Thanks for the links Stefan and for your input John :)
I've resolved the AWR creation for a PDB. There is a 12.2 new parameter
called AWR_PDB_AUTOFLUSH_ENABLED that defaults to FALSE. The setting
enables or disables auto AWR snaps for PDB's. It can be set true or false
within each PDB. If set in the CDB, it impacts all PDBs unless set
differently at the PDB level.
There is another parameter, AWR_SNAPSHOT_TIME_OFFSET (in seconds) to help
if you encounter cpu spikes from all PDBs taking snapshots at the same
time. If set to 1,000,000 then the offset is randomized.
Still waiting on Oracle dev to fix the parse error for STANDARD...
On Tue, Apr 25, 2017 at 2:43 PM, Stefan Koehler <contact@xxxxxxxx> wrote:
Our design has a customer in each PDB. I don't see how you cantroubleshoot performance for a single PDB from a CDB based AWR report.
Well if AWR is the best way to troubleshoot performance issues is another
topic but you can take AWR snapshots on PDB level with 12.2. For more
details please check these references:
multitenant-wp-12c-2078248.pdf ("Automatic Workload Repository (AWR)
reports at a
per-PDB basis are now supported")
("What’s new in 12cR2 is the ability to run AWR snapshots at CDB or PDB
Independent Oracle performance consultant and researcher
Upcoming online seminar: http://tinyurl.com/17-06-13-Shared-Pool-Internals
Woody McKay <woody.mckay@xxxxxxxxx> hat am 25. April 2017 um 18:46geschrieben:
Has anyone started using Oracle 12.2 multi-tenant (non-engineered) in
SRs have hit or created two new bugs:
We've started evaluating it for our production database. So far, my
DATABASE USING AWR_PDB OPTION
- BUG 25941188 - ORA-20200 WHEN ADDM REPORT GENERATED FROM A PDB
STANDARD OBJECT NAME IN 12.2
- BUG 25818707 - PLSQL COMPILATION FAILS WITH ORA-06555 WHEN USING
production with these bugs.
Has anyone else experienced this or been able to work around them?
Oracle dev is working on fixes, but no ETA. We can't take 12.2 into
troubleshoot performance for a single PDB from a CDB based AWR report.
Our design has a customer in each PDB. I don't see how you can
be many PDB for each CDB.
Started looking at using Statspack for each PDB...
12.1 (non-engineered) does not have the isolation features we want.