| streams and required_checkpoint_scn [message #543015] |
Sun, 12 February 2012 10:15  |
 |
piotrtal
Messages: 168
Registered: June 2011
Location: Poland/Czestochowa
|
Senior Member |
|
|
could you explain me what forces required_checpoint_scn change forward during normal database work?
i know that when i enforce checkpoint for capture process using...
call dbms_capture_adm.set_parameter('P2ORA19_CAP','_CHECKPOINT_FORCE','Y');
...and restart capture process, the required_checkpoint_scn changes, and capture process scan only newest archive redo logs.
but my question is: in what situations required_checkpoint_scn changes forward during normal database processing (without enforcing checpoint).
|
|
|
|
|
|
| Re: streams and required_checkpoint_scn [message #543027 is a reply to message #543017] |
Sun, 12 February 2012 12:02  |
John Watson
Messages: 8964
Registered: January 2010
Location: Global Village
|
Senior Member |
|
|
Quote:size of transaction and parameter _CHECKPOINT_FREQUENCY.
value for this parameter was 1500 that means every 1.5GB capture process makes checkpoint.
i changed it for 1M and now checkpoint performs frequent. Thank you for this information, I knew the update was not in real tme but I did know what controls it.
As for first_scn, you can use dbms_capture.alter_capture to adjust the checkpoint_retention_time to a certain number of days. This will force Oracle to advance the first_scn automatically up to the archive log from that far back. I think the default is 60 days.
|
|
|
|
| Re: streams and required_checkpoint_scn [message #543036 is a reply to message #543027] |
Sun, 12 February 2012 13:57 |
|
piotrtal
Messages: 168
Registered: June 2011
Location: Poland/Czestochowa
|
Senior Member |
|
|
i have set checkpoint_retention_time as 0.00694444444444444 what means that every 10 minutes dba_capture.first_scn should go forward.
but it doesn't.
... and i have not idea what this "pushing forward" is dependable from.
for me it changes unpredictable. i spend several hours in my test enviroment with testing it and finally i surrender.
[Updated on: Sun, 12 February 2012 13:58] Report message to a moderator |
|
|
|
| Re: streams and required_checkpoint_scn [message #543039 is a reply to message #543036] |
Sun, 12 February 2012 15:41 |
John Watson
Messages: 8964
Registered: January 2010
Location: Global Village
|
Senior Member |
|
|
My understanding of this is that the first_scn will advance to the first_change# of the archivelog that covers the check_point_retention time. Every 10 monutes is really very short, and many databases do not log switch so frequently, so you will not get a very close correlation with the check_point_retention time. You could try a test: temporarily set archive_lag_target=60 to force log switches every minute, and see if that gives you the result you expect.
But I could be wrong - I don't happen to be working on a streaming database at the moment, so I'm going by memory.
|
|
|
|
] 
Blog
Search
Help
Members
Register
Login
Home
