On Sat, 2021-05-08 at 13:37 +1000, Venkata B Nagothi wrote: > We are thinking to upgrade to PG 11 instead so that we can avoid doing ALTER TABLE.. SET WITHOUT OIDs. > Does that makes sense ? Please advise if there are any gotchas !
It makes sense, but it means that you will have to face the same problem later. However, for upgrading from v11 with little down time you may be able to use logical replication.
Yes, we will have a lot of time to deal with the OID problem later, good thing is we will be out of 9.5 with less trouble. Hopefully we will be on a better replication architecture soon which will make it much easier for us.
We have upgraded our test Postgres-9.5 cluster to 11.11 and it went smooth without complaining about OIDs which is great. However, we see the following errors in our Postgres log files, not sure if its upgrade related.
Error :
ERROR: column c.relhaspkey does not exist at character 33
Below is the query generating the error :
STATEMENT: SELECT c.relname AS table_name, c.relhaspkey AS has_primary_key FROM pg_catalog.pg_class c, pg_catalog.pg_namespace n WHERE c.relkind = 'r' AND c.relnamespace = n.oid AND n.nspname = $1
Any advice on how bad it is and how to get rid of it would be great.