devops4me/fluentd

By devops4me

•Updated almost 7 years ago

A fluentd logstash container for docker logs that can forward to both elasticsearch and s3 buckets.

Image

911

Overview Tags

devops4me/fluentd repository overview

⁠The primary use case involves containerized apps using a fluentd docker log-driver to push logs to a fluentd container that in turn forwards them to an elasticsearch instance. The secondary use case is visualizing the logs via a Kibana container linked to elasticsearch.

⁠Send Docker Logs to Fluentd and Elasticsearch

fluentd will pump logs from docker containers to an elasticsearch database. These logs can then be viewed via a docker kibana user interface that reads from the elasticsearch database. With this plan you

run an elasticsearch docker container
run a kibana docker container
run a fluentd (logstash) docker container
use docker's fluentd log-driver switch to run a container
login to the kibana ui to visualize the logs

⁠1 | docker run elasticsearch

docker run --detach --rm             \
    --name elastic-db                \
    --publish 9200:9200              \
    --publish 9300:9300              \
    --env discovery.type=single-node \
    --env transport.host=127.0.0.1   \
    --env ELASTIC_PASSWORD=secret    \
    docker.elastic.co/elasticsearch/elasticsearch-platinum:6.0.0 && sleep 20

The sleep commands give the containers breathing space before client connections are made.

⁠2 | docker run kibana

⁠@todo - attempt setting ( --network host ) and then eradicate the elastic-db container name which is then replaceable by localhost

⁠@todo - try removing the legacy link option switch if the above --network host is successful

⁠@todo - try updating the versions of both elasticsearch and kibana

docker run --detach --rm \
    --name kibana-ui     \
    --publish 5601:5601  \
    --link elastic-db    \
    --env "ELASTICSEARCH_URL=http://elastic-db:9200" \
    --env ELASTICSEARCH_PASSWORD=secret              \
    docker.elastic.co/kibana/kibana:6.0.0 && sleep 20

Use http://localhost:5601 with username elastic and password secret to access Kibana.

⁠3 | docker run fluentd (logstash)

Within the Dockerfile⁠ for devops4me/fluentd every command

either installs a fluentd plugin
or it copies in a configuration template

Select the configuration file that suits your needs and reference it in the FLUENTD_CONF environment variable in docker run along with other options documented in the tables below.

⁠fluentd to elasticsearch | http

Fluentd/logstash sends its logs to an elasticsearch instance configured with a username password.

docker run --interactive --tty                           \
    --name fluentd.es.logs                               \
    --network host                                       \
    --publish 24224:24224                                \
    --env FLUENTD_CONF=fluentd-simple.conf \
    --env ELASTICSEARCH_HOSTNAME=<<hostname>             \
    --env ELASTICSEARCH_PORT=<<port>>                    \
    devops4me/fluentd

Environment Variable	Mandatory?	Fluentd Configuration Explained
FLUENTD_CONF	Mandatory	Always use `fluentd-simple.conf`
ELASTICSEARCH_HOSTNAME	Mandatory	Hostname, url or IP Address
ELASTICSEARCH_PORT	Mandatory	Usually 9200 or 443 (https) and 80 for (http)
ELASTICSEARCH_SCHEME	Optional	Defaults to `http` but you can pass in `https`

⁠simple elasticsearch| example docker run

docker run --interactive --tty                           \
    --name fluentd.simple.logs                           \
    --network host                                       \
    --publish 24224:24224                                \
    --env FLUENTD_CONF=fluentd-simple.conf \
    --env ELASTICSEARCH_HOSTNAME=localhost               \
    --env ELASTICSEARCH_PORT=9200                        \
    devops4me/fluentd

This localhost reaches elasticsearch if we use --network=host to run both the fluentd container as well as the docker containers that source the logs through the log-driver.

⁠s3 with AWS credentials | fluentd logs copied to S3 and elasticsearch | https

This configuration uses AWS credentials so you can post it from your laptop. The other one uses roles so the ec2 instance would be configured to access the bucket without credentials.

⁠BEWARE - the S3_BUCKET_REGION is against the S3 bucket and is not the same as `AWS_REGION` | `AWS_DEFAULT_REGION` which decide where infrastructure gets created.

To discover the S3_BUCKET_REGION you must go to the S3 bucket in the AWS console, click on it and at the top right you will see the region name. Use that to set the region ID in your fluentd docker run environment variable.

<source>
    type forward
    port 24224
    bind 0.0.0.0
</source>

<match *.*>
    @type stdout
</match>

<match **>
    @type copy

    <store>
        @type elasticsearch
        host "#{ENV['ELASTICSEARCH_HOSTNAME']}"
        port "#{ENV['ELASTICSEARCH_PORT']}"
        scheme "#{ENV['ELASTICSEARCH_SCHEME'] || 'http'}"
        logstash_format true
        logstash_prefix "#{ENV['ELASTICSEARCH_PREFIX'] || 'app.logs'}"
        type_name "#{ENV['ELASTICSEARCH_TYPE_NAME'] || 'log.msg'}"
    </store>

    <store>
        @type s3
        aws_key_id "#{ENV['S3_AWS_ACCESS_KEY']}"
        aws_sec_key "#{ENV['S3_AWS_SECRET_KEY']}"
        s3_bucket "#{ENV['S3_BUCKET_NAME']}"
        s3_region "#{ENV['S3_BUCKET_REGION']}"
        utc
	store_as "#{ENV['S3_BUCKET_FILE_TYPE'] || 'json'}"
        path logs.%Y.%m-%B/%d-%A/
        s3_object_key_format "%{path}%{time_slice}-#{Socket.gethostname}-%{index}.%{file_extension}"
        time_slice_format %Y.%m.%d-%H%M
    </store>

</match>

⁠fluentd logs copied to S3 and elasticsearch | https

Many production systems copy logs to both S3 and an elasticsearch kibana instance. This docker run triggers a fluentd container that knows

which host and port elasticsearch is listening on
which s3 bucket to send logs to
the S3 path and object (file) name format

There are no IAM user credentials here because your ec2 instances should have the correct roles to write into the S3 bucket.

docker run --interactive --tty                       \
    --name fluentd.s3.es.logs                        \
    --publish 24224:24224                            \
    --env FLUENTD_CONF=fluentd-elasticsearch-s3.conf \
    --env ELASTICSEARCH_HOSTNAME=<<hostname>         \
    --env ELASTICSEARCH_PORT=<<port>>                \
    --env ELASTICSEARCH_SCHEME=<<scheme>             \
    --env S3_BUCKET_NAME=<<s3-bucket-name>>          \
    --env S3_BUCKET_REGION=<<aws-region>>            \
    devops4me/fluentd

⁠fluentd-elasticsearch-s3.conf

With this config fluentd pushes out logs to an elasticsearch instance and an AWS S3 bucket.

Environment Variable	Mandatory?	Fluentd Configuration Explained
FLUENTD_CONF	Mandatory	Always use `fluentd-elasticsearch-s3.conf`
ELASTICSEARCH_HOSTNAME	Mandatory	Hostname, url or IP Address
ELASTICSEARCH_PORT	Mandatory	Usually 9200 or 443 (https) and 80 for (http)
ELASTICSEARCH_SCHEME	Optional	Defaults to `http` but you can pass in `https`
ELASTICSEARCH_PREFIX	Optional	elasticsearch document index name will be prefixed with this.
ELASTICSEARCH_TYPE_NAME	Optional	elasticsearch document type name.
S3_BUCKET_NAME	Mandatory	The name of the S3 bucket to copy logs to.
S3_BUCKET_REGION	Mandatory	The AWS region the bucket is in.
S3_BUCKET_FILE_TYPE	Optional	Default is json but you can also have gzip and txt.

⁠s3 and elasticsearch| example docker run

docker run --interactive --tty                       \
    --name fluentd.s3.es.logs                        \
    --network host                                   \
    --publish 24224:24224                            \
    --env FLUENTD_CONF=fluentd-elasticsearch-s3.conf \
    --env ELASTICSEARCH_HOSTNAME=<<es-hostname>>     \
    --env ELASTICSEARCH_PORT=443                     \
    --env ELASTICSEARCH_SCHEME=https                 \
    --env S3_BUCKET_NAME=<<s3-bucket-name>>          \
    --env S3_BUCKET_REGION=<<aws-region>>            \
    devops4me/fluentd

⁠4 | docker run --log-driver fluentd

You append --log-driver fluentd and --log-opt fluentd-address=localhost:24224 to the docker run of any container you wish to collect logs from. Docker will then push its stdout logs to our on-board fluentd / logstash collector.

docker run                \
    --name fluent.smoke.test \
    --log-driver fluentd  \
    ubuntu \
    echo "Early mornings and late nights are meat and drink for the best life $(date)."

Again the --network host switch (in both) allows us to access the fluentd (logstash) log collector without stating the precise ip address or hostname.

⁠query doc from elasticsearch

⁠`curl 'localhost:9200/_search?q=supercalifragilistic&pretty' -u 'elastic:secret'`

This query verifies that our supercalifragilistic log document is in the elasticsearch database. The response reads like this.

{
  "took" : 41,
  "timed_out" : false,
  "_shards" : {
    "total" : 12,
    "successful" : 12,
    "skipped" : 0,
    "failed" : 0
  },
  "hits" : {
    "total" : 1,
    "max_score" : 0.2876821,
    "hits" : [
      {
        "_index" : "logstash-2019.01.21",
        "_type" : "fluentd-es-type",
        "_id" : "lhZgbmgBUreNY9GGjvzF",
        "_score" : 0.2876821,
        "_source" : {
          "container_id" : "356b3b051426294eb666122e66e16924af1b18b49ab262f146d77d47f1cc1c1a",
          "container_name" : "/log.smoke.test",
          "source" : "stdout",
          "log" : "supercalifragilistic log to elasticsearch via fluentd on Mon 21 Jan 03:06:25 GMT 2019.",
          "@timestamp" : "2019-01-21T03:06:25.000000000+00:00"
        }
      }
    ]
  }
}

⁠5 | kibana visualize

The smoke test in step 4 used docker log-driver to write to our fluentd container which (after one minute due to near real time) makes the log available in the Kibana user interface. To view the log you

visit url http://localhost:5601
enter username elastic
enter password secret
click on Discover
create a logstash-* index pattern
find last 15 minutes at the top right
change to last 7 days

The very first time we need to create an index pattern. The fluentd container always sends data with indices beginning with logstash- and ending in a date. Set the time field to @timestamp. Once created set the index pattern as the default and select an appropriate time range.

Our supercalifragilistic log should appear. Open it up with the little arrow or choose to view it in JSON.

⁠kibana screenshot of log from docker container

The screenshot shows that our smoke test log is in the elasticsearch database and accessible via the Kibana visualization user interface.

⁠documentation | fluentd | elasticsearch | kibana | logstash

Aside from this wiki page, some excellent documentation exists out there to help you implement the unified logging layer pattern with the ELK stack.

⁠Appendix - Useful Commands

curl 'localhost:9200/logstash-2019.01.21/_doc/1?pretty' -u 'elastic:secret'
curl 'localhost:9200/_search?q=apollo&pretty' -u 'elastic:secret'
curl 'localhost:9200/_search?q=jenkins&pretty' -u 'elastic:secret'
curl 'localhost:9200/logstash-2019.01.21/_mapping/group?pretty' -u 'elastic:secret'
curl 'localhost:9200/logstash-2019.01.21/_mapping/group?pretty' -u 'elastic:secret'
curl 'localhost:9200/_search?q=smoke&pretty' -u 'elastic:secret'
curl 'localhost:9200/_search?q=jenkins&pretty' -u 'elastic:secret'
curl 'localhost:9200/_cat/indices?v&pretty' -u 'elastic:secret'

⁠`curl 'localhost:9200/_cat/indices?v&pretty' -u 'elastic:secret'`

health status index                           uuid                   pri rep docs.count docs.deleted store.size pri.store.size
yellow open   .triggered_watches              fK2iGqDQTWue7ZmQkT6t3A   1   1          0            0     16.2kb         16.2kb
yellow open   .monitoring-es-6-2019.01.21     t1a2c6SWQdCI9De3oufHWw   1   1       4084           24      1.9mb          1.9mb
yellow open   .watches                        qAt3cJKXTcq6o8HJUowMXw   1   1          5            0     33.4kb         33.4kb
yellow open   logstash-2019.01.21             oBk-QrcvRmi22DIOI7t77g   5   1          1            0        8kb            8kb
yellow open   .monitoring-kibana-6-2019.01.21 TpoKwbPbRTa7mix75KqxaQ   1   1        334            0      218kb          218kb
yellow open   .kibana                         3aV5WqlIQoyBMWsqmYEFJg   1   1          2            0     33.7kb         33.7kb
yellow open   .monitoring-alerts-6            0fa4o5CARvCyFdejexm4yA   1   1          1            0      6.5kb          6.5kb
yellow open   .watcher-history-6-2019.01.21   xn8EFUksTpOj0ZkDXA7dWw   1   1        406            0    562.8kb        562.8kb

curl -XPUT 'https://<<secure-elasticsearch-url>>:443/devops4me/clothing/1?pretty' -d '{
"name": "itsybitsyteenyweenyyellowpolkerdotbikini",
"organizer": "Apollo Akora"
}' -H 'Content-Type: application/json'

curl 'https://<<url>>:443/_search?q=itsybitsyteenyweenyyellowpolkerdotbikini&pretty'


curl -XPUT 'localhost:9200/get-together/group/1?pretty' -u 'elastic:secret' -d '{
"name": "Elasticsearch Denver",
"organizer": "Lee"
}' -H 'Content-Type: application/json'

curl -XPUT 'localhost:9200/get-together/group/1?pretty' -u 'elastic:secret' -d '{
"name": "Elasticsearch Denver",
"organizer": "Lee"
}' -H 'Content-Type: application/json'

curl 'localhost:9200/get-together/_mapping/group?pretty' -u 'elastic:secret'

curl 'localhost:9200/get-together/_search?q=*&pretty' -u 'elastic:secret'

curl 'localhost:9200/.monitoring-es-6-2019.01.19/_search?q=*&pretty' -u 'elastic:secret'

curl 'localhost:9200/.watcher-history-6-2019.01.19/_search?q=*&pretty' -u 'elastic:secret'

⁠Appendix - Count Documents in ElasticSearch

Count the number of documents in the elasticsearch database. Choose the command below and substitute the url and username / password if necessary.

curl "http://localhost:9200/_count" -u 'elastic:secret' | jq .
curl "http://localhost/_count?pretty"
curl -k "https://localhost/_count?pretty"
curl "http://localhost/_count?pretty" -u '<<username>>:<<password>>'
curl "http://localhost:9200/_count?pretty" -u '<<username>>:<<password>>'

⁠Appendix - Put Documents into ElasticSearch

Choose the command below and substitute url and username password if necessary.

⁠Put Without Username / Password

curl -XPUT http://localhost/film-index/film/1 -d \
'{
"director": "Burton, Tim",
"genre": ["Comedy","Sci-Fi"],
"year": 1996,
"actor": ["Jack Nicholson","Pierce Brosnan","Sarah Jessica Parker"],
"title": "Apollo 13"
}' \
-H 'Content-Type: application/json' | jq .

curl -XPUT http://localhost/song-index/song/1 -d \
'{
"director": "Michael Jackson",
"genre": ["Comedy","Sci-Fi"],
"year": 2001,
"actor": ["Jack Nicholson","Pierce Brosnan","Sarah Jessica Parker"],
"title": "Billi Jean"
}' \
-H 'Content-Type: application/json' | jq .

⁠Put With Username / Password

curl -XPUT http://localhost/song-index/song/1 -u '<<username>>:<<password>>' -d \
'{
"director": "Michael Jackson",
"genre": ["Comedy","Sci-Fi"],
"year": 2001,
"actor": ["Jack Nicholson","Pierce Brosnan","Sarah Jessica Parker"],
"title": "Billi Jean"
}' \
-H 'Content-Type: application/json' | jq .

⁠Appendix - send jenkins⁠ logs to fluentd

Let's adapt the Jenkins 2.0⁠ container to send its logs via fluentd to an elasticsearch instance in localhost.

docker run --tty --privileged --detach                 \
    --network host                                     \
    --log-driver fluentd                               \
    --volume /var/run/docker.sock:/var/run/docker.sock \
    --volume /usr/bin/docker:/usr/bin/docker           \
    --publish 8080:8080                                \
    --name jenkins-2.0                                 \
    devops4me/jenkins-2.0

⁠Appendix - fluentd | docker build | docker push

This example employs the safe credentials manager for keeping dockerhub credentials safe.

cd /path/to/fluentd/project
docker build --rm --no-cache --tag devops4me/fluentd .
safe open dockerhub devops4me
safe docker login
docker push devops4me/fluentd
docker tag devops4me/fluentd devops4me/fluentd:v0.1.0001
docker push devops4me/fluentd:v0.1.0001
safe docker logout

⁠Appendix - Remove all Docker Containers and Images

Wipe the docker slate clean by removing all containers and images with these commands.

docker rm -vf $(docker ps -aq)
docker rmi $(docker images -aq) --force
docker ps -a
docker images -a

⁠Appendix - Documentation for the S3 Plugin

https://www.elastic.co/guide/en/logstash/current/plugins-outputs-s3.html⁠

⁠Appendix - ElasticSearch and S3

⁠excellent information

https://www.fluentd.org/guides/recipes/elasticsearch-and-s3⁠

https://raw.githubusercontent.com/fluent/fluentd-docker-image/master/v1.3/alpine-onbuild/fluent.conf⁠

https://www.fluentd.org/guides/recipes/docker-logging⁠

⁠Appendix - Quickly Install an ELK Stack for Logging

The below commands switch off security for elasticsearch and then kibana automatically picks this up and does not display the login screen.

⁠Run ElasticSearch

docker run --detach                    \
    --name elasticsearch               \
    --publish 9200:9200                \
    --publish 9300:9300                \
    --restart always                   \
    --log-driver json-file             \
    --env discovery.type=single-node   \
    --env transport.host=127.0.0.1     \
    --env xpack.security.enabled=false \
    docker.elastic.co/elasticsearch/elasticsearch-platinum:6.0.0 && sleep 20

⁠Run Kibana

docker run --detach        \
    --name vm.kibana       \
    --publish 5601:5601    \
    --restart always       \
    --link elasticsearch   \
    --log-driver json-file \
    --env "ELASTICSEARCH_URL=http://elasticsearch:9200" \
    docker.elastic.co/kibana/kibana:6.0.0 && sleep 20

⁠Run Fluentd

@todo - remove the ELASTICSEARCH_USERNAME and ELASTICSEARCH_PASSWORD from the fluent configuration file in the devops4me/fluentd image and then remove the environment variables in the docker run for fluentd.

docker run --interactive --tty             \
    --name vm.fluentd                      \
    --network host                         \
    --restart always                       \
    --log-driver json-file                 \
    --env FLUENTD_CONF=fluentd-simple.conf \
    --env ELASTICSEARCH_HOSTNAME=localhost \
    --env ELASTICSEARCH_PORT=9200          \
    --env ELASTICSEARCH_USERNAME=elastic   \
    --env ELASTICSEARCH_PASSWORD=secret    \
    devops4me/fluentd

Tag summary

Recent tags

Content type

Image

Digest

Size

14 MB

Last updated

almost 7 years ago

docker pull devops4me/fluentd