Usn’s IT Blog

Before everyone diappears from The Net(tm) for Christmas:
Have a nice yule-tide, don’t eat too much of the feast, don’t drink and drive, and talk the RDBMS of your choice into operating itself for at least two or three days: You won’t miss the usual 6-in-the-morning phone calls on Christmas!

If you are lucky enough to have real holidays over the next days: A happy new year and let’s work on big red letters in January again! ;)

Yours
Martin Klier

PS: I hope you recognized: The usual omnipresent word starting with an X – it’s missing here. And, it hasn’t been too difficult. :)

Hi,

it’s old news, but oracle 11g expires passwords after 180 days, and locks you out if you mistype your password 10 times, and most DBAs don’t like that. It’s unsecure, but for client-server-applications a locked account is no single-user-annoyance, it’s a downtime killing SLAs, nerves and – hopefully not – DBA jobs. So just make your DEFAULT user profile less secure:

select * from dba_profiles where resource_type='PASSWORD' \
  order by resource_name;
alter profile DEFAULT limit FAILED_LOGIN_ATTEMPTS UNLIMITED;
alter profile DEFAULT limit PASSWORD_LIFE_TIME UNLIMITED;
select * from dba_profiles where resource_type='PASSWORD' \
  order by resource_name;

Be careful
Usn

Thursday 19th was the last day of DOAG Confernce 2009. Compared with the second day, the program was much more dense: For my (DBA) profile there have been several hours, where more that one talk would have been appropriate.

This list is, what I collected during the day.

Read more…

As promised, today my impressions of yesterday (confusing, I know). Clarification: This is about Wednesday, November 18th. I’m still at DOAG Conference 2009 in Nuremberg. Wednesday was a silent day for my interests, especially at noon there have been less talks being of interest, but maybe they have done something for my general knowledge…

So once again, this are impressions, a digital copy of my scratchpad, unordered and without warranty. :)

For SQL Trace analysis:

• let’s look at HOTSOS Profiler by Mr. Milsap, TRCANLZR (Oracle), TV$XTAT, ORASRP and MERTITS Profiler. For the latter, a limited free edition is available (without realtime V$ associations)
• Review the Index usage when date/timestamp datatypes are used
• 11g cursor close type (in a trace file) might enlighten something

Oracle on VM:

• Oracle VM 2.2 uses XEN 3.4 core now
• nice feature of Oracl VM: IO Priority (time slicing)
• Oracle DB in an Oracle VM will have about 10% losses on CPU power
• Oracle VM is limited to 60.000 IOPS :)
• PVM with Direct Driver option is fastest
• Activate Direct IO
• Windows will be faster than on bare metal when running on Oracle VM, due to better IO scheduling

Read more…

These days, the DOAG Conference 2009 happens in Nuremberg, Germany. First day was yesterday, Tuesday November 11th 2009. This is a machine-readable copy of my scratchpad, coming from the different talks I attended.

• Sun Oracle Database Machine Exadata 2: The “Quarter Rack” (consisting of 2 Servers and 3 storage nodes) sounds interesting.
• 11gR2 has some features only available with Exadata: For example, a special “Storage Index”, grouping several lines with theit MIN and MAX values in order to minimize scan ranges (called Smart Scan)
• Exadata provides inter-instance IO ressource management possibilities, it might be useful to look at the IO ressource management capabilities of 11g without Exadata as well!
• Really interesting stuff was “Righttime Data Warehousing – reducing latency” of Sven Bosinger; his presentation and the paper will be great material to explain some stuff for conceptional discussions. Thanks, Sven!
• Look at issues for your DB release version! Metalink Note 161818.1 is great for that!!
• Flashback Data Archive may be a good tool for an archiving concept (Oracle sales tells us about Basel II and stuff), but have care!Firstly, DDL on tables prepared for FDA is slow, and secondly and worse, if you have tablespace quotas on your schema, the archiving might stop if it’s full. And that without error messages due to asynchronous archiving. Only way to monitor that is to scan for new trace files all the time. :(
• The new Oracle Grid Infrastructure and its Instance Caging sounds interesting as well. Especially the CPU assignment with CPU_COUNT parameter sound nice, as well as CPU overcommitment (assigning more CPUs than you have).
• Oracle OMotion and RAC One Node are other nice ones: But at the moment, RAC One Node is not available with Standard Edition!

At a glance: The first day was great, and absolutely interesting. Let’s see what Wednesday brings, the program looks promising so far.

Best regards from Nuremberg
Martin Klier

In the last days, My Oracle Support was redesigned. But since flash is a performance eater and not always available, people like to have an option to use My Oracle Support (formerly Metalink) with HTML only.

The link to it is well-hidden, but nevertheless useful!

Give it a try: https://supporthtml.oracle.com

Best regards
Usn

drop table plan_table;
@?/rdbms/admin/utlxplan.sql

Did you ever change lock_sga = true for an Oracle Instance on AIX 5.3, and did see

ORA-27126: unable to lock shared memory segment in core
IBM AIX RISC System/6000 Error: 1: Not owner

in Return?

Do the following:

# vmo -o v_pinshm
v_pinshm = 0
# vmo -o v_pinshm=1
Setting v_pinshm to 1
# lsuser -a capabilities oracle
oracle
# chuser capabilities=CAP_BYPASS_RAC_VMM,CAP_PROPAGATE oracle
# lsuser -a capabilities oracle
oracle capabilities=CAP_BYPASS_RAC_VMM,CAP_PROPAGATE

It should do now.

Regards
Usn

Hi folks,

since Oracle 11g Release 2 is out now, I had to test one of the most-missed ASM features: the ASM cluster file system ACFS.

My Setup:

• Two VMware nodes with 2 CPUs and 1,5GB of RAM each
• Oracle Enterprise Linux 5.3 x86_64
• Four virtual cluster disks from the ESX server, 10GB in size each
• Building disk group DATA from them, with redundancy NORMAL
• containing four failgroups with each one of the disks within
• In DATA, one ACFS volume of 1 GB in size, mounted to /acfs1

Read more…