Re: checkpoint over 15 minutes?

Don't need to look but I remember by heart - from seconds to minutes. Peak time in active production databases about a minute for incremental checkpoint.

It depends, for example, on cache size, number of DBWriters (or slaves) and async IO on/off, fast_start_mttr_target, number of datafile. Of course, a lot depends on you IO subsystem performance.

Are you concerned? Unless you see the impact on perfromance than it will only affect crash recovery time. For example, you can have a lot of "free buffers" waits if DB writers can't keep up flushing dirty blocks and checkpoiting (I observed that on one of our systems).

2006/5/23, LS Cheng <exriscer_at_gmail.com>:
> Do you mind have a look at your alerts (if you have log_checkpoints_to_alert
> set to TRUE) and see how long does your CKTP take :-?

-- 
Best regards,
Alex Gorbachev

http://oracloid.blogspot.com
--
http://www.freelists.org/webpage/oracle-l