Error: F_JD02FD: PostgreSQL Streaming Replication Error

Issue

Channel capture fails during PostgreSQL logical replication with the following error:

F_JD02FD: PostgreSQL Streaming Replication error while retrieving PostgreSQL logical log records: -2 : SSL SYSCALL error: EOF detected. LOG: terminating walsender process due to replication timeout

Environment

• HVR 6
• Source: PostgreSQL

Resolution

To resolve this issue:

1. Adjust the PostgreSQL server settings:

   i. Set wal_sender_timeout to 0 to disable the walsender timeout, or increase it to a longer interval, such as 5 minutes.

   ii. Set HVR_PQ_STREAMING_REPLICATION_WAIT_MAX to 1024 using an HVR environment action.

   iii. Set HVR_PQ_STREAMING_REPLICATION_RESTART_AT_END_OF_WAL to 0 using an HVR environment action.

   iv. Reactivate the channel to apply the environment variable changes.

2. Check the replication slot by running the following query. Replace hvr_<channel_name> with your replication slot name:

   SELECT slot_name, active, restart_lsn, confirmed_flush_lsn 
   FROM pg_replication_slots 
   WHERE slot_name = 'hvr_<channel_name>';
   

   If the slot is missing or invalidated, perform a full data refresh.

3. Create a heartbeat table and use a cron job to update it every 15 minutes. This generates periodic transactions that help prevent inactivity. PostgreSQL doesn't include a built-in scheduler, so an external schedule such as cron is required.

For more information, see Stuck or Slow Replication Slot.

Cause

This issue occurs when database activity is low and the replication connection times out, or when heavy transaction loads delay data streaming. If the server doesn't receive a keepalive message or acknowledgment within the configured timeout period, it terminates the connection.