Refactor: reuse single_node mode for playground and standalone

[kwannoel]

• Add a profile for playground.
• Add a profile for standalone mode.
• Cloud needs to use the new standalone profile.
• Docs need to be updated to use the new playground profile via single_node.
• Docker needs to use the new standalone profile.

[kwannoel]

See #14951 (review)

[xxchan]

What about the naming? I quite like --preset in-memory/local-disk (or --profile) proposed here #14951 (review).

And maybe we can reuse standalone as the command name? (just like implement the map_opts feature directly on top of old standalone mode.

[kwannoel]

What about the naming? I quite like --preset in-memory/local-disk (or --profile) proposed here #14951 (review).

Yes, profile.

And maybe we can reuse standalone as the command name? (just like implement the map_opts feature directly on top of old standalone mode.

Yes, after we deprecate standalone, we can use single-node as an alias, and rename it to standalone.

[xxchan]

Yes, profile.

I mean the profiles' names. e.g., I like in-memory better than playground 🥺

[kwannoel]

Yes, profile.

I mean the profiles' names. e.g., I like in-memory better than playground 🥺

Yes, I prefer:

• in-memory (playground)
• persistent (single-node)

For standalone mode, can just choose either and supply extra-args for further configuration.

[fuyufjh]

We may remove playground mode in the future. Users who want to take a try of RW can just use single-node mode.

Discussing here https://risingwave-labs.slack.com/archives/C05TXGYUL6N/p1708578235388599?thread_ts=1707372310.943919&cid=C05TXGYUL6N