This PR updates elasticsearch to a supported version. Subjectively, this speeds up the tests on my machine a good amount.
It adds a Dockerfile to build our own image, which includes some configuration data for the decompounder. Which is now the hyphenation compounder, included in Elasticsearch itself, so that the elasticsearch-analysis-decompound plugin can be removed. The respective tests are still green, so I'm assuming this works as expected. Also, see dc3a97d for some links about all of this.
The elastichq admin interface is moved into a separate service using the official docker image.
The docker-compose config now uses an explizit docker volume for persisting the data. Docs here. I'm not entirely sure how this will impact the setup we currently have on the server - the index may have to be re-created when deploying this. @jrlover do you see any problem with this?
Also:
replaces the deprecated elasticsearch "string" type in papers.rb with text and keyword respectively.
add "dockerize wait" to the test runner script, in order to wait for the elasticsearch service to be fully booted before starting the test run, making that a bit less flaky.
Update elasticsearch to 6.8.6
This PR updates elasticsearch to a supported version. Subjectively, this speeds up the tests on my machine a good amount.
It adds a Dockerfile to build our own image, which includes some configuration data for the decompounder. Which is now the hyphenation compounder, included in Elasticsearch itself, so that the elasticsearch-analysis-decompound plugin can be removed. The respective tests are still green, so I'm assuming this works as expected. Also, see dc3a97d for some links about all of this.
The elastichq admin interface is moved into a separate service using the official docker image.
The docker-compose config now uses an explizit docker volume for persisting the data. Docs here. I'm not entirely sure how this will impact the setup we currently have on the server - the index may have to be re-created when deploying this. @jrlover do you see any problem with this?
Also:
papers.rbwithtextandkeywordrespectively.