What happens when INSERT/DELETE/UPDATE/COPY run on a secondary subcluster?

Deb_Chatigny

What happens when INSERT/DELETE/UPDATE/COPY run on a secondary subcluster? Will the client get a warning/error?
@skeswani , @skamat
Sizing and Configuring Vertica in Eon Mode for Different Use Cases

ybei

it should work still. The downside is that mergeout will need to read file directly from s3, rather than depot.

Pan

And mergeout runs on primary subcluster instead of secondary subcluster, so it will always need to read from s3 if keep loading from secondary subcluster.