Re: Redo Log help

Rogbaker wrote:
>
> We are using Oracle 8.1.7 and are in the process of a PeopleSoft
> upgrade test move. We have an APP Engine program running very long and
> and sql script running very long simultaneously. Looking in the alert
> log, I have seen multiple
> 'Cannot allocate new log' and 'Checkpoing not complete' messages.
> These upgrade scripts/programs do move a ton of data. I have added
> more 2 more log groups, which seem to have handled the error messages,
> but how frequently should the database advance to a new log group? It
> is happening like every 30 seconds.
> Right now I have:
> groups #1-5 with 2 10M members
> group #6 with 2 15M members
> group #7 with 2 20M members
>
> Thanks for any help,
> Roger Baker
> Seminole Co. Public Schools

'Checkpoint not complete' basically means checkpointing could not finished before redo logs wrapped around. So there are two plans of attack:

1. make checkpointing faster: better I/O, better load balancing etc etc
2. make redo logs take longer to wrap: Maybe have (say) 6 groups of 200m or so

Many people just advocate (b) but that sometimes just masks an underlying I/O deficiency

hth
connor

-- 
==============================
Connor McDonald

http://www.oracledba.co.uk

"Some days you're the pigeon, some days you're the statue..."