treasuretron
commented
4 years ago In GitLab by @thomasmurphy on Mar 21, 2019, 21:01
2 kinds of errors we've seen so far: 1) Platform intermittently losing MySQL server connection, which causes Drupal config-import to fail 2) Entity UUID of content or group types not matching the dev database (eg if someone downloads a database from a branch or pr env), which can cause drush config-import to attempt to delete and recreate content or group types, which fails if they already have entities in the database, which causes the whole import to fail. These errors are visible when running
platform drush --environment=dev config-import
Using the platform CLI, but not as a part of the gitlab > platform pipeline automation.
When this happens, config on the dev DB then matches config pre-merge, and if the database is downloaded for local development on a different branch, and a developer runs
lando drush config-export
It will overwrite the yml in the config/sync directory of that branch with config from the previous commit, which may remove functionality previously merged in to dev when that new branch is also merged into dev.
Running
Lando composer install Lando drush updatedb lando drush config-import
After every merge into dev and before begging any development should massively reduce these kinds of errors. Adding this comment for cross-issue context but I will update the wiki around this issue soon.
In GitLab by @richbodo on Mar 15, 2019, 06:14
Tom, correct me if I'm wrong on this in any way:
Although we won't change our current development pipeline, we will document our understanding of how it can be improved, and what steps are required, in as much detail as we can at this point, to get there.