I currently have an issue where writes to my hypertable are causing gigantic WAL segments which can’t be logically decoded causing the WAL sender to run into a memory issue:
ERROR: could not receive data from WAL stream: ERROR: invalid memory alloc request size 1213919920
The table doesn’t have any large object in either. Only 19 columns and either uuid, varhcar(255), int8 and timestamps. The insert size we are doing is max 1000 rows in a single transaction.
This is causing Logical Replication to break in our cluster. Has anyone got any ideas on how to fix this and could it be directly linked to the fact it is a HyperTable rather than a normal PG table?