My investigation was:
Case 1:
mongodb: db.<collection>.remove({}) # configured admin user will remain in db, though copy config.json in place exec nodebb setup copy config.json in place (req'd as nodebb setup truncates the config which is in itself bad enough)config.json contains a valid admin/passwd/email configuration. Mongodb has the same nodebb admin username configured.
Result: no login possible.
Case 2:
=> Running nodebb setup without cleaning up mongod database beforehand but with valid admin config in config.json makes nodebb setup create "admin" username with random password.
Result: login using "admin" username and provided random password is successful.
Case 3:
Using environment variables with nodebb setup like this (w/ database cleaned up):
yields:
[...] This looks like a new installation, so you'll have to answer a few questions about your environment before we can proceed. Press enter to accept the default setting (shown in brackets). 2021-06-23T21:31:03.849Z [1261005] - info: Now configuring mongo database: 2021-06-23T21:31:03.979Z [1261005] - info: [database] Checking database indices. 2021-06-23T21:31:03.984Z [1261005] - info: [database] Checking database indices done! 2021-06-23T21:31:04.851Z [1261005] - verbose: [minifier] utilizing a maximum of 0 additional threads Configuration Saved OK Populating database with default configs, if not already set... [...] Schema update complete!Result: login with "my-nodebb-admin" username and preconfigured password is successful.
Tests done with nodebb/v1.17.1 . Switched midway from npm to yarn as npm failed to accomplish it's task once too often.
Conclusion: automated nodebb setup
using environment variables as of the given nodebb/node/npm versions is working using preconfigured config.json as of the given nodebb/node/npm versions is NOT workingFurthermore, nodebb setup truncates a preconfigured config.json, preventing successful startup of nodebb.
In case preconfigured config.json was once working there is some regression. A possible fix for the above discrepencies/failures should prevent future regressions.
NB: If there is interest, the final ansible playbook could possibly be open-sourced on github.
