Cherry-pick #24817 to 7.x: [Elastic Agent] Fix Docker container to allow state to properly be handled

[blakerouse]

Cherry-pick of PR #24817 to 7.x branch. Original message:

What does this PR do?

This fixes the Elastic Agent Docker image to allow a proper state to be stored for the image. Previously this was not possible for the state to be placed into a volume which prevent restarts of the container to maintain its previous state.

This changes the Elastic Agent to not version its directory when running under Docker. This is key to allowing the state to be storable and being that an Elastic Agent container cannot be self-upgraded it is acceptable that it is not versioned.

Because of how the Elastic Agent ships with the builds in the download directory the state could not mount over that directory of Elastic Agent would need to re-download those files. This changes the container sub-command to rsync the builds download directory into the state directory before actually running. This allows new downloads to stay in the state directory and allows the running Elastic Agent to use the binaries that shipped with the image.

By default the state of the Docker container goes to /usr/share/elastic-agent/state. This can changed with STATE_PATH environment. Elastic Agent logs now default to output stderr so they can be collected by the container runtime. In the case that LOGS_PATH variable is set then the Elastic Agent will log to $LOGS_PATH/elastic-agent.log instead of stderr.

Other fixes include using the $STATE_PATH/data/tmp instead of /tmp for the unix sockets as the container cannot write to /tmp.

Why is it important?

Critical to allow the state of the Elastic Agent to persist restarts of the Elastic Agent Docker image.

Checklist

• My code follows the style guidelines of this project
• I have commented my code, particularly in hard-to-understand areas
• [ ] I have made corresponding changes to the documentation
• [ ] I have made corresponding change to the default configuration files
• I have added tests that prove my fix is effective or that my feature works
• I have added an entry in CHANGELOG.next.asciidoc or CHANGELOG-developer.next.asciidoc.

Author's Checklist

• State is stored in /usr/share/elastic-agent/state.
• State path can be changed with STATE_PATH.

How to test this PR locally

$ # Build the docker container in dev-mode
$ cd x-pack/elastic-agent
$ PLATFORMS="+all linux/amd64" mage dev:package
$ # Run the docker container
$ docker run -it -e KIBANA_FLEET_SETUP=1 -e ELASTICSEARCH_HOST=http://host.docker.internal:9200 -e KIBANA_HOST=http://host.docker.internal:5601 -e FLEET_ENROLL=1 -e FLEET_URL=http://host.docker.internal:8220 -e FLEET_INSECURE=1 docker.elastic.co/beats/elastic-agent:8.0.0

Related issues

• Closes [Elastic Agent] Allow separation of runtime state location and installation and executable location.  #24160
• Closes [Elastic Agent] Accept logs path configuration #24271
• Closes [Elastic Agent] empty downloads folder when providing --path.home #24801

[elasticmachine]

Pinging @elastic/agent (Team:Agent)

[elasticmachine]

💚 Build Succeeded

the below badges are clickable and redirect to their specific view in the CI or DOCS

Expand to view the summary

Build stats

• Build Cause: Pull request #24879 opened

• Start Time: 2021-03-31T20:14:24.474+0000

• Duration: 57 min 26 sec

• Commit: ea7ef3c

Test stats 🧪

Test	Results
Failed	0
Passed	6532
Skipped	24
Total	6556

Trends 🧪

💚 Flaky test report

Tests succeeded.

Expand to view the summary

Test stats 🧪

Test	Results
Failed	0
Passed	6532
Skipped	24
Total	6556