Notes Link
Notes Link
1) Martins Blog
https://martincarstenbach.wordpress.com/2012/06/20/little-things-worth-knowing-
static-and-dynamic-listener-registration/
2) Dominic
https://orastory.wordpress.com/
https://community.oracle.com/message/9362003#9362003
SQL_TRACE (10046), TKProf and Explain Plan - Overview Reference (Doc ID 199081.1)
For Install Support & Certification you can refer this note:
https://community.oracle.com/community/support/support-blogs/database-support-
blog
Question: Direct path read temp
https://community.oracle.com/thread/3695655?start=0&tstart=0
https://community.oracle.com/community/support/support-blogs/database-support-
blog/blog/2014/10/16/oracle-database-11g-release-2-patchset-11204-availability
In DWH there are two main concepts: 1. Kimball 2. Immon. The book from latter is already suggested,
you can read The Data Warehouse Toolkit - Kimball & Ross. You can also search Google for "Kimball
Vs. Immon" to read the battle of ideas.
Go to Oracle Support and search patch set - p10404530 and you can find both;
1. You can upgrade current 11.2 home to 12.1 using DBUA or Manual. Install 12.1
and start DBUA.
2. Create a new PDB in a CDB, use expdp (11.2) /impdp (12.1) to copy the data
across.
3. Transport Database,
4. You can use Golden Gate
http://docs.oracle.com/database/121/UPGRD/toc.htm
How to find Objects Fragmented below High water mark (Doc ID 337651.1)
Script to Report Tablespace Free and Fragmentation (Doc ID 1019709.6)
https://community.oracle.com/thread/3684816
Nologging can only help for "create table", "alter table move", direct path load on tables, "create
index", "alter index rebuild" anyway. It can't help on a delete. And for the direct path load (also insert
/*+ append */) will eliminate (most of) the logging on the table - index maintenance operations will still
be logged.
Question: Index Monitoring
Enabling Index monitoring won't tell you if that index has been used in hour X or even in the last day.
It will only tell you that the index has been used since monitoring was turned on. So it won't be able to
tell you that the index was only used during business hours. Its possible that the index was used off-
hours. But does that matter? If the index is used, then presumably it is needed, business hours or not.
https://community.oracle.com/thread/2160267
set pages0
spool enable_monitoring.sql
select 'set echo on' from dual;
select 'alter index ' || index_name || ' monitoring usage ;'
from user_indexes
order by 1;
spool disable_monitoring.sql
select 'set echo on' from dual;
select 'alter index ' || index_name || ' nomonitoring usage ;'
from user_indexes
order by 1;
spool off
But enabling monitoring is not the right thing to do for at least --3-- 4 important reasons :
1. It can falsely capture an index that is being incorrectly used by the optimizer -- i.e. a sub-optimal
that the optimizer selects (for whatever reasons, including bad statistics)
2. It can falsely capture an index that is used only once in one query out of a million queries that have
been executed
3. It can falsely NOT capture an index that is used only on specific days -- e.g. month-end reports
which require a few specific indexes not used for the 28 other days of the month
4. It can falsely NOT capture indexes that are required for constraints, simply because they are not
used in queries (although they are important to maintain data consistency !)
5. It can falsely NOT capture FK indexes that are used by Oracle behind the scene to avoid the
locking threat
[http://richardfoote.wordpress.com/2008/09/12/index-monitoring-and-foreign-keys-caution-kid-a/]
Try looking at the problem via ASH raw data - V$ACTIVE_SESSION_HISTORY and
DBA_HIST_ACTIVE_SESS_HISTORY.
It should help to add information to the blocked/blocker scenarios.
Question: Export / Import help
1. use flashback_scn to ensure the data is consistent - if you have an outage on that application it
may be OK anyway - but i would always want to include this
2. If you have Oracle EE use parallel - this is where you'll get the most performance improvement
from
3. don't use compression on export it will slow it down (and you need an extra licence anyway)
4 Find out if some of the internal metadata operations are slow (extracting DDL and doing the
estimates etc) - this could be bugs and there may be workarounds
5 If the problem is a certain table - find out why it is so slow - does it have lobs?
http://www.datadisk.co.uk/html_docs/oracle_dg/cheatsheet.htm
Differences Between Enterprise, Standard and Standard One Editions on Oracle 12.1
(Doc ID 1628809.1)
https://community.oracle.com/thread/3687755
https://apexapps.oracle.com/pls/apex/f?
p=44785:141:0::::P141_PAGE_ID,P141_SECTION_ID:6,84
The potential size of bigfile tablespaces is why people use BCT (block change tracking) to minimise
the backup load, and flashback logging to give them some chance of flashing back the database and
then rolling it forward rather than restoring a very large file which (ultimately) is the inevitable threat
with bigfile tablespaces. You need to consider how much you're prepared to pay, and how much
time/data you're prepared to lose, for each type of scenario you will need to recover from before you
can sensibly work out a best (for you) strategy. (And then you need to review the strategy and
underlying assumptions every year or so.)
No specific "demerits" that I can think of, just a couple more opportunities for making a couple more
mistakes.
a) If you're running on filesystem make sure you're using directio to bypass the file system otherwise a
small number of bigfiles could lead to excessive inode contention
b) If you're doing backups with RMAN then remember to specify the section size if you want to
maximise your use of concurrent channels on backup
Things you also expect to do if you think bigfiles are a good idea
a) Enable change tracking to get proper incremental backups from rman
b) Think carefully about the impact of the tablespace being the unit of restore in the event of a
catastrophe
Question: How can you restore latest backup using3 months Old Control file?
Of course you can. A RECOVER DATABASE USING BACKUP CONTROLFILE lets Oracle know
that the controlfile is "old" -- it doesn't matter if the "old"ness is older than the lowest checkpoint SCN
in the database backup.
https://community.oracle.com/thread/3689660
https://community.oracle.com/thread/3691213
Simple:
1. row directory pointer - the PHYSICAL address of a row within a block
2. rowid - the LOGICAL address of a row within a block
The row can be moved within a block. That WILL change the PHYSICAL location of the row but it will
often NOT CHANGE the logical number of the row in the block.
The PHYSICAL location of a row has NOTHING to do with the ROW NUMBER of that row in the
block. A row stored at physical location starting at byte 1023 could be row 1, 17, 84 or whatever in
that block depending on where the other rows are stored.
How to Move/Restore DB to New Host and File System using RMAN (Doc ID
1338193.1)
If you want to move database only, there are different way you can do it;
1) Golden Gate
2) Data Guard - Active Data Guard
3) Data Pump
4) RMAN Backup/Restore
5) Storage level Copy/Move
6) CTAS
If you want to move the whole exadata machine, you need to contact vendor who actually does this
job. In my opinion taking help from Oracle is better.
We recently had a data center move and moved some exadata machines to a new location.
1) Get the network ready on new Data Center and make sure all connectivity (e.g. you can remote
access to it) are working. In our case the new center has same IPs network.
2) Get the Vendor who does transporting of Hardware Machines.
3) On the last day before move, Shutdown the Exadata Machine ( shut down database, RAC,
shutdown db nodes, cell nodes, infinity band etc.. )
4) Power off Exadata Machine
5) Unplug all network cables.
6) Let the Hardware cool down.
7) Transport to new location and move the Machines to correct Bay and wait for machines to be
acclimatized
8) Plug Network cables.
9) Power On Exadata.
I found a Support note which also you can refer; What Needs to be Done When Moving
Cluster Nodes to a New Data Center (Doc ID 1332423.1)
https://community.oracle.com/thread/3691874
https://community.oracle.com/message/11241625#11241625
1. select name,spare6 from user$ where type#=1 and spare6 < sysdate - 30 ;
2. If you were on 12.x, you'd have another option (dba_users.last_login).
3. If auditing is enabled you can use select * from DBA_AUDIT_TRAIL where username =
'SCOTT'
DBA_AUDIT_TRAIL.TIMESTAMP – column specifies the logon time.
https://community.oracle.com/thread/3695030
The recommendation about per process Memlock does not appear when the database is started
manually.
https://community.oracle.com/thread/3704819
select regexp_replace(substr('ABCD-234-3245-A',instr('ABCD-234-3245-
A','-',1,1)),'[^[:digit:]]') position_underscore from dual;
or
select replace(substr('ABCD-234-3245-A',instr('ABCD-234-3245-A','-')) , '-', '')
position_underscore from dual ;
You can also look at this link which caters for 11gR1 as well though title say 10g,
How to Convert 10g Single-Instance database to 10g RAC using Manual Conversion
procedure (Doc ID 747457.1)
http://skillbuilders.com/webinars/webinar.cfm?id=94&w=convert-to-oracle-12c-rac-
in-10-minutes
HTH,
Pradeep
Rconfig Error When Converting From Single Instance To RAC (Doc ID 371519.1)
How to Convert 10g Single-Instance database to 10g RAC using Manual Conversion
procedure (Doc ID 747457.1)
How To Convert A Single Instance Database To RAC In A Cluster File System
Configuration (Doc ID 208375.1)
How to Convert a Single-Instance ASM to Cluster ASM (Doc ID 452758.1)
http://docs.oracle.com/cd/B19306_01/install.102/b14205/cvrt2rac.htm#sthref1165