I’ve got a controller on lxd locally. This morning I noticed that Juju has the wrong idea what the controller’s IPs are. Not sure what happened but perhaps the container got restarted.
This is the controller:
$ lxc list
...
+------------------+---------+----------------------+-------------------------------------------+-----------------------+-----------+
| juju-568192-0 | RUNNING | 192.168.1.43 (eth0) | fd00::1:216:3eff:fe1a:b851 (eth0) | CONTAINER | 0 |
| | | | another.ipv6.address (eth0) | | |
+------------------+---------+----------------------+-------------------------------------------+-----------------------+-----------+
and this is what juju thinks it needs to hit
$ juju --debug status ~
10:02:05 INFO juju.cmd supercommand.go:91 running juju [2.8-rc3 0 7af5ef814c95d4ea13d3039ee1f59763c52cc5d6 gc go1.14.3]
10:02:05 DEBUG juju.cmd supercommand.go:92 args: []string{"/snap/juju/12214/bin/juju", "--debug", "status"}
10:02:05 INFO juju.juju api.go:67 connecting to API addresses: [[fd00::1:216:3eff:fe2d:128b]:17070 [some.other.ipv6.address.in.my.prefix]:17070 192.168.1.98:17070]
^C
I broke into the controller using lxc shell and tried to add 192.168.1.98 as an address manually. This worked in as much as I could telnet to it from the outside, but it didn’t fix any juju commands for some reason.
$ telnet 192.168.1.98 17070 ~
Trying 192.168.1.98...
Connected to 192.168.1.98.
Escape character is '^]'.
Any way I can fix up juju’s idea of what reality is?