Reverts b1b4ba501f, 90c9801c56, a3c84f78ca, ..
I haven't really traced it (yet), but on some servers, I'm observing
`ansible-playbook ... --tags=start` completing very slowly, waiting
to stop services. I can't reproduce this on all Matrix servers I manage.
I suspect that either the systemd version is to blame or that some
specific service is not responding well to some `docker kill/rm` command.
`ExecStop` seems to work great in all cases and it's what we've been
using for a very long time, so I'm reverting to that.
4.95-r0-1 was problematic, because `/etc/exim/exim.conf` in the
container had the wrong permissions (writable by the `exim` user).
Fixed in 697f3cff7e
which is built as 4.95-r0-2
4.95-r0-1 is the first container image tag that is available as a multi-arch image
with support for linux/amd64, linux/arm64/v8 (arm64) and linux/arm/v7 (arm32),
so self-building is no longer necessary on all these platforms.
Until now, we were leaving services "enabled"
(symlinks in /etc/systemd/system/multi-user.target.wants/).
We clean these up now. Broken symlinks may still exist in older
installations that enabled/disabled services. We're not taking care
to fix these up. It's just a cosmetic defect anyway.
These are just defensive cleanup tasks that we run.
In the good case, there's nothing to kill or remove, so they trigger an
error like this:
> Error response from daemon: Cannot kill container: something: No such container: something
and:
> Error: No such container: something
People often ask us if this is a problem, so instead of always having to
answer with "no, this is to be expected", we'd rather eliminate it now
and make logs cleaner.
In the event that:
- a container is really stuck and needs cleanup using kill/rm
- and cleanup fails, and we fail to report it because of error
suppression (`2>/dev/null`)
.. we'd still get an error when launching ("container name already in use .."),
so it shouldn't be too hard to investigate.
Starting with Docker 20.10, `--hostname` seems to have the side-effect
of making Docker's internal DNS server resolve said hostname to the IP
address of the container.
Because we were giving the mailer service a hostname of `matrix.DOMAIN`,
all requests destined for `matrix.DOMAIN` originating from other
services on the container network were resolving to `matrix-mailer`.
This is obviously wrong.
Initially reported here: https://github.com/spantaleev/matrix-docker-ansible-deploy/pull/748
We normally try to not use the public hostname (and IP address) on the
container network and try to make services talk to one another locally,
but it sometimes could happen.
With this, we use a `matrix-mailer` hostname for the matrix-mailer
container. My testing shows that it doesn't cause any trouble with
email deliverability.
The Docker 19.04 -> 20.10 upgrade contains the following change
in `/usr/lib/systemd/system/docker.service`:
```
-BindsTo=containerd.service
-After=network-online.target firewalld.service containerd.service
+After=network-online.target firewalld.service containerd.service multi-user.target
-Requires=docker.socket
+Requires=docker.socket containerd.service
Wants=network-online.target
```
The `multi-user.target` requirement in `After` seems to be in conflict
with our `WantedBy=multi-user.target` and `After=docker.service` /
`Requires=docker.service` definitions, causing the following error on
startup for all of our systemd services:
> Job matrix-synapse.service/start deleted to break ordering cycle starting with multi-user.target/start
A workaround which appears to work is to add `DefaultDependencies=no`
to all of our services.
Fixes https://github.com/spantaleev/matrix-docker-ansible-deploy/issues/716
This patch makes us use more fully-qualified container image names
(either prefixed with docker.io/ or with localhost/).
The latter happens when self-building is enabled.
We've recently had issues where if an image was removed manually
and the service was restarted (making `docker run` fetch it from Docker Hub, etc.),
we'd end up with a pulled image, even though we're aiming for a self-built one.
Re-running the playbook would then not do a rebuild, because:
- the image with that name already exists (even though it's something
else)
- we sometimes had conditional logic where we'd build only if the git
repo changed
By explicitly changing the name of the images (prefixing with localhost/),
we avoid such confusion and the possibility that we'd automatically pul something
which is not what we expect.
Also, I've removed that condition where building would happen on git
changes only. We now always build (unless an image with that name
already exists). We just force-build when the git repo changes.
Depending on the distro, common commands like sleep and chown may either
be located in /bin or /usr/bin.
Systemd added path lookup to ExecStart in v239, allowing only the
command name to be put in unit files and not the full path as
historically required. At least Ubuntu 18.04 LTS is however still on
v237 so we should maintain portability for a while longer.
It adds support for a new `DISABLE_SENDER_VERIFICATION` environment
variable that can be used to disable verification of sender addresses.
It doesn't matter for us, but we upgrade to keep up with latest.
We do use some `:latest` images by default for the following services:
- matrix-dimension
- Goofys (in the matrix-synapse role)
- matrix-bridge-appservice-irc
- matrix-bridge-appservice-discord
- matrix-bridge-mautrix-facebook
- matrix-bridge-mautrix-whatsapp
It's terribly unfortunate that those software projects don't release
anything other than `:latest`, but that's how it is for now.
Updating that software requires that users manually do `docker pull`
on the server. The playbook didn't force-repull images that it already
had.
With this patch, it starts doing so. Any image tagged `:latest` will be
force re-pulled by the playbook every time it's executed.
It should be noted that even though we ask the `docker_image` module to
force-pull, it only reports "changed" when it actually pulls something
new. This is nice, because it lets people know exactly when something
gets updated, as opposed to giving the indication that it's always
updating the images (even though it isn't).
Reasoning is the same as for matrix-org/synapse#5023.
For us, the journal used to contain `docker` for all services, which
is not very helpful when looking at them all together (`journalctl -f`).
We run containers as a non-root user (no effective capabilities).
Still, if a setuid binary is available in a container image, it could
potentially be used to give the user the default capabilities that the
container was started with. For Docker, the default set currently is:
- "CAP_CHOWN"
- "CAP_DAC_OVERRIDE"
- "CAP_FSETID"
- "CAP_FOWNER"
- "CAP_MKNOD"
- "CAP_NET_RAW"
- "CAP_SETGID"
- "CAP_SETUID"
- "CAP_SETFCAP"
- "CAP_SETPCAP"
- "CAP_NET_BIND_SERVICE"
- "CAP_SYS_CHROOT"
- "CAP_KILL"
- "CAP_AUDIT_WRITE"
We'd rather prevent such a potential escalation by dropping ALL
capabilities.
The problem is nicely explained here: https://github.com/projectatomic/atomic-site/issues/203
This makes all containers (except mautrix-telegram and
mautrix-whatsapp), start as a non-root user.
We do this, because we don't trust some of the images.
In any case, we'd rather not trust ALL images and avoid giving
`root` access at all. We can't be sure they would drop privileges
or what they might do before they do it.
Because Postfix doesn't support running as non-root,
it had to be replaced by an Exim mail server.
The matrix-nginx-proxy nginx container image is patched up
(by replacing its main configuration) so that it can work as non-root.
It seems like there's no other good image that we can use and that is up-to-date
(https://hub.docker.com/r/nginxinc/nginx-unprivileged is outdated).
Likewise for riot-web (https://hub.docker.com/r/bubuntux/riot-web/),
we patch it up ourselves when starting (replacing the main nginx
configuration).
Ideally, it would be fixed upstream so we can simplify.
