topic Re: Table ACLs, secrets, and compute clusters in Data Governance

Table ACLs, secrets, and compute clusters

Mr__E — Sat, 12 Feb 2022 01:34:51 GMT

I am adding a credential to access MongoDB through the Python MongoDB Spark connector. It is stored as a secret and the `users` group has `MANAGE` ACL privileges. I turned on table ACLs and created a `High Concurrency` compute cluster. I then run a script that attempts to read from MongoDB, but fails due to authentication issues. If I put in the password manually, it succeeds. If I use the same script on a non-High Concurrency (aka no table ACL) cluster, it succeeds. Only when I use table ACLs does it fail. Anyone know if I'm missing a setting?

Re: Table ACLs, secrets, and compute clusters

AmanSehgal — Sun, 13 Feb 2022 22:26:36 GMT

Interesting. Check if privileges are set properly.

Also, maybe check that passwords are exactly same - the one in your script and the one you're pasting manually.

Re: Table ACLs, secrets, and compute clusters

Mr__E — Mon, 14 Feb 2022 20:12:16 GMT

| Check if privileges are set properly.

I added `MANAGE` permissions for `users` to the secret scope and also gave `users` restart access to the compute cluster. Are there other permissions I should be setting?

| Also, maybe check that passwords are exactly same - the one in your script and the one you're pasting manually.

Thanks for getting back. I checked that they are the same by copy-pasting and pasting the same one during `secrets put` and in the notebook (I made sure to remove the extra line from the file). It is working correctly when I don't use the high-concurrency cluster as noted before, so that suggests the secrets is correct.

Re: Table ACLs, secrets, and compute clusters

Atanu — Wed, 16 Mar 2022 05:26:18 GMT

@Erik Louie looks the issue is resolved. Please let me know if you need further help on this. Thanks!