Troubleshooting active deployment

[w0rd-driven]

After some time off I went back to the deployed application to work through issues as I noticed things weren't working as I thought I left it.

I'm currently in the process of adding these to the README and will use this as a generic issue instead of one per each problem as I likely should do it.

• "Only valid bearer authentication supported"
  • The SPOTIFY_ environment variables/secrets appeared to be missing. I'm not sure I left it in a working state tbh.
• "Constraint error when attempting to insert struct: * albums_pkey (unique_constraint)"
  • The migration has a create unique_index(:albums, [:artist_id, :name]) set which isn't the correct order according to documentation.
  • I'll likely take the bad approach and adjust the migration directly. I'm not the biggest fan of tiny migrations but I also consider this to be pre-release alpha where breaking changes will happen. If this were any closer to 1.0 I wouldn't do it.
  • I may end up doing the right thing but it doesn't address the underlying issue because this works perfectly fine locally.
  • I wonder if importing the data in postgres means the primary key identifier isn't incremented but this is my ignorance in Postgres. MySQL uses an AUTO_INCREMENT=<number> part of the create table definition where I can see the value.
  • This looks to be the case because SELECT last_value FROM albums_id_seq; returns 3 not select max(id) from albums; or 366.

[w0rd-driven]

TablePlus has limitations around backup and restore primarily because the database names are different (beatseek in prod vs beatseek_dev). If I would compile to production locally it wouldn't be a problem but I move quickly in ignorance.

What I do is truncate tables then choose Copy Rows As > INSERT Statement to insert data for the artists table and then albums. The correct way to adjust the sequences is to use SELECT setval('artists_id_seq', (SELECT MAX(id) FROM artists)); and repeat for each table.

I'm looking to create a revolving demonstration at some point where it'd be useful to have ecto.dump combined with data to return to a pristine state. That file will need both of these techniques.

[w0rd-driven]

Added a troubleshooting section to the readme to help future me