Hi Jared,
It is possible to setup an event generated incident like this:
alter system set events '1555 incident(sto)';
Then, a standard flood-control kicks in after a certain number of incidents
(please see the following MOS article for the thresholds: DIA-49435 When
Trying To Create An Incident Packet (Doc ID 1358087.1)).
For instance, in my 19c database I have got five incidents and the sixth
was flood controlled.
Here is a relevant excerpt from the trace file:
DDE: Problem Key 'ORA 700 [EVENT_CREATED_INCIDENT] [1555] [STO]' was flood
controlled (0x2) (incident: 111062)
That is not exactly what is required but it is the closest I can come up
with.
Best regards,
Mikhail Velikikh
On Tue, 20 Aug 2019 at 17:49, Jared Still <jkstill@xxxxxxxxx> wrote:
Hello,
Does anyone know if it is possible to control the total number of trace
files created by 'alter system set events' ?
For example, the following command will cause all sessions to dump stack
traces when there have been at least 2 occurrences (armcount) of ORA-1555
in a session:
alter system set events '1555 trace name errorstack level 3, armcount 2,
lifetime 2';
After 2 events have occurred (lifetime) then the session will stop dumping
stacks for this error.
What I would like to be able to do is limit this system wide.
Say for instance 3 sessions dump stack traces for ORA-1555.
I would like to limit this to N system wide dumps of stack trace.
So after 3 sessions have created a trace file, no other sessions should do
so.
The reason is that if left unattended, when many session incur ORA-1555,
then this could potentially be quite a few rather large files.
The idea of course is to be able to set and forget, until the next event.
And yes, I know that TFA could be setup to deal with this, but that is
another story...
Jared Still
Certifiable Oracle DBA and Part Time Perl Evangelist
Principal Consultant at Pythian
Pythian Blog http://www.pythian.com/blog/author/still/
Github: https://github.com/jkstill
