Redo logs

[itwizzkid]

I was getting checkpoint not complete so I followed advice and created a couple of logs using following SQL

ALTER DATABASE
ADD LOGFILE GROUP 5 ('/oradata/ORD/redo05.log', '/oradata/ORD/redo05m.log') SIZE 10240K

This seems to have worked but when I checked the location of the disk where the logfile should be I could only find redo05.log. The other pair does not seemed to have been created ot does not exist.

But looking at the alert_log it should

Wed Aug 7 06:45:52 2002
Thread 1 advanced to log sequence 5708
Current log# 5 seq# 5708 mem# 0: /oradata/ORD/redo05.log
Current log# 5 seq# 5708 mem# 1: /oradata/ORD/
redo05m.log

How can I correct this. If a recovery is needed will this be a problem?

Thanks in advance

[julian]

Could you run first this

col member for A40
select * from v$logfile order by group#;

and post the output.

BTW, why have you tried to create the 2 members in the same physical directory?

[itwizzkid]

Thanks for your reply Julian. I now fixed the problem. It looks like a carriage return was inserted in the middle of the file name.

As to your question about everything being in one volume, this was something done by my company with imput from HP before I got here. Apparently every thing is triple mirrored.

[itwizzkid]

Julian,


select * from v$logfile order by group#

GROUP# STATUS MEMBER
---------- ------- ----------------------------------------
1 /oradata/ORD/redo01.log
1 /oradata/ORD/redo01m.log
2 /oradata/ORD/redo02.log
2 /oradata/ORD/redo02m.log
3 /oradata/ORD/redo03.log
3 /oradata/ORD/redo03m.log
4 /oradata/ORD/redo04.log
4 /oradata/ORD/redo04m.log
5 /oradata/ORD/redo05.log
5 /oradata/ORD/
redo05m.log
10 rows selected

[pando]

Originally posted by itwizzkid
Apparently every thing is triple mirrored.

triple mirrored so what, it does not prevent you logical corruptions

[itwizzkid]

Pando,

The history behind this was that, they are using business copy and if anything should happen to the database for instance the database crashes, they will restore from backup then replay missing data from the application. The application that were are using allows this with no data loss.

When I arrived here I argued against this but since we are looking after the database for a client of us, we can only make suggestions to them and thet make the business decissions.

The HP boxes are based in Atlanta looked after by HP. To date I have been requesting the disk architecture with not much success

[julian]

Originally posted by itwizzkid

5 /oradata/ORD/
redo05m.log

It seems that you have pressed ENTER after ORD/ and I would suggest that you drop this member and recreate it:

alter database drop logfile member /oradata/ORD/
redo05m.log';

You may have to run alter system switch logfile; before the drop command if group 5 is current.

Then run:

alter database add logfile member '/oradata/ORD/redo05m.log' to group 5;

[JMac]

Re: earlier comment about logical corruptions in Redo Logs ...

If one Group member was logically corrupt - which group member would be written to the Archive destination? And how can Oracle detect a logical corruption?

[itwizzkid]

Thanks Julian,

I have already done this