Currently, database backup is performed using RMAN from shell script.
The shell script has has 2,347 lines. Does anyone KISS anymore?
Anyhow, I recalled very early in my career when someone coined the phrased, “Doing It The Dinh’s Way”.
Basically, it’s not effective nor efficient to debug and troubleshoot shell script.
Client provided the following info and off I went.
Database backup changed from filesperset 4 to filesperset 1.
Rollback change and still facing performance issues.
Level 1 increment backup with filesperset = 4 still running > 24 hours – killed.
Can you try a few things not using the rman script in crontab, but using your own rman command?
Request was to disable block change tracking – DONE.
Database size is 272G with the following datafiles.
SQL> select file_id, file_name, round(bytes/1024/1024/1024) gb from dba_data_files order by 1 asc;
FILE_ID FILE_NAME GB
-------- ------------------------------------------------------------ --------
1 +DATA_XXXDWP01/XXXdwp01/datafile/system.256.964878417 1
2 +DATA_XXXDWP01/XXXdwp01/datafile/sysaux.257.964878419 4
3 +DATA_XXXDWP01/XXXdwp01/datafile/undotbs1.258.964878421 21
4 +DATA_XXXDWP01/XXXdwp01/datafile/gg_tbs.261.964878453 5
5 +DATA_XXXDWP01/XXXdwp01/datafile/undotbs2.262.964878457 19
6 +DATA_XXXDWP01/XXXdwp01/datafile/users.264.965655143 31
7 +DATA_XXXDWP01/XXXdwp01/datafile/users.267.1020339295 32
8 +DATA_XXXDWP01/XXXdwp01/datafile/dev_users.268.1027263011 0
9 +DATA_XXXDWP01/XXXdwp01/datafile/dev_users.269.1027263013 0
10 +DATA_XXXDWP01/XXXdwp01/datafile/dev_users.270.1027263013 0
11 +DATA_XXXDWP01/XXXdwp01/datafile/dev_users.271.1027263013 0
12 +DATA_XXXDWP01/XXXdwp01/datafile/dev_users.272.1027263015 0
13 +DATA_XXXDWP01/XXXdwp01/datafile/users.274.1046508647 32
14 +DATA_XXXDWP01/XXXdwp01/datafile/users.275.1056870315 32
15 +DATA_XXXDWP01/XXXdwp01/datafile/users.276.1066387693 32
16 +DATA_XXXDWP01/XXXdwp01/datafile/users.277.1066387695 32
17 +DATA_XXXDWP01/XXXdwp01/datafile/users.278.1066387697 32
17 rows selected.
SQL>
Backup completed in 3.25 hours using MAXPIECESIZE 2G MAXOPENFILES 1.
Believe further improvements by increasing MAXPIECESIZE to 4G; however, client wanted to leave as is.
Backupset size is 24G.
I believe performance may be related to Pure Storage Deduplication; however, there’s no more time nor different destination to test with.
MAXOPENFILES ≤ 4
Each buffer = 1MB, total buffer size for channel is up to 16MB
MAXOPENFILES = 1 => 16 buffers/file, 1 MB/buffer = 16 MB/file
Optimal for ASM or striped system
Recovery Manager: Release 19.0.0.0.0 - Production on Fri Jul 16 11:08:18 2021 Version 19.8.0.0.0
RMAN> connect target;
2> set echo on
3> spool log to backup.log
4> show all;
5> run {
6> allocate channel c1 device type disk format '/backup/XXXDWP01/%d_%I_%T_%U_%s' MAXPIECESIZE 2G MAXOPENFILES 1;
7> allocate channel c2 device type disk format '/backup/XXXDWP01/%d_%I_%T_%U_%s' MAXPIECESIZE 2G MAXOPENFILES 1;
8> allocate channel c3 device type disk format '/backup/XXXDWP01/%d_%I_%T_%U_%s' MAXPIECESIZE 2G MAXOPENFILES 1;
9> allocate channel c4 device type disk format '/backup/XXXDWP01/%d_%I_%T_%U_%s' MAXPIECESIZE 2G MAXOPENFILES 1;
10> backup incremental level 1 filesperset 1 database;
11> }
12> exit