Right now integration of this container in production is tricky.
Using custom fields needs currently one of these things:
- a whole new managed-schema using volume mounts, with the downside i miss new fields or field changes from upstream
- a custom dockerfile using sed, which causes huge amount of data transfer on every deployment (around 500MB). Not practicable on daily deployments.
- using sed as startup command and then running ckan as root, as managed-schema can be written only as root user. Bad practice at all.
All of these aspects seem inefficient to me, as each of them causes a drawback.
In my opinion it would be best to add custom fields via an Environment Variable and chmod/chown the managed-schema file to be writeable for solr user.
What do you think? Is there any downside of this approach?
And by the way: huge thank you for providing a preconfigured solr. Less headache on setup. :-)
Right now integration of this container in production is tricky.
Using custom fields needs currently one of these things:
All of these aspects seem inefficient to me, as each of them causes a drawback.
In my opinion it would be best to add custom fields via an Environment Variable and chmod/chown the managed-schema file to be writeable for solr user.
What do you think? Is there any downside of this approach?
And by the way: huge thank you for providing a preconfigured solr. Less headache on setup. :-)