Re: concurrent re-partitioning of declarative partitioned tables

2020-12-02 Thread Nick Cleaton
On Wed, 2 Dec 2020 at 15:59, Michael Lewis wrote: > You can not have overlapping partitions that are both attached. > Not directly, no. That's why I'm considering the _partition_channel hack. Why do you want to merge partitions that you are "done with" instead of > just leaving them partitioned

Re: concurrent re-partitioning of declarative partitioned tables

2020-12-02 Thread Nick Cleaton
On Wed, 2 Dec 2020 at 16:07, David G. Johnston wrote: > On Mon, Nov 30, 2020 at 8:36 AM Nick Cleaton wrote: > >> I want to set up a large table on postgresql 12.4, using declarative >> partitioning to partition by record creation date. I'd like to have recent >> records in small partitions but o

Re: concurrent re-partitioning of declarative partitioned tables

2020-12-02 Thread David G. Johnston
On Mon, Nov 30, 2020 at 8:36 AM Nick Cleaton wrote: > I want to set up a large table on postgresql 12.4, using declarative > partitioning to partition by record creation date. I'd like to have recent > records in small partitions but old records in a few larger partitions, so > I want merges. The

Re: concurrent re-partitioning of declarative partitioned tables

2020-12-02 Thread Michael Lewis
You can not have overlapping partitions that are both attached. Why do you want to merge partitions that you are "done with" instead of just leaving them partitioned by day? Why are you partitioning at all? Are you confident that you need partitions for performance & that the trade-offs are worth

concurrent re-partitioning of declarative partitioned tables

2020-11-30 Thread Nick Cleaton
I want to set up a large table on postgresql 12.4, using declarative partitioning to partition by record creation date. I'd like to have recent records in small partitions but old records in a few larger partitions, so I want merges. The merges should be concurrent, in the sense that they lock out