CKPT checkpoint queue

The buffer cache memory of the CKPT checkpoint queue contains the chain and bufferLRU to concatenate available blocks (based on the Access Frequency). LRUW concatenates dirty blocks (Access Frequency) and cold-end blocks are first written to the hard disk. When CBC reads data blocks into the memory, it uses CKPT (chain): 1. It links dirty blocks. 2. It links the first dirty time (from small to large) the address of the RBA redo block address log block will generate a log when the block is modified. The log has an address. This address is RBA. At the same time, the RBA address is stored in the modified block record. There are two addresses for each block: LRBA (the first dirty address) HRBA (the last dirty address) the data blocks in the checkpoint queue are linked by the LRBA address of the data blocks. CKPT works in two ways: 1. Full checkpoint: it will trigger DBWR to write all dirty blocks to the disk (close the database) 2. incremental checkpoint: a. Record the LRBA (log address corresponding to the first dirty block) of the checkpoint queue to the control file (every 3 seconds) b. When an incremental checkpoint occurs, it checks the checkpoint queue. When there are too many dirty blocks in the queue and the system is not very busy, dbwr is triggered to shorten the checkpoint queue. Dbwr writes two types of data: 1. LRUW 2. The earliest dirty parts of the checkpoint queue are written on the disk.

The last log address saved by the on disk rba logbuffer database (the last log address in logbuffer is the latest log address recorded) col LowRBA for a20col OndiskRBA for a20select CPDRT, CPLRBA_SEQ | '. '| CPLRBA_BNO | '. '| CPLRBA_BOF "LowRBA", CPODR_SEQ | '. '| CPODR_BNO | '. '| CPODR_BOF "OndiskRBA", CPODS, CPODT, CPHBT from x $ kcccp; alter system flush buffer_cacheCPDRT column is the number of dirty blocks in the checkpoint queue. the CPODS column is the scn CPODT column of on disk rba and the timestamp of on disk rba. The CPHBT column is the heartbeat.

 

Related Operations

select checkpoint_change# from v$databasealter system checkpointalter system switch logfileselect name,checkpoint_change# from v$datafileselect name,checkpoint_change# from v$datafile_headerselect * from v$log;beginfor i in 1..10000 loopinsert into  t2 values(1,'xkj');commit;end loop;end;select * from t2