if you delete the storage/xtdb directory in prod, that'll delete the index (and the transaction log + doc store, unless you set up Postgres). then do a bb restart

i set up postgres 👀

I deleted all the users in the db and then tried to sign up with the same email… just deleted storage/xtdb and ran bb restart but still got this error:

clojure.lang.ExceptionInfo: Transaction violated a constraint {:tx ([:xtdb.api/match #uuid "059f8246-da94-46ce-b0db-5e879e208adf" nil] [:xtdb.api/put {:user/joined-at #inst "2023-07-22T17:55:50.887-00:00", :user/email "", :xt/id #uuid "059f8246-da94-46ce-b0db-5e879e208adf"}] [:xtdb.api/fn :biff/ensure-unique {:user/email ""}])}

to really make sure everything's cleared out, you can: 1. ssh into the server as root and run systemctl stop app 2. delete the storage/xtdb directory 3. drop the Postgres db. I think you can do this from the digitalocean website 4. then run bb restart

I think that worked 🙏

no worries 🙂 the way you're doing it is how I'd do it to. agreed there's less of a need for migration tools. even when I do want to do some sort of migration (e.g. add a new required attribute to all existing docs in the database) I just code it up as a plain fn and call it from the repl.

do you every see performance hits if it’s touching large amounts of entities?

do you mean during or after the migration? the more entities are involved, the longer the migration will take to run, same as all queries/transactions

during. That’s true, just wondering if you hit a certain number of entities in a transaction where using the repl causes problems. I think i’m getting ahead of myself 😄

🙂 I do usually batch the transaction, so instead of one big transaction, I'll do multiple transactions with, say, 500 documents at a time