backwards and may offer a PR to invert this logic.