* TimeZone: autoinstall and API
Add support for Get/Set timezone methods. Get means that we inquire
with GeoIP as to which timezone is suggested. Non-availability of
GeoIP, or a previous explicit Set, means that we return the system
timezone. Set of timezone by Post results in set of the live system
timzeone, and queuing a set of the target system by way of cloud-init.
* Add clarifying comment about _request.
Move mock_app to common location.
Move run_coro to subiquitycore so that subiquitycore doesn't have to
reference things in subiquity, even for test.
Move task tracking things from mirror to geoip.
Server app owns the geoip instance.
Create EventCallback as an alternative to MessageHub that should
hopefully express clearer intermodule dependencies.
* locale - let it check interactive-sections again
* Turn Serial into a whole new screen
When in serial, first offer the rich/basic choice (or SSH button),
and only show the current welcome screen if we choose rich mode.
Add a back button on Welcome if we are on serial.
LP: #1919251
This fixes a problem where you drop to a shell and refresh subiquity
from that shell -- the client tries to restart but it is running in the
background and so crashes trying to modify the terminal settings. So
this kills the subprocess before restarting. This required the extremely
angry PR I sent before: forcefully killing the subprocess also crashes
the client before restart in a similar way.
Here is something you can try with a live server installer today: drop
into a subshell and type "kill -9 $$". The installer will crash with an
'Input/output error' from tcsetattr.
What's going on is some deep unix arcana: when the subshell exits via
signal somehow the shell's process group is still in the foreground
(even though there are no processes in it?) and calling tcsetattr() from
a background process is not allowed. So the fix for this is reasonably
simple (or at least: short): call tcsetpgrp() to put our process group
back into the foreground. A background process doing this gets sent
SIGTTOU, but if we ignore that, all is well.
Frankly this is all very strange and I would like a lie down now.
Splitting subiquity into server and client means that in general
old versions of the client can still be running when the server is
updated (the client running on tty1 will be restarted by snapd/systemd
when the snap is updated but clients running via e.g. ssh will not). I
implemented a way for the client to detect this and restart itself: the
server sets a header in all responses that indicates if it has been
updated. So far so good. But the way it knows that it has been updated
is to check the presence of a file that is only created when subiquity
itself triggers the refresh, so it's not there in the case of manual
refresh, and as reported in https://bugs.launchpad.net/bugs/1921820 this
can lead to the client crashing because it cannot parse the new server's
response. This simply changes to creating the marker file in the snap
post-refresh hook, which will be executed for manual snap refreshes as
well.
While I'm at it, remove the rest of the post-install hook that restarted
subiquity clients running on the serial line as the generic machinery
will work for these too.
The log file names have pids in now, but when subiquity re-execs itself
to fake a snap refresh the pid doesn't change. Having the pre "refresh"
logs get overwritten does not help anything and is sometimes very
annoying.