Migrated data from rabbit_hole_update_8106 is lost when running drush deploy

Never mind, I was wrong. Post update hooks also run before config import. Only deploy hooks run after config import, but those are a Drush specific concept so probably not a good idea to introduce here. The alternative is to make sure the fields aren't recreated, so no data is lost. I'll try to come up with a solution.

Update hooks updating data should still be post update hooks though, so I'll leave that in the MR.

The alternative is to make sure the fields aren't recreated, so no data is lost.

Yeah, there doesn't seem to be a way to do that. updatedb creates and fills the fields and config:import overwrites them. The only way I can think about is to add a rabbit_hole_field_config_insert() hook and migrate the field there, after it's created. Not sure if we can use the Batch API at that point though.

For my own purposes, I'm going to create a separate MR that makes this a deploy hook. It's a relatively easy and sufficient fix for people using Drush.

I can confirm that the deploy hook in MR !86 does the job, but like I said before only for people using drush deploy.

The issue can be that drush deploy assumes up to date configuration (it's for deployment only) as it doesn't run config:export after updatedb, if there are any schema changes the old updatedb will update the schema, but config:import will import old schema in - that might create issues, like missing config, etc

Follow steps in Deployment of a new feature with configuration management - note actions on LOCAL ENVIRONMENT → before running drush deploy on server.

I'm aware of that, that's not the problem here. I described those steps in 'Steps to reproduce'.