Subject: Process for changing producer partition assignment strategy


That is puzzling to me, too. Could it be that you have `null` keys for
the "new topic" you mentioned in your original email? For `null` keys,
the fallback would be round-robin.

Or you just got lucky and the keys you write get distributed evenly "by
chance" -- in general, if the data is not skewed, hash partitioning
should result in a fairly even distribution, too.

-Matthias

On 11/15/19 1:21 AM, Mikkel Gadegaard wrote:
Comment: GPGTools - https://gpgtools.org
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=xWKW
-----END PGP SIGNATURE-----