Re: Redo Log Status

Hi:

Oracle marks redo logs as stale when the Oracle software cannot determine if the contents of a particular redo log are valid. This is one reason that you mirror redo logs at the Oracle level, even if the redo logs are on logical drives that are mirrored at the hardware level. If the Oracle software has any write errors to a redo log, it will mark it as stale. If your case, sounds like it had write errors on both members of a particular group. A stale redo log cannot be archived (since it might worthless for recovery). I've never seen more than one member get marked stale. (Oracle will simply archive the redo log that isn't stale.) Since both members were marked stale, I would assume that you have both members on the same drive. You should move one member of each group to a different drive, then you shouldn't have this problem again. I'd suggest possibly even adding a 3rd member to each group (also on a different drive) if you are using non-raided drives for redo logs. Make sure you get a backup of your database after any stale logs that could not be archived (otherwise, you may have some serious difficulties trying to recovery if needed.)

Try performing ALTER SYSTEM SWITCH LOGFILE until you cycle through the stale redo logs. If they are still marked STALE, then drop that group and re-create it, then cycle through again to make sure it shows as valid.

Assuming all your members are currently on the same drive, this could be an indication of an impending drive failure, which is all the more reason to move/add members onto other drives. And, get a hot or cold backup!

Hope this helps.

Dave Sisk, OCP DBA
davesisk_at_ipass.net
http://www.ipass.net/~davesisk/ORACLEonNT.htm

Anne-Marie Ternes wrote in message <36554F86.689ABFDA_at_cie.etat.lu>...
>Hello Oracle fellows,
>
>A few days ago, I saw in my alert log a message like : "can not allocate
>redo thread xy, redo needs archiving".
>I added a new redo group with two redo members, so now I have three redo
>groups with each having two members. There have been no more alert
>messages of this kind.
>
>Today I checked the V$LOG table, and the status field told me that
>group1 and group2 are "inactive". What is the exact difference between
>"inactive", "active", and "unused"? What should be the ideal status of a
>redo group that is not the current redo group?
>I also checked the V$LOGFILE table, and there the status field tells me
>that both members of redo group 2 are "stale". According to the Oracle
>Server Reference book, this means that the "file's contents are
>incomplete". That's very nice, but it does not tell me what I should do
>next.
>
>Could somebody please explain to me in detail what is happening to my
>redo logs, and give me suggestions as to what I can do to get back to
>normal?
>
>Thank you very much for any help,
>
>Anne-Marie Ternes
>
Received on Fri Nov 27 1998 - 00:00:00 CET