Uploaded image for project: 'Debezium'
  1. Debezium
  2. DBZ-4884

Oracle connector stops calling logminer without any error message

XMLWordPrintable

    • Icon: Bug Bug
    • Resolution: Done
    • Icon: Major Major
    • 1.9.0.Final
    • 1.8.1.Final
    • oracle-connector
    • None
    • False
    • None
    • False
    • Hide

      Start the connector and wait overnight. It does not happen every night but it has happened three times in 7 nights since we started. We do have a job that runs between 2 and 3 am Central time which corresponds to when the issue happens. It truncates two moderate sized tables, inserts 3 million rows into them, then does merge statements updating all 3 million rows one at a time committing every 10,000 rows.

      Show
      Start the connector and wait overnight. It does not happen every night but it has happened three times in 7 nights since we started. We do have a job that runs between 2 and 3 am Central time which corresponds to when the issue happens. It truncates two moderate sized tables, inserts 3 million rows into them, then does merge statements updating all 3 million rows one at a time committing every 10,000 rows.

      We have an Oracle connector setup to connect to an on premises 11.2.0.3 non-RAC Oracle database running on HP-UX. The Debezium software runs in an AWS EC2 and connects to an AWS MSK Kafka cluster. It is connected to a production database. It is replicating 6 moderate size tables with the largest having 2 million rows. We started up the connector on 3/11/22 and it has failed three times since then around 2-3 am Chicago time which is about 5 hours later UTC. There are no error messages in the connect log. On the Oracle database alert log we can see that the LOGMINER calls stop at that time. When we check in the morning after each issue we find that the connector is in RUNNING status but the the Oracle session (we use user DBZ) is not there. So, the Oracle session has dropped and the connector is no longer making LOGMINER calls but nothing in the connect log says why this would be. Also, when we look at the six Kafka topics that are being populated we see that they have not been updated since the LOGMINER calls stopped.

       

      See https://debezium.zulipchat.com/#narrow/stream/302529-users/topic/oracle.20with.20debezium.201.2E8.2E1.20stop.20sending.20data 

        1. LongSQL-1.java
          0.8 kB

              ccranfor@redhat.com Chris Cranford
              bobbydurrett-1 Bobby Durrett (Inactive)
              Votes:
              0 Vote for this issue
              Watchers:
              4 Start watching this issue

                Created:
                Updated:
                Resolved: