First of all, please make sure the Helm version is larger than 3.7. It is because the chart leverages an undocumented feature of helm. Upgrade the helm and try again.
If the problem isn't solved, please open a new issue and provide detailed information about the problem, including steps to reproduce, expected behavior, and any relevant logs or error messages.
Sometimes, after several rounds of re-installation, you may find the hook for creating database or setting the username/password failed. For example, when you inspect the logs of the hook, you may find the following ones:
kubectl -n risingwave-poc logs pod/risingwave-hook-create-databases-8xvj4
Password for user root:
psal: error: connection to server at "risingwave.risingwave-poc.svc"(172.20.23.179), port 4567 failed: ERROR:PasswordError: Invalid password
This is probably caused by re-installation with dirty meta store and state store backends. Please make sure you clean them properly and then retry.
Warning
Restart a RisingWave without persistent stores will cause data loss and DOS. The best way to recover in this case is to rebuild everything.
If the meta store and state store backends that RisingWave is using are all persistent (remote backend or local
persistent), simply delete the RisingWave pods and let them restart to get a full restart of the cluster. Note
that SQLite meta store and Local File System state store are not persistent unless the data paths are in a
persistent volume. Please refer to the installation guide for more details.
What to do if risingwave fails to start with PostgreSQL meta backend giving an uppercase database name?
When starting RisingWave with PostgreSQL as the metadata backend, if the startup fails and you provided an uppercase database name, please ensure that the corresponding uppercase database has been created properly. In PostgreSQL, if you create a database with an uppercase name without using double quotes, the database name will be converted to lowercase. Check SQL-SYNTAX-IDENTIFIERS for more details.