Regarding Checkpoint & Dirty Buffers
It's very misleading. Some explainaton say that "During checkpoint, dirty buffers are written into the disk" and at someother point we also come across statements like "A checkpoint occurs when the dirty buffers are written to the disk". I mean, which one of them is true? There is a sea of difference between the two hypothesises. Can someone clarify my obscurity?
Many thanks in advance..
Re: Regarding Checkpoint & Dirty Buffers
Hi,
The truth is "During checkpoint ,dirty buffers are written to the disk".From Oracle 7.3 ,there is a event called the checkpoint event.This event occurs because DBWR is very lazy in writing the dirty database buffers from database buffer cache to the datafiles.Whenever a checkpoint happens,LGWR writes the contents of the redolog buffer to the online redolog files.It then signals DBWR to write the contents of the database buffer cache to the datafiles. Also checkpoint occurs due to various situations such as a log switch,when the database is shutdown normal,transactional or immediate option,when u put ur tablespace in read-only mode or when you take ur tablespace offline with normal option or through checkpoint parameters in ur init.ora parameter file i.e LOG_CHECKPOINT_TIMEOUT and LOG_CHECKPOINT_INTERVAL.
Regards,
Rohit nirkhe,oracle/Apps DBA,OCP 8i
oracle-support@indiatimes.com
Regarding Checkpoint & Dirty Buffers
Quote:
Originally posted by Sameer
Its wrong... By writting dirty blocks to datafile doesn't force or occure checkpoint. You can force checkpoint to write dirty blocks to datafile.. The question is quite well answered by Sanjay.
It is definetely under the auspicious of the LRU mechanism but the Dirty blocks are subjected to the mechanism of MRU due to which the checkpoint is enforced writing the dirty block back to the datafile.
So far very good discussion.
VAST