All Cloudron containers need to have the nameserver 172.18.0.1. This was
being done at the daemon level, however since there are also iptables
rules restricting access to the nameserver from containers that aren't
on the Cloudron Docker network, this broke DNS for non-Cloudron
containers.
Since the DNS is only required for Cloudron containers in the first
place, this patch specifies 172.18.0.1 as the nameserver when Cloudron
creates a container and reverts the change at the daemon level
The core issue we want to solve is to debug a running app.
Let's make it explicit that it is in debugging mode because
functions like update/backup/restore don't work.
Part of #171
When turned off, it will put the app in a writable rootfs. This
allows us to debug live/production apps (like change start.sh) and
just get them up and running. Once turned off, this app cannot be
updated anymore (unless the force flag is set). This way we can
then update it using the CLI if we are convinced that the upcoming
update fixes the problem.
Part of #171
do no set hostname of containers to location as it might conflict with addons names. for example, an app installed in mail
location may not reach mail container anymore by DNS. We cannot set hostname to fqdn either as that sets up the dns
name to look up the internal docker ip. this makes curl from within container fail
Note that Hostname has no effect on DNS. We have to use the --net-alias for dns.
Hostname cannot be set with container NetworkMode
The main issue is that multiple cloudron exec sessions do not
share the same rootfs. Which makes it annoying to debug.
We also have some nginx timeout which drops you out of exec
now and then resulting in loss of all state.
scheduler.json gets nuked during updates. When the box code restarts,
the scheduler is unable to remove old container because the state file
scheduler.json is now gone. It proceeds to create new container but that
does not work because of name conflict.
Fixes#531
Dec 01 08:36:53 girish.cloudron.us node[5431]: Error: HTTP code is 409 which indicates error: undefined - Conflicting options: --net=container can't be used with links. This would result in undefined behavior
Some apps like pasteboard try to curl the public app url from inside
the container. This fails because we set the hostname and the hostname
maps to the internal docker IP.
To fix this, simply export two environment variables providing the
app's domain and origin. The hostname is set to the app location instead
of the FQDN for debugging.
Fixes#521
