logspout

logspout

A log router for Docker container output that runs entirely inside Docker. It attaches to all containers on a host, then routes their logs wherever you want.

It's a 100% stateless log appliance (unless you persist routes). It's not meant for managing log files or looking at history. It is just a means to get your logs out to live somewhere else, where they belong.

For now it only captures stdout and stderr, but soon Docker will let us hook into more ... perhaps getting everything from every container's /dev/log.

Getting logspout

Logspout is a very small Docker container (14MB virtual, based on busybox), so you can just pull it from the index:

$ docker pull progrium/logspout

Using logspout

Route all container output to remote syslog

The simplest way to use logspout is to just take all logs and ship to a remote syslog. Just pass a default syslog target URI as the command. Also, we always mount the Docker Unix socket with -v to /tmp/docker.sock:

$ docker run -v=/var/run/docker.sock:/tmp/docker.sock progrium/logspout syslog://logs.papertrailapp.com:55555

Logs will be tagged with the container name. The hostname will be the hostname of the logspout container, so you probably want to set the container hostname to the actual hostname by adding -h $HOSTNAME.

Inspect log streams using curl

Whether or not you run it with a default routing target, if you publish its port 8000, you can connect with curl to see your local aggregated logs in realtime.

$ docker run -d -p 8000:8000 \
	-v=/var/run/docker.sock:/tmp/docker.sock \
	progrium/logspout
$ curl $(docker port `docker ps -lq` 8000)/logs

You should see a nicely colored stream of all your container logs. You can filter by container name, log type, and more. You can also get JSON objects, or you can upgrade to WebSocket and get JSON logs in your browser.

See Streaming Endpoints for all options.

Create custom routes via HTTP

Along with streaming endpoints, logspout also exposes a /routes resource to create and manage routes.

$ curl $(docker port `docker ps -lq` 8000)/logs -X POST \
	-d '{"source": {"filter": "db", "types": ["stderr"]}, target": {"type": "syslog", "addr": "logs.papertrailapp.com:55555"}}'

That example creates a new syslog route to Papertrail of only stderr for containers with db in their name.

By default, routes are ephemeral. But if you mount a volume to /mnt/routes, they will be persisted to disk.

See Routes Resource for all options.

HTTP API

Streaming Endpoints

You can use these chunked transfer streaming endpoints for quick debugging with curl or for setting up easy TCP subscriptions to log sources. They also support WebSocket upgrades.

GET /logs
GET /logs/filter:<container-name-substring>
GET /logs/id:<container-id>
GET /logs/name:<container-name>

You can select specific log types from a source using a comma-delimited list in the query param types. Right now the only types are stdout and stderr, but when Docker properly takes over each container's syslog socket (or however they end up doing it), other types will be possible.

If you include a request Accept: application/json header, the output will be JSON objects including the name and ID of the container and the log type. Note that when upgrading to WebSocket, it will always use JSON.

Since /logs and /logs/filter:<string> endpoints can return logs from multiple source, they will by default return color-coded loglines prefixed with the name of the container. You can turn off the color escape codes with query param colors=off or the alternative is to stream the data in JSON format, which won't use colors or prefixes.

Routes Resource

Routes let you configure logspout to hand-off logs to another system. Right now the only supported target type is via UDP syslog, but hey that's pretty much everything.

Creating a route

POST /routes

Takes a JSON object like this:

{
	"source": {
		"filter": "_db"
		"types": ["stdout"]
	},
	"target": {
		"type": "syslog",
		"addr": "logaggregator.service.consul"
		"append_tag": ".db"
	}
}

The source field should be an object with filter, name, prefix, or id fields. prefix allows a string match against the start of a container name (e.g. "frontend" will match containers named like "frontend-1"). You can specify specific log types with the types field to collect only stdout or stderr. If you don't specify types, it will route all types.

To route all logs of all types on all containers, don't specify a source.

The append_tag field of target is optional and specific to syslog. It lets you append to the tag of syslog packets for this route. By default the tag is <container-name>, so an append_tag value of .app would make the tag <container-name>.app.

And yes, you can just specify an IP and port for addr, but you can also specify a name that resolves via DNS to one or more SRV records. That means this works great with Consul for service discovery.

Listing routes

GET /routes

Returns a JSON list of current routes:

[
	{
		"id": "3631c027fb1b",
		"source": {
			"name": "mycontainer"
		},
		"target": {
			"type": "syslog",
			"addr": "192.168.1.111:514"
		}
	}
]

Viewing a route

GET /routes/<id>

Returns a JSON route object:

{
	"id": "3631c027fb1b",
	"source": {
		"id": "a9efd0aeb470"
		"types": ["stderr"]
	},
	"target": {
		"type": "syslog",
		"addr": "192.168.1.111:514"
	}
}

Deleting a route

DELETE /routes/<id>

Sponsor

This project was made possible by DigitalOcean.

License

BSD