Do you see anything in background trace files about "time drift"?
Is the environment real or virtual ? I do see odd timing anomalies occasionally
on an OVM Linux running as a guest on a multi CPU windows machine.
Can you modify the code to capture current SCN as well?
(From my iPad mini; please excuse typos and auto-correct)
On 13 Jan 2018, at 08:01, Stefan Knecht <knecht.stefan@xxxxxxxxx> wrote:
If you have a code structure similar to this:
update t set start_time=systimestamp;
update t set end_time=systimestamp;
-- run some code here that takes anywhere between 0.1 and 2 seconds to
I have some executions of this code, 40 out of 6000 runs, where I end up with
the end_time being a timestamp BEFORE the begin_time and not just a
microsecond or two. 0.4 seconds, 0.6 seconds. A significant amount of time.
This leads me to believe that Oracle somehow fires autonomous transactions
asynchronously or out of order of the code structure?
Before I go off on a digging spree - has anyone seen this before or has an
Oracle 12.1 on *nix.
zztat - The Next-Gen Oracle Performance Monitoring and Reaction Framework!
Visit us at zztat.net | Support our Indiegogo campaign at igg.me/at/zztat |