Move coop cloud content under "uncategorised"

docs/uncategorised/coop-cloud/logging.md

@@ -0,0 +1,48 @@
+# Logging with systemd-journald
+
+The default Docker swarm logging driver is `json-file`, which is nice but
+whenever a container is kiled or replaced, the logs are lost. This doesn't help
+us when we want to look back and try to diagnose what hapened.
+
+So, we change the default logging driver to use the systemd based journal
+logging system. This is configured in the `/etc/systemd/journal.conf` and in
+the `/etc/docker/daemon.json`. In practice, it means that logs are persistently
+stored after containers go away, they are rotated and can be analysed later on.
+
+`/etc/docker/daemon.json`:
+
+```
+{
+    "log-driver": "journald",
+    "log-opts": {
+      "labels":"com.docker.swarm.service.name"
+    }
+}
+```
+
+`/etc/systemd/journal.conf`:
+
+```
+[Journal]
+Storage=persistent
+SystemMaxUse=5G
+MaxFileSec=1month
+
+```
+
+There is also the [official docker
+documentation](https://docs.docker.com/config/containers/logging/journald/) on
+the journald logging driver.
+
+Some useful commands:
+
+- `journalctl -f`
+- `journalctl CONTAINER_NAME=gitea_gitea.1.jxn9r85el63pdz42ykjnmh792 -f`
+- `journalctl COM_DOCKER_SWARM_SERVICE_NAME=gitea_gitea --since="2020-09-18 13:00:00" --until="2020-09-18 13:01:00"`
+- `journalctl CONTAINER_ID=$(docker ps -qf name=gitea_gitea) -f`
+
+Also, for more system wide analysis stuff:
+
+- `journalctl --disk-usage`
+- `du -sh /var/log/journal/*`
+- `man journalctl` / `man systemd-journald` / `man journald.conf`