Re: Help: huge amount of redo log switches

> > Hello,
> >
> > I have a problem on my development database. It's generating huge
> > amount of redo log switches CONSISTENTLY which give me a large amount
> > of archive logs(2GB a day).

I'm sorry. I'm in a bloody mood, I guess. But databases simply do not suddenly take it upon themselves to start generating huge amounts of redo. A mere modicum of logical thought would reveal the truth of the matter: someone or something on YOUR side of the fence is causing redo to be generated.

>>Why I say consistent is because it starts
> > generating at 7am to 5pm on every 3 minutes basis EVERY DAY.
> >
> > It's a very small database(only 1GB) with multiple schemas and the only
> > users are developers. log_buffer=32768 and two sets of redo logs each
> > 4M. I've asked my developers and none of them say they have a job
> > running since 7am every day.

So then of course it must be the database's fault. Natch. If the developers say it's not them, clearly the huge Oracle Corporation has developed a dog of a product that can't be trusted.

>>And I don't think there is any large batch
> > job because it's a development db.
> >

And your 'thinking' this is so makes it so, huh Captain?

> > I tried resizing(increasing and decreasing) log_buffer and redo log file
> > size but it doesn't help.

I suspect that you should qualify that statement. You increased the size of your redo logs and it made precisely NO difference??? I suspect you are being economical with the truth, or someone is playing you false. If it takes 3 minutes to fill up 4M, then doubling the log file size would mean (ceteris paribus) that it would take 6 minutes to activate a log switch. Yet you claim ZERO difference?

The ONLY reason there could be zero difference in the time taken to fill up the double-sized logs is if the transaction rate had just doubled as well.

>>I think I need to identify the db
> > transactions which are responsible for all those redo logs.
> >

That would indeed be a start

> > 1GB database with 2GB archive logs doesn't make sence to me!

Why not?! A 1Gb database presumably means that there is 1Gb of data in the tables. If every piece of that data was updated simultaneously, you'd still end up with a '1Gb database', but 1Gb of redo would have been generated. If every piece of the data was updated and then the updates rolled back, you'd generate 2Gb of redo -and there would still be 1Gb of data sitting in the tables.

This is not rocket science

HJR
>>Any help
> > is appreciated!
> >

> > Susana
> >
> > Sent via Deja.com http://www.deja.com/
> > Before you buy.
>
> Someone/something is generating a lot of redo...look at v$sesstat to
> identify which sessions are writing lots of redo...
>
> ... Then start tracing those ones to see what jobs need to be looked ,or
> which developer needs to be paid a visit.. :-)
> --
> ===========================================
> Connor McDonald
> http://www.oracledba.co.uk
>
> We are born naked, wet and hungry...then things get worse
Received on Wed Aug 09 2000 - 00:00:00 CDT