ERROR: could not map filenumber after Aurora upgrade to 16.1

Question

Hello,

After upgrading our Aurora cluster from 15.3 to 16.1 we've got the following error in the logs.

We have a logical replication between two aurora cluster one in 14.10 and this one.

When looking at the publisher logs (16.1) every time the subscriber try to start the replication it gots the following error in the publisher logs :

`could not map filenumber "base/16402/3147867235" to relation OID`

I've tried :
1. Dropping the replication slot and creating a new one
2. Starting over from a clone
3. Starting the replication from a snapshot

Not sure what can I do. I've tried deleting the table corresponding to the OID 3147867235. But then it starts giving me another OID failling.

Answer

Hello There,

Thank you for posting your question here.

The issue you are facing is very DB specific and need further analysis however, on best effort basis I would request you to check if the replication slots are active. You can use the following query to check the same:

select * from pg_catalog.pg_replication_slots;

In the output, please refer the column  'temporary' and 'active'.

You can also verify WAL level using 'SHOW wal_level;' it should be set to logical.

[+] Best practices for Amazon RDS PostgreSQL replication - https://aws.amazon.com/blogs/database/best-practices-for-amazon-rds-postgresql-replication/

If both the above outputs are true, I would request you to create a case with the Support Team to further deep dive the error.

I hope this is helpful!

ERROR: could not map filenumber after Aurora upgrade to 16.1

相關內容