OCP 11gR2: Administration II 練習筆記 (三)

Lession 03
=================================================
01. backup as backupset
- RMAN backup method are [backupset | copy] default is backupset.
- backup command run by serial

- Whole database backup command is "backup database plus archivelog"
  plus archivelog purpose is maintain log folder size.

- backup as copy [datafile|tablespace|database|archivelog];
  only on disk and not in FRA
  show records by "list copy"
  it support incremental backup



02. RMAN backup types
- backup database; <- Whole database backup.
- backup incremental level 0 database; <- Full database backup.

- backup incremental level 1 database; <- Incremental backup base on level 0.
  Level 1 with 2 types [differential  | cumulative] default is defferential

SUN	MON	TUE	WED	THU	FRI	SAT
LV0	LV1	LV1	LV1	LV1	LV1	LV1
D	<------	<------	<------	<------	<------	<------	24HR/DAY
C	<------						24HR/DAY
C	<--------------------						48HR/DAY
C	<----------------------------------						72HR/DAY
C	<------------------------------------------------						96HR/DAY
C	<--------------------------------------------------------------						120HR/DAY
C	<-----------------------------------------------------------------------------						148HR/DAY

03. Fast incremental backup
- not support save log in FRA
- block change tracking default is off
- enable command "alter database enable block change tracking using file 'location';
  after enable, it will not affect on coming incremental backup, it effective on next.

- OMF, parameter db_create_file_dest
  include [datafile, tempfile, redo, controlfile, block change tracking]



04. Proxy copy (OCP exam not test)
- backup to SAN via MML (media manage libarary)



05. create duplexed backup
- duplex support disk and sbt
- duplex not support image
- duplex cannot backup in FRA, must use "FORMAT"
  backup as backup set copies 2 database format '/tmp';
  or by configure



06. backup the backupset and FRA
- move all backupset to sbt by command "backup device type sbt backupset all;"
- backup FRA by command "backup recovery area;" or "backup recovery files;"


07. backup readonly tablespace (do not use this function)
- consider retention policy with backup optimization
- skip readonly no relationship with optimization
  the command is "backup database skip readonly;"



08. Multisection backup
- if datafile very large, use multisection to limited backuppiece size
- fast complete backup with parallel (common use)
- use multisection with parallel command as below:

RMAN> configure device type disk parallelism 4;
run {
allocate channel ch1 device type disk format '/tmp/bak1_%U';
allocate channel ch2 device type disk format '/tmp/bak2_%U';
allocate channel ch3 device type disk format '/tmp/bak3_%U';
allocate channel ch4 device type disk format '/tmp/bak4_%U';
backup section size 500M database; }

- if you have only 1 disk and config parallel to 4, the performance will be affect.



09. Archival backup
- backupset not affect by retention policy
- cannot use on FRA
- change backupset to archival backup as below:
  change backup tag 'name' keep forever;



10. RMAN report & dynamic view
report schema;
report need backup;
report obsolete
v$backup_set;
v$backup_piece;
v$datafile_copy;
v$backup_files;

11. Recovery Non-critical & critical recovery

Tempfile lost:

- Cannot sort table from "order by", also EM will have error

- automatic re-create of tempfile when restart <- if file corrupted, restart cannot fix

- manual re-create:

 "alter tablespace temp add tempfile '/location/temp02.dbf' size 500M autoextend on;"

 "alter tablespace temp drop tempfile '/location/temp01.dbf';"

Loss of a redo log group:

- current log corrupted and instance crashed, need cancel-based point in time recovery.

- current log corrupted but instance not crash. Perform alter log switch and checkpoint.

- inactive & active log corrupted, use clear log command.

        Below command for reference:

        "alter database clear logfile group [1|2|3];"  <- inactive status run, no data loss.

        "alter database clear unarchived logfile group [1|2|3];" <- active status run, when recovery will data loss.

        "alter database clear unarchived logfile group [1|2|3] unrecoverable datafile;" <- if the datafile offline and     need to clear unarchived logfile you need to add unrecoverable datafile

Index tablespace recovery:

- if datafile lose only contains indexes, no need perform recover task.

 Drop datafile, drop the tablespace(index only), re-create the index tablespace, re-create indexes that in tablespace.

- use parallel to reduce time for re-create index

 create index 'name' on 'schema.table'('column') parallel 4;

all controlfile loss

- open resetlogs

Readonly tablespace

- take tablespace offline

- restore tablespace

- bring tablespace online

No logging object

- drop it and re-create

 Use report unrecoverable

 Drop table 'table_name';

 Create table 'table_name' nologging;

 Insert /*+ append */ into 'table_name' select * from 'source_table';

Authentication for database

- SYS can use O/S auth but other role cannot, they are maintenance by password file.

 Recovery or reset

 orapwd filename='/location/name.ora' password='password' enries=10

 re-grant role to user

12. Complete & Incomplete recovery

- complete recovery - no data loss

                                - need datafile, archivelog, redo log and undo tablespace apply

- incomplete recovery - bring database to point in time recovery

                                - no need redo log and some archive missing

                                - undo tablespace apply and resetlogs to open db.

OCP 11gR2: Administration II 練習筆記 (四)
OCP 11gR2: Administration II 練習筆記 (二)