Checkpoint not complete.

The "checkpoint not complete" messages are generated because the logs are switching so fast that the checkpoint associated with the log switch isn’t complete.

During that time, Oracle’s LGWR process has filled up the other redo log groups and is now waiting for the first checkpoint to successfully execute.Oracle will stop processing until the checkpoint completes successfully.

you see this "checkpoint not complete" error in your alert log:

/u01/app/oracle/mydb/redo5.log
Thu Oct 05 10:50:46 2012
Thread 1 cannot allocate new log, sequence 55998
Checkpoint not complete

Oracle suggests that you might consider setting archive_lag_target to zero to reduce "checkpoint not complete" messages:

alter system set archive_lag_target=0 scope=both;

Overall Oracle performance can be dramatically improved by increasing the log sizes so that logs switch at the recommended interval of 15 to 30 minutes.

Identify the current size of the redo log members from v$log, record the number of log switches per hour and increase the size of the log to allow Oracle to switch at the recommended rate of one switch per 15 to 30 minutes.


Suppose there are 4 redo buffers named as redo1,redo2,redo3 and redo4. At every log switch the redo buffer's content are written to the online redo log by lgwr.At every check point the dirty block in the data buffers are written to the datafile.
For example for a successful log switch from redo1 to redo2 occurs only if
i)the contents of redo2 are written to the redo log file
ii)the last SCN of the redo buffer has been written to the data file during the last checkpoint.

If point ii) fails then its gives the warning that check point not found.
Then the lgwr forces the checkpoint and dbwr writes the dirty block to data file.
If the redo buffer size is too small then the log switch occurs very frequently and it gives the warning checkpoint not found and forces the checkpoint to execute more frequently and hence the performance problem occurs.