Oracle


Go to MOS directly from Google results – part 1

I’m sure you know this annoying sequence, when you click on a My Oracle Support result from Google:

It would be nice to go from the Google results page directly into the MOS website.

This can be easily done with a Google Chrome (I believe also exists for Firefox and Edge) that rewrites the URL.

I used Redirector which I configured with:

To make copy/paste easier:

Redirect: ^(?:https?://)support.oracle.com/knowledge/(.*)/(.*)_([0-9]).html
To: https://support.oracle.com/epmos/faces/DocumentDisplay?id=$2.$3

Now from Google results I go directly the Oracle Support note page. 🙂


Explore multiple optimizer features and fixes with SQLT Xplore

Yesterday I described how I come to SQLT Xplore and it helped me to find out, on Oracle 12.2.0.1, that optimizer_features_enable=8.1.3 decreased the parsing time of a query from 5 seconds to 0.2 seconds. Today I show how to use SQLT Xplore.

What is SQLT Xplore?

SQL Xplore automatises the test of almost 2’000 optimizer parameters and bug fixes control against one query, allowing to discover which parameter was eventually the reason of a performance regression.

(more…)

Long parsing on Oracle 12.2 and the discovery of SQLT Xplore

Today I discovered a fantastic free tool from Oracle: SQLT Xplore!

At a client I had one query that took long time parsing. The tkprof result of the 10046 trace showed it took 5 seconds to parse:

call     count       cpu    elapsed       disk      query    current        rows
------- ------  -------- ---------- ---------- ---------- ----------  ----------
Parse        1      5.17       5.30          0          0          0           0
Execute      1      0.00       0.00          0          0          0           0
Fetch        1      0.00       0.00          0        665          0          26
------- ------  -------- ---------- ---------- ---------- ----------  ----------
total        3      5.17       5.30          0        665          0          26

Looking around I’ve noticed that disabling the cost based transformation:

ALTER SESSIONS SET "_optimizer_cost_based_transformation"=off;  

…the query was much faster to parse: only 0,5 seconds.

(more…)

Why PGA_AGGREGATE_LIMIT appears to be set, when it is not?

Today I had to explain why the pga_aggregate_target was showing a value, but one does not have to care about it most of the cases. Here the same explanation.

When doing:

SQL> show parameter pga

NAME                                 TYPE                                         VALUE
------------------------------------ -------------------------------------------- ----------------------
pga_aggregate_limit                  big integer                                  3000M
pga_aggregate_target                 big integer                                  1G

One would think that pga_aggregate_target is set to 3000M. However when one checks on the spfile, there is nothing defined:

(more…)

Strange behaviours on setting PDB parameters 3

[Tests performed on Oracle 19.8]

All started when I wanted to create a query to check which parameters are set on a PDB and the difference from the CDB$ROOT container.

set pages 110
col pdb_name for a10 
col name for a30 
col value for a20
col pdb_value for a20
col root_value for a20
col source for a10
select a.pdb_name, a.name, a.value PDB_VALUE, b.value ROOT_VALUE,source from
  (select pdb_name,name,value,a.con_id, decode(ismodified,'MODIFIED','PDB SPFILE','PDB$SEED') SOURCE 
   from v$system_parameter a left join dba_pdbs b on (a.CON_ID=b.pdb_id)
   where a.con_id>2 and (ismodified='MODIFIED' or isdefault='FALSE')) a,
  (select 'CDB$ROOT' pdb_name,name,value,con_id,null
   from v$system_parameter where con_id=0) b
where a.name=b.name and a.con_id>2
order by 1,2;

But I know there is also one view called pdb_spfile$ that would show the parameters on the PDB pseudo-spfiles:

col pdb_name for a10 
col name for a20
col value$ for a20
select pdb_name,name,value$ 
from pdb_spfile$ left join dba_pdbs on (CON_UID=pdb_uid) 
where con_id>2 
and bitand(nvl(spare2,0),1)=0
order by name;

The V$SYSTEM_PARAMETER is well documented, while the PDB_SPFILE$ is not.

A slight more complete query to check parameters which are not set the same between cdb$root and the PDB is this one:

with pdb_params as (select pdb_name,name,trim(both '''' from value$) value$ 
                                   from pdb_spfile$ left join dba_pdbs on (CON_UID=pdb_uid)
                                   where con_id>1 and bitand(nvl(spare2,0),1)=0 ),
        cdb_params as (select name, value,inst_id,con_id from gv$parameter ),
        spfile_params as (select name,value, sid from v$spparameter where isspecified='TRUE')
  select pdb_name, pdb_params.name, value$ pdb_value, cdb_params.value root_value, spfile_params.value spfile_value
    from pdb_params, cdb_params, spfile_params
    where pdb_params.name(+)=cdb_params.name and cdb_params.name(+)=spfile_params.name
        and (value$!=cdb_params.value or value$!=spfile_params.value)
union all 
  select case when sys_context('USERENV','CON_NAME') !='CDB$ROOT' then 'CALL THIS SCRIPT FROM CDB$ROOT' end pdb_name,
        null,null,null,null from dual
order by 1,2;

Now, the set or unset parameters do not work the same way, as I expected, it trigger some strange behaviours.

(more…)

HugePages not used when starting DB with srvctl (but works with sqlplus)

Once again I end up with my clients database swapping. Why? After quick investigation, could see that HugePages were not used on the last restart of the database.

oracle@myvm1:./trace/ [oracle19] grep -B1 -A4 PAGESIZE alert*.log
2020-04-14T04:36:34.601494+02:00
  PAGESIZE  AVAILABLE_PAGES  EXPECTED_PAGES  ALLOCATED_PAGES  ERROR(s)
2020-04-14T04:36:34.601550+02:00
        4K       Configured              10              10        NONE
2020-04-14T04:36:34.601642+02:00
     2048K           247816            8193            8193        NONE
--
2020-10-13T22:59:28.856763+02:00
  PAGESIZE  AVAILABLE_PAGES  EXPECTED_PAGES  ALLOCATED_PAGES  ERROR(s)
2020-10-13T22:59:28.856818+02:00
        4K       Configured              10         4186122        NONE
2020-10-13T22:59:28.856925+02:00
     2048K           202479            8193              17        NONE

Why was that? I did use a normal start command:

oracle@myvm1:./trace/ [oracle19] srvctl start database -db mydb

Let’s put the context. This is a Oracle Restart server, with separation between oracle and grid users.

(more…)

Get unlock from PDB Lockdown profiles

PDB Lockdown profiles allow, on a multitenant database, to limit what a user can do inside a PDB.

One use case is when we want to avoid high privileged users (like Application DBAs) to perform ALTER SYSTEM or ALTER SESSION commands.

Quickly we would think we can do a PDB Lockdown profile like:

SQL> create lockdown profile lock_test;
SQL> alter lockdown profile lock_test disable statement=('ALTER SESSION');
SQL> alter lockdown profile lock_test disable statement=('ALTER SYSTEM');

The problem of this simple profile is that we can lock ourselves, also as common user, inside the lock profile.

Image that you want to enable this profile on several PDBs:

SQL> alter session set container=pdb01;
Session altered.

SQL> alter system set pdb_lockdown=lock_test;
System altered.

SQL> alter session set container=samplepdb;
ERROR:
ORA-01031: insufficient privileges

Oups, you cannot anymore change the active container!

(more…)

Which OPatch to use? One OPatch version fits (almost) all.

The download page of Oracle OPatch has quite some room for improvement: put some ‘order by’ on the version and platform would be welcome. Also, make clear that there are very few versions of it.

In fact, for database, there are just two versions of OPatch! One OPatch version that covers all database supported versions from 12.1 to 20c. For paid long-term supported Oracle 11.2 there is another version.

So, in summary, here the OPatch version you need to patch your DBs:

Oracle Database versionOPatch version
11.211.2.x
12.112..2.x
12.212..2.x
18c12..2.x
19c12..2.x
20c12..2.x

The download link is this one:

https://support.oracle.com/epmos/faces/ui/patch/PatchDetail.jspx?patchId=6880880

For Enterprise Manager (middleware) there is another OPatch version, 13.9.x which I don’t have experience with.

The information about which OPatch versions is needed to apply the Database RU, RUR, is now part of the Patch Availability Document. For instance for OCtober 2020, this is what we can see:

Note 1: For Enterprise Manager (middleware) there is another OPatch version, 13.9.x which I don’t have experience with.

Note 2 – for Oracle guys out there: when we see the current size of the Release Updates, maybe it would be worth to include the latest version of OPatch within it. It would not increase so much the size and avoid the need of checking if we have the latest OPatch.


SQL query for queue table with priorities 6

Queue tables are used to take care of events. There is a mechanism that insert rows and another that takes care of the existing rows, usually deleting them at the end.

When possible one should use Oracle Advanced Queuing mechanism which takes care of managing the queue and a simple request will give you the next in the line.

Some applications develop their own queuing systems and it is a good way to learn how queue works.

Imagine a table with a list of elements and two specific columns:

  • order of arrival
  • priority

The first to be served is the one that, having the highest priority, was the first to arrive.

(more…)

Impact of Azure-Oracle OCI interconnect on a multi-cloud solution 3

Introduction

Oracle and Microsoft announced in June 2019 a cloud interoperability partnership which enables workloads across Microsoft Azure and Oracle Cloud. By creating a first joint multi-cloud solution, the software giants can each continue to provide the best of their services. At the same time customers do not need to decide which vendor they opt-out when moving their on-premises constructs.

Being myself an Oracle Database Administrator, this article aims to check what is the impact of distributing resources in multiple clouds, with databases remaining on Oracle Cloud Infrastructure.

It is not aim of the article to discuss the costs of resources on any of the clouds.

(more…)