redo logs

[manjunathk]

Hi friends,
In my alert_log file, I often get the following message.

Thread 1 cannot allocate new log, sequence 1030
Checkpoint not complete
Current log# 2 seq# 1038 mem# 0: /tools/oracle/database/dev/redo_2.log

What is wrong and how does it may affect the performance?

Thanks
manjunath

[uday]

1. How many redo logs do you have in your system.

2. How often are they switching ( Whats the average time between switches )

3. How large are they

uday

[manjunathk]

Details of redo logs are as under:

GROUP# THREAD# SEQUENCE# BYTES MEMBERS ARC STATUS FIRST_CHAN FIRST_TIME
---------- ---------- ---------- ---------- ---------- --- ---------------- ---------- --------------------
1 1 1179 52428800 1 NO INACTIVE 3263039 09-MAR-01
2 1 1180 52428800 1 NO CURRENT 3263961 09-MAR-01


they are switching every minute.

[jmodic]

You have only two redo groups, which is not enough for your database activity. Are they switching ecery menute on average or only during peek hours? If they realy do switch so often, you should consider enlarging your redo log files.

I would advise at least 4 groups of redo logs, I would also enlarge them from current 50M to at least 200M. It also appears that you only have one member in each group. This is not safe at all, consider adding another member to each group, if possible locate them on another disk drive, possibly under different controler.

You also obviously run in noarchive mode. If yor database crashes you won't be able to recover your database, you will only be able to restore it from your last cold backup.

[uday]

I agree with jmodic

I would even go further due to your noarchivelog situation and increase each log to 25M so that it switches at alteast a 5 min interval.

also consider adding 6 groups so that you always have 30 min between overwrites of the online redo logs.

If you make the redo logs the right size, the checkpoints due to log switches decrease, hence better performance