Skip to content

Latest commit

 

History

History
207 lines (157 loc) · 6.95 KB

README.md

File metadata and controls

207 lines (157 loc) · 6.95 KB
Raw

Using InsightOps?

Please use our updated library which supports specifying the region to send your log data to.

docker-logentries

Forward all your logs to Logentries, like a breeze.

logentries dashboard

See the Logentries community pack at http://revelops.com/community/packs/docker/.

Usage as a Container

The simplest way to forward all your container's log to Logentries is to run this repository as a container, with:

docker run -v /var/run/docker.sock:/var/run/docker.sock \
           --read-only \
           --security-opt=no-new-privileges \
           logentries/docker-logentries \
           -t <TOKEN> \
           -j \
           -a host=`uname -n`

You can also use different tokens for logging, stats and events:

docker run -v /var/run/docker.sock:/var/run/docker.sock \
           --read-only \
           --security-opt=no-new-privileges \
           logentries/docker-logentries \
           -l <LOGSTOKEN> \
           -k <STATSTOKEN> \
           -e <EVENTSTOKEN> \
           -j \
           -a host=`uname -n`

The --read-only docker flag specifies that the container file system will be read-only. This is not a requirement but since currently there's no need for writing, it makes the container more secure.

The --security-opt=no-new-privileges docker flag sets a kernel bit which stops the process or its children from gaining additional privileges via setuid or sgid.
Once again not required, but increases security.

You can pass the --no-stats flag if you do not want stats to be published to Logentries every second. You need this flag for Docker version < 1.5.

You can pass the --no-logs flag if you do not want logs to be published to Logentries.

You can pass the --no-dockerEvents flag if you do not want events to be published to Logentries.

The -i/--statsinterval <STATSINTERVAL> downsamples the logs sent to Logentries. It collects samples and averages them before sending to Logentries.

If you don't use -a a default host=`uname -n` value will be added.

You can also filter the containers for which the logs/stats are forwarded with:

  • --matchByName REGEXP: forward logs/stats only for the containers whose name matches the given REGEXP.
  • --matchByImage REGEXP: forward logs/stats only for the containers whose image matches the given REGEXP.
  • --skipByName REGEXP: do not forward logs/stats for the containers whose name matches the given REGEXP.
  • --skipByImage REGEXP: do not forward logs/stats for the containers whose image matches the given REGEXP.

Running container in a restricted environment.

Some environments(such as Google Compute Engine) does not allow to access the docker socket without special privileges. You will get EACCES(Error: read EACCES) error if you try to run the container. To run the container in such environments add --privileged to the docker run command.

Example:

docker run --privileged \
           -v /var/run/docker.sock:/var/run/docker.sock \
           --read-only \
           --security-opt=no-new-privileges \
           logentries/docker-logentries \
           -t <TOKEN> \
           -j \
           -a host=`uname -n`

Usage as a CLI

  1. npm install docker-logentries -g
  2. docker-logentries -t TOKEN -a host=\uname -n``
  3. ..there is no step 3

You can also pass the -j switch if you log in JSON format, like bunyan.

You can pass the --no-stats flag if you do not want stats to be published to Logentries every second.

You can pass the --no-logs flag if you do not want logs to be published to Logentries.

You can pass the --no-dockerEvents flag if you do not want events to be published to Logentries.

The -a/--add flag allows to add fixed values to the data being published. This follows the format 'name=value'.

The -i/--statsinterval downsamples the logs sent to Logentries. It collects samples and averages them before sending to Logentries.

You can also filter the containers for which the logs/stats are forwarded with:

  • --matchByName REGEXP: forward logs/stats only for the containers whose name matches the given REGEXP.
  • --matchByImage REGEXP: forward logs/stats only for the containers whose image matches the given REGEXP.
  • --skipByName REGEXP: do not forward logs/stats for the containers whose name matches the given REGEXP.
  • --skipByImage REGEXP: do not forward logs/stats for the containers whose image matches the given REGEXP.

Embedded usage

Install it with: npm install docker-logentries --save

Then, in your JS file:

var logentries = require('docker-logentries')({
  json: false, // or true to parse lines as JSON
  secure: true, // or false to connect over plain TCP
  token: process.env.TOKEN, // logentries TOKEN
  newline: true, // Split on newline delimited entries
  stats: true, // disable stats if false
  add: null, // an object whose properties will be added

  // the following options limit the containers being matched
  // so we can avoid catching logs for unwanted containers
  matchByName: /hello/, // optional
  matchByImage: /matteocollina/, //optional
  skipByName: /.*pasteur.*/, //optional
  skipByImage: /.*dockerfile.*/ //optional
})

// logentries is the source stream with all the
// log lines

setTimeout(function() {
  logentries.destroy()
}, 5000)

Building a docker repo from this repository

Using the plain docker file

First clone this repository, then:

docker build -t logentries .

docker run -v /var/run/docker.sock:/var/run/docker.sock \
           --read-only \
           --security-opt=no-new-privileges \
           logentries \
           -t <TOKEN> \
           -j \
           -a host=`uname -n`

Using Make - the official nodejs onbuild image

export BUILD_TYPE=node-onbuild
make build
make test
make tag

Using Make - the alpine linx build (~42Mb)

export BUILD_TYPE=alpine-node
make build
make test
make tag

Pushing to your own repo

After you've build, tested, tagged it locally

export DOCKER_REGISTRY_PREFIX=you-dockerhub-user/yourimage-name
make push

How it works

This module wraps four Docker APIs:

  • POST /containers/{id}/attach, to fetch the logs
  • GET /containers/{id}/stats, to fetch the stats of the container
  • GET /containers/json, to detect the containers that are running when this module starts
  • GET /events, to detect new containers that will start after the module has started

This module wraps docker-loghose and docker-stats to fetch the logs and the stats as a never ending stream of data.

All the originating requests are wrapped in a never-ending-stream.

License

MIT