PostgreSQL : terminating connection due to administrator command postgresql

Question

I found this error occurred after SQL insert statement was running. The SQL is called by store procedure which executed from Java. Please see as below.

=== 2012-03-23 20:39:13  Insert statement procedure executed ===

2012-03-23 21:16:01 CST:[local]:user@report:[1591]:FATAL:  57P01: terminating connection due to administrator command
2012-03-23 21:16:01 CST:[local]:user@report:[1591]:LOCATION:  ProcessInterrupts, postgres.c:2662
2012-03-23 21:16:01 CST:[local]:user@report:[656]:NOTICE:  00000: Slony-I: cleanup stale sl_nodelock entry for pid=1583
2012-03-23 21:16:01 CST:[local]:user@report:[656]:LOCATION:  exec_stmt_raise, pl_exec.c:2298
2012-03-23 21:22:55 CST:[local]:user@app:[17236]:NOTICE:  00000: Slony-I: cleanup stale sl_nodelock entry for pid=1590
2012-03-23 21:22:55 CST:[local]:user@app:[17236]:CONTEXT:  SQL statement "SELECT  "_slon_rep".cleanupNodelock()"
        PL/pgSQL function "cleanupevent" line 77 at PERFORM
2012-03-23 21:22:55 CST:[local]:user@app:[17236]:LOCATION:  exec_stmt_raise, pl_exec.c:2298
2012-03-23 21:22:55 CST:[local]:user@app:[17236]:NOTICE:  00000: Slony-I: cleanup stale sl_nodelock entry for pid=1634
2012-03-23 21:22:55 CST:[local]:user@app:[17236]:CONTEXT:  SQL statement "SELECT  "_slon_rep".cleanupNodelock()"
        PL/pgSQL function "cleanupevent" line 77 at PERFORM
2012-03-23 21:22:55 CST:[local]:user@app:[17236]:LOCATION:  exec_stmt_raise, pl_exec.c:2298

=== 2012-03-23 22:15:04 Insert statement procedure return release signal to Java ===

There were 2 schema in my database. "app" and "report" and both of them were synchronized mutually by Slony. I've checked in table and found those records were inserted since 20:39:13. My concerning point is why the Postgres return release signal to Java at 22:15:04 since the insert statement already executed at 20:39:13. I suppose that the above process was affected to the insert statement process was hung and need to wait until its' finish then released to Java.

Thanks in advance Stop :)

To allow anyone to find an answer can you please enable log_statement="all" and apply it to PostgreSQL and edit your question to include the logs from PostgreSQL at the time this happens? — Chris Travers, Mar 26 '13 at 01:45

score 0 · Answer 1 · answered Mar 27 '13 at 01:21

There were 2 schema in my database. "app" and "report" and both of them were synchronized mutually by Slony.

My guess is that Slony is detecting records inserted into replicas (remember, Slony is Master-Slave only) and aborting your connections. Please replicate only one way.

PostgreSQL : terminating connection due to administrator command postgresql

1 Answers1