Active Dataguard Switchover Issue

From: Patrick Jolliffe <jolliffe_at_gmail.com>
Date: Tue, 1 Mar 2016 18:17:51 +0800
Message-ID: <CABx0cSUF0kv9Z2xDC+ibJpYW3QNGKdvmY-1gjVgF_9EbM8GiRA_at_mail.gmail.com>

Testing Active DataGuard (12.1.0.2 Database and Grid Infrastructure, patched to Jan 2016 PSU on RHEL6 64-bit.)

I create GTT:
CREATE GLOBAL TEMPORARY TABLE poc_gtt as select * from all_objects where 1=0;

If I connect to standby instance and execute the following: insert into poc_gtt2 select * from all_objects;

Without disconnecing that session, I attempt to switchover, the switchover fails.
The alert log on the standby fills up with messages like the following:

Active process 29512 user 'grid' program 'oracle_at_vmhklftorcpoc2' Active process 29512 user 'grid' program 'oracle_at_vmhklftorcpoc2' Tue Mar 01 17:39:18 2016
Errors in file /u01/app/oracle/diag/rdbms/poc2/POC/trace/POC_pmon_29335.trc: ORA-03170: deadlocked on readable physical standby (undo segment 65535)

Below is contents of the trace file:

kssxdl: error deleting SO: 0x2b33b3990 = transaction (66)

---

  SO: 0x2b33b3990, type: 66, owner: 0x2b8ce45f8, flag: INIT/-/-/0x00 if: 0x3 c: 0x3

   proc=0x2b8b3aea0, name=transaction, file=ktccts.h LINE:444, pg=0 conuid=0 ORA-03170: deadlocked on readable physical standby (undo segment 65535)

I have opened SR with support, but anyone has a test configuration and want to try this out?
Confirm the problem, or is there something special on my install (can't think what, it is a fresh install)?
Any ideas from trace what is going on and how to move forwards with it?

Thanks in advance

Patrick

--
http://www.freelists.org/webpage/oracle-l

Received on Tue Mar 01 2016 - 11:17:51 CET