Credentials - unable to add any API keys

[ZimM-LostPolygon]

Describe the bug

Trying to add credentials for any connector always fails. Standalone client works just fine with those same credentials.

Steps to reproduce bug

1. Install using the "deploy" repo according to docs: https://hummingbot.org/installation/docker/#installation-hummingbot-20
2. Go to Credentials page. Create a new account
3. Try to add credentials for 'kucoin' connector
4. Observe the 'NoneType' object has no attribute 'encrypt_secret_value' error.

In the backend-api container log, there's a line:
"POST /add-connector-keys/kucoin_test/kucoin HTTP/1.1" 400 Bad Request

UPD: this happens with any connector, not just Kucoin

[rapcmia]

Kucoin and other connectors works on latest and using setup.sh from deploy repo

Can you try again to reproduce the error then:

1. Go to terminal and run docker compose log backend-api
2. Send the screenshot of the terminal so we can check the traceback of the error

Thank you

[ZimM-LostPolygon]

There is no traceback. The only relevant line is the one I've already posted:
"POST /add-connector-keys/kucoin_test/kucoin HTTP/1.1" 400 Bad Request

I've tried 3 different machines, it's always the same error for me, both with docker-compose.yml and docker-compose-dev.yml

[rapcmia]

Thanks for the response @ZimM-LostPolygon i will have a look if I can reproduce this issue this week

[ZimM-LostPolygon]

Looking at backend-api code though, there are only two places encrypt_secret_value is mentioned:

And judging from the 'NoneType' object has no attribute 'encrypt_secret_value' error, that means secrets_manager is None 🤔 I can't think of any reason why that could be...

I thought about running backend-api locally to debug it, but Conda activation always there as well on any of my machines, unfortunately.

[gdhaworth]

Looking at backend-api code though, there are only two places encrypt_secret_value is mentioned:

And judging from the 'NoneType' object has no attribute 'encrypt_secret_value' error, that means secrets_manager is None 🤔 I can't think of any reason why that could be...

I thought about running backend-api locally to debug it, but Conda activation always there as well on any of my machines, unfortunately.

Seems to be fixed by changing the environment variable CONFIG_PASSWORD=hunter2 to CONFIG_PASSWORD=a in the backend-api container. I looked through the code briefly and think this might have something to do with https://github.com/hummingbot/deploy/blob/main/bots/credentials/master_account/.password_verification but I'm not sure how that's generated.