On 01/30/2004 01:24:26 PM, A Joshi wrote:
Hi,
In view v$log there is a column status. This changes from current
(if the redo log is in use) to atcive then to inactive. Documentation
says :
ACTIVE: The log is active but is not the current log. It is needed
for
crash recovery. It may be in use
There is a pretty good explanation in this usenet thread:
Mladen,
Thanks for info. So all the dirty blocks need to be written to disk after eachcheckpoint. After that is donethe status becomes 'INACTIVE'. Just that sometimes this is very unpredictable.
My question : Ifa log switch always causes a implicit checkpoint then what is the need for this
David Hau explained this much better and in much more clear
fashion then me. Here is the most important part from the usenet
thread he was referring you to:
**
This makes sense if you think about where the various v$ dynamic
Thanks Mladen and David
It is great to get such detailed info.
If we take your example of three redo logs then you could have a situation where one is is 'CURRENT' status and two are in 'ACTIVE' status. As follows :
GROUP# STATUS-- 1 ACTIVE2 CURRENT3 ACTIVE
Now if group
LGWR will always write to the redo-logs in a round-robin
fashion. So, on completing Group2, it would wait
for Group3 to be INACTIVE. {One exception is when you ADD a new Log
Group. If you added a new log
file without specifying a Group#, it would be allocated Group#4, would be
set to status