[DBZ-1579] Regression of postgres Connector times out in schema discovery for DBs with many tables - Red Hat Issue Tracker

XML

Word

Printable

Steps to Reproduce:
Hide

In any snapshot mode (but easy to repro with "never"), start a replication stream with many tables (we have over 1000)

observe that the schema refresh will happen, but then a read will fail with a closed connection as soon s the replicationStream is used
Show
In any snapshot mode (but easy to repro with "never"), start a replication stream with many tables (we have over 1000) observe that the schema refresh will happen, but then a read will fail with a closed connection as soon s the replicationStream is used

In ~~DBZ-1214~~, there was an issue where the replication connection would time out during a schema refresh.

This was fixed, but it appears was re-introduced as part of the effort to have postgres use the standard framework.

The code for doing the keep alive thread still exists, it simply isn't called.