ERROR: unexpected pageaddr for logical replication service after Aurora maintenance window

Question

Hello,

Since a Sunday maintenance window, we began to receive 
```
ERROR: unexpected pageaddr
```
 for all logical replication apps (AWS DMS and Debezium on Kafka Connect).
This error is somehow recoverable and periodic:
![DB write instance logs](/media/postImages/original/IMS7FwUZkkRxebe1LNGNEKeA)

Otherwise DMS shows no error on this but Debezium kind of fails and recovered by connecting Postgres again.
We are wondering if this could be related to Aurora DB minor version update? Or more like we are wrongly using some DB parameters (reboot actually activate the changes).
Thanks a lot in advance :).

Best regards,
David

Accepted Answer

Hi David,

This error log is a known issue and will be patched likely in our next Aurora PostgreSQL release.

In the interim, in order to mitigate the issue, disable the logical WAL cache feature by setting the cluster parameter 'rds.logical_wal_cache' to 0 and restarting your writer instance. Please rest assured that disabling this will not result in any loss of data from existing logical replication publications.

I hope that helps :)

ERROR: unexpected pageaddr for logical replication service after Aurora maintenance window

Contenuto pertinente