On Fri, Mar 29, 2019 at 03:10:23PM +0000, Shinoda, Noriyoshi (PN Japan A&PS Delivery) wrote: > I tried this great feature for partition index. > The first time the REINDEX TABLE CONCURRENTLY statement is executed > to the partition, then an error occurs.
Yes, that's a problem. I am adding an open item. > The second run succeeds but leaves an index with an INVALID status. > I think this is not the desired behaviour. This one is partially expected actually. Invalid indexes are ignored when processing as that would cause at least a table-level reindex to double its amount of processing. However, it is not possible either to detach an index from an partition tree for indexes, hence the index cannot be dropped directly either. It seems to me that the root of the problem is that the partition indexes created as copycats of the original ones should never be in a state where they are attached to the index tree. -- Michael
signature.asc
Description: PGP signature
