Why pulsar retention quota must exceed configured backlog quota

[michaeljmarshall]

Discussed in #17697

^{Originally posted by mnit016 September 16, 2022}
Hello everyone,

I'm new to Pulsar, recently I get into a problem when configure the backlog/retention quota for namespace.

The problem:
retention: {time: -1, size: 10M} => then I can set backlog quota as much as I want.
But if retention: {time: 1, size: 10M} => when I set backlog quota exceed 10M, it said "Backlog Quota exceeds configured Retention quota for namespace. Please increase retention quota and retry", I have no idea why.

Could someone please help to explain for me?
Thanks

[michaeljmarshall]

This appears to be a bug, based on @Jason918's analysis here:

#17697 (comment)