On Sun, Dec 17, 2017 at 10:50 AM, Alvaro Herrera <alvhe...@alvh.no-ip.org> wrote: >> If ALTER INDEX .. ATTACH is already taking AEL on the parent, then I >> think it might as well try to validate while it's at it. But if not >> then we might want to go with #2. > > The problem I have with it is that restoring a dump containing indexes > on partitions becomes a O(N^2) deal as it has to do the full check once > for every index we attach.
Sure. If the constant factor is high enough to matter, then VALIDATE makes sense. IMHO, anyway. -- Robert Haas EnterpriseDB: http://www.enterprisedb.com The Enterprise PostgreSQL Company
