Problems with migrations for 0.12.0

[hypnoglow]

CHANGELOG upgrade guide recommends running index migration manually for large tables. After that, "normal" river migration should be run to finalize the process.

The problem is that the existing index is tolerated, but the existing column is not, which results in the following error:

column "unique_states" of relation "river_job" already exists

[bgentry]

@hypnoglow hmm, there wasn't a unique_states column prior to this migration v6. Did you perhaps try to run that migration line adding that column twice?

[hypnoglow]

there wasn't a unique_states column prior to this migration v6

@bgentry You're right. But your migration guide suggests to run the following query to create it manually, because it is used in the index:

ALTER TABLE river_job ADD COLUMN unique_states BIT(8);

CREATE UNIQUE INDEX CONCURRENTLY river_job_unique_idx ON river_job (unique_key)
    WHERE unique_key IS NOT NULL
      AND unique_states IS NOT NULL
      AND river_job_state_in_bitmask(unique_states, state);

Then the guide recommends running river migrator, and so ALTER is executed second time.

[brandur]

@hypnoglow Ah yes that's quite right. We'll get a fix shipped for this immediately. Thanks for reporting!

[hypnoglow]

Thank you 💙

[hypnoglow]

Just FYI

In release descripion https://github.com/riverqueue/river/releases/tag/v0.14.3 PR links are mixed up with each other.

[bgentry]

@hypnoglow thanks, fixed the GitHub release and will tweak the changelog too 🙏