Re: checkpoint not complete

From: Neil Greene <ngreene_at_ni.net>
Date: 1996/06/01
Message-ID: <31B0A62E.788B_at_ni.net>#1/1

vermaat_at_gemini.tfdl.agro.nl wrote:
>
> In article <31add637.5284444_at_news2.ios.com>, chuckh@dvol.com (Chuck Hamilton) writes:
 

>
> What happens is that the checkpoint interval has a value that does not match
> exactly with the size of the redo log file. Therefore the redo logg "session"
> arrives at the end of the file while the interval is not filled completety.
> What happens is that a message is put in the ALERT.LOG-file and a checkpoint is
> forced to jump to the next redo log file.
> In fact it is just an informational message - nothing to worry about.

I would have to disagree with the comment "nothing to worry about". If you have a heavy OLTP environment, you do not want these errors. This is a tunning issue which can lead to a tunning problem. Oracle dumps the log buffers to disk. These buffers must be flushed before Oracle will reuse an online redo log. If this has not been completed, Oracle will not switch to the next redo log group. Thus, producing the error message.

Consider running the checkpoint process (CKPT), adding redo logs, or tunning your LOG_CHECKPOINT_INTERVAL. Chances are, if you see an error, it is an error. Best to fix these errors before they become real problems.

---

Neil Greene
Sr Oracle DBA / Unix Admin
SHL Systemhouse, Inc.
email: ngreene_at_laoc.SHL.com Received on Sat Jun 01 1996 - 00:00:00 CEST