I've encountered a problem, where setting new allocation settings on an index doesn't seem to trigger reallocation of shards. I'm trying to implement rollover-logic, and I'm failing on this step, which is necessary before shrinking the index.
I'm following this blogpost here:
I've reduced my problem to a simple test case, and included the resulting log output here:
Maybe someone spots an error? Or are there settings on the cluster level I should look out for?
Actually found the problem. The code is perfectly fine, it was just my nodes were not registering with their correct hostname as name. This is why the allocation couldn't be executed, as there was no such node as specified.