Clojurians Log v2

That is an impressive performance improvement, looking forward to testing this release soon.

I don't believe I specified :scope in my app in the older deployments but specifying it now (with the new db credentials) doesn't seem to help or override the error. I can't even connect to the current deployment db to make a copy of it, because datahike will throw config-does-not-match-stored-db.

Somewhat at a loss. I don't want to destroy anything without making a copy but I can't access the db!

Which datahike version are you using?

I relaxed the config check in a recent version, but I might have to expand that.

How are the credentials passed? I assume as part of the store config.

just updated to most recent. Still failed

Same error

Ok

Yes, in the store config

Like so:

(def cfg
  {:store {:backend :jdbc
           :dbtype "postgresql"
           :host ""
           :port 5432
           :user ""
           :password ""
           :dbname "dc6bbrav2u4lve"
           :scope "already-set-this-huh"}})

Is the dbname changing for you?

The store identity for JDBC should not depend on the password.

This is how the store is identified https://github.com/replikativ/datahike-jdbc/blob/main/src/datahike_jdbc/core.clj#L21

I assume that table stays the same, then the culprit should be dbname here.

@alekcz360 has worked with JDBC a lot, but I think not with Heroku yet.

We can relax that further, the config check is there to not accidentally corrupt a database.

By using the wrong configuration.

@feedmyinbox02_clojuri in the error it tells you what has changed. Is the dbname the culprit?

dbname does change, actually

Let me check the error report

:diff
({:store
  {:dbname "dc6bbrav2u4lve", 
   :password "", 
   :user "u10ue9c465kc7o", 
   :host "c4[...]."}}
 {:store 
  {:dbname "d3ldb5vhikjho2", 
   :password "", 
   :user "jhhyishypdwycq", 
   :host "ec2[...]."}}
 {:keep-history? true,
  :search-cache-size 10000,
  :index :datahike.index/persistent-set,
  :store {:port 5432, :dbtype "postgresql", :backend :jdbc}, :store-cache-size 1000, :attribute-refs? false, :crypto-hash? false, :schema-flexibility :write, :branch :db})

Ok, this is a bit odd, but I guess it is what it is.

Probably the best way to deal with oddities like this is to allow deactivation of the config check altogether at your own risk.

How would I do that?

That is a boolean I can set somewhere in the store?

I need to add an option for that.

I think I would probably add a field to the config itself to avoid dynamic bindings.

We'd have to update the datahike with a property like allow-unsafe-configor something along those lines

Alternatively it could be an environment variable, but then it is global.

Is there anything I can do at the moment less nuclear than reconstituting the entire production db?

... oh wait, I can't do a migration or copy. I don't know how to get a new conn. This is blocking me from making a new conn even

Wait, this is bad. How can I make a copy of the production db using Datahike tools if this error is blocking?

@whilo you want to work on a new cut? or should I pick it up? I can do it this evening?

@feedmyinbox02_clojuri try this PR https://github.com/replikativ/datahike/pull/693

Lmk whether it works. You need to set :allow-unsafe-config true in the configuration.

I might have to add stuff the config specification, I have not checked this fix yet.

Ok, trying now. Thank you

Hi whilo, it didn't seem to work. I am DM'ing you direct copy-paste of error

The PR (#693) provided by whio worked! 😊 Thank you

You are welcome! Lmk if there is anything else causing you issues.