| Oracle REDO log switching - query / help [message #230174] |
Wed, 11 April 2007 02:35  |
kumarpk108
Messages: 7
Registered: December 2006
|
Junior Member |
|
|
Hi all,
We are running Oracle 9.2.0.7 / SAP 4.6C.
Recently we performed the oracle version
upgrade from 9.2.0.6 -> 9.2.0.7.
Our REDO log grouping as follows:
LOGFILE
GROUP 11 (
'/oracle/C11/origlogA/log_g11m1.dbf',
'/oracle/C11/mirrlogA/log_g11m2.dbf'
) SIZE 400M,
GROUP 12 (
'/oracle/C11/origlogB/log_g12m1.dbf',
'/oracle/C11/mirrlogB/log_g12m2.dbf'
) SIZE 400M,
GROUP 13 (
'/oracle/C11/origlogA/log_g13m1.dbf',
'/oracle/C11/mirrlogA/log_g13m2.dbf'
) SIZE 400M,
GROUP 14 (
'/oracle/C11/origlogB/log_g14m1.dbf',
'/oracle/C11/mirrlogB/log_g14m2.dbf'
) SIZE 400M
Last year we had an issue with Oracle redo log switching
where in the logs were switched in the following way.
Group Thread Sequence Size Nr of Archive
Nr Nr Nr (bytes) Members Status
12 1 84606 419430400 2 YES
14 1 84607 419430400 2 YES
13 1 84608 419430400 2 YES
11 1 84609 419430400 2 NO
12 -> 14 -> 13 -> 11 which is logB -> logB -> logA -> logA
Our understanding is that oracle should write to one redo log,
once filled switch to another redo log on a different file system.
This is not the case for C11. We switch (to write in new)
& read (to archive out) on the same file system as specified above.
Last year we did an experiment and dropped and recreated LOGFILE
groups and the redo log sequence was put in correctly. i.e
12 -> 13 -> 14 -> 11 which is logB -> logA -> logB -> logA
When we examined the redo log sequence recently, it is again
showing the older way - which is
12 -> 14 -> 13 -> 11 which is logB -> logB -> logA -> logA
and that is not the best option.
We would like to know what made this switch happen
even though we manually dropped/recreated the
logfile groups and ensured the switching in an optimal way.
Is there any parameter or v$ table entries to examine whether
this type redo log switching has any issues.
Also what will be implications of this type of
switching - will it adversely affect the log
switching and hence the wait time if we leave
it like this. What kind of performance implication
can this result in?
What would be the recommended steps to effect
the change to logfile groups so that the redo log
switching happens in a more optimal way i.e
12 -> 13 -> 14 -> 11 which is logB -> logA -> logB -> logA
Any help on the subject will be greatly appreiciated.
Thanks
Kumar
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
] 
Blog
Search
Help
Members
Register
Login
Home
