Re: Message in alert file

The system is attempting to advance into a redo log that has not had its checkpoint successfully finished (ie, the log to which you are wanting to move is still sitting there with a status of "ACTIVE"). That's not allowed, so the system actually hangs until the checkpoint completes.

That will happen if the number of Redo Log groups is insufficient (ie, by adding a new group or several, you switch to other logs instead of trying to switch back on top of yourself. Theoretically, if you add enough groups, you'll give the system long enough to complete the original checkpoint.

So: first cure... how many redo log groups have you already? If you've only got the minimum of 2, start adding extras. When you have around half a dozen, I suggest you stop adding log groups, and consider starting 'intra-log' checkpointing (ie, use log_checkpoint_interval and log_checkpoint_timeout to trigger extra checkpoints -this way, the checkpoint at the log switch only has the last bit of the file to worry about).

There is a theoretical limit of 32 log groups -I'm suggesting that managing a system with more than around half-a-dozen gets a bit dodgy, and hence you take alternative routes. However, extra checkpointing will cause performance slow-down, and so you might want to keep adding as many log groups as necessary, up to the 32 limit, before adopting that approach. (Incidentally, just to further muddy the waters, your own database may not permit the maximum of 32 log groups to be added -there's actually a parameter set at database creation time which determines how much room in the Control File is reserved for pointers to the log groups -MAXLOGFILES I think from memory. You may need to increase this for your database, which involves editing the backup controlfile to trace script, and re-creating your control files. But we're getting way ahead of ourselves there!)

For now, add a couple more groups and see how things go. Don't forget to mirror them, just as (I hope!) your current groups are mirrored.

Incidentally, this problem has nothing to do with the size of the logs... A bigger file takes longer to switch away from, that's true... but it also takes longer to checkpoint.

Regards
HJR

--
--------------------------------------------------------------------------
Opinions expressed are my own, and not those of Oracle Corporation
Oracle DBA Resources:               http://www.geocities.com/howardjr2000
--------------------------------------------------------------------------



"Igor V. Podkorytov" <piv_at_countrymen.omsk.su> wrote in message
news:39D2D067.45946992_at_countrymen.omsk.su...


> Please, help me to identify the problem.

> i've got the message

> "Thread 1 cannot allocate new log, sequence 2444

> Checkpoint not complete"

> Sizes of redo log files are large enough.

>

>

>