Database not stable
Date: Tue, 31 Aug 1999 08:52:20 +0200
From: George Fungayi Mamvura mamvura@alpha.co.zw
To: oraapps-l@cpa.qc.ca
Subject: Database not stable
Hi
I am having problems with my apps database (PRD).
It keeps failing over.
Only limmited Users are given access to the database even if it is up.
Can anybody tell why and what I should do?
Urgent
Thank you
George Mamvura Alpha Systems
Oracle Customer Support Analyst Oracle Distributor for Zimbabwe,
Oracle Support Department Zambia, Mozambique & Malawi
email: mamvura@alpha.co.zw P.O. Box CY 1504 Causeway
mamvura_george@hotmail.com
tel: +263 4 730715/9 Harare
fax: +263 4 792204 Zimbabwe
Date: Tue, 31 Aug 1999 09:31:29 -0400
From: "Mark W. Farnham" mwf@rsiz.com
To: "'oraapps-l@cpa.qc.ca'" oraapps-l@cpa.qc.ca
Subject: RE: Database not stable
George:
If someone can determine your problem without a few more details, I will be
impressed.
Please send a little more information such as error messages from the alert
log or background job dumps.
These might be in $ORACLE_BASE/admin/PRD/bdump, if there is such a place on
your system.
Of particular interest would be any messages from the first crash and any
messages you get if you try to restart the database via svrmgrl.
It is sometimes helpful to know what operating system, hardware
manufacturer, Oracle RDBMS version, and Oracle Applications release you are
running (or not running in this case).
Good luck!
Mark
Date: Tue, 31 Aug 1999 17:36:31 -0400
From: "Innamuri, Chakrapani" cinnamuri@btg.com
To: "'oraapps-l@cpa.qc.ca'" oraapps-l@cpa.qc.ca
Subject: RE: Database not stable
I think you need to change the init.ora parameter for no. of processes.
Also there are other constraints like
- you may need to increase the memory of the machine etc.
-chakrapani
Date: Wed, 1 Sep 1999 08:29:41 -0500
From: "Hostetler, Don" Don.Hostetler@impacgroup.com
To: "'oraapps-l@cpa.qc.ca'" oraapps-l@cpa.qc.ca
Subject: RE: Database not stable
In the init.ora file put open_cursors = 255. Then bounce the RDBMS.
Don
Date: Thu, 2 Sep 1999 17:25:50 +0400
From: "Mehul Zaveri" mehulapps@hotmail.com
To: oraapps-l@cpa.qc.ca
Subject: Re: Database not stable
Just to add on :-
Also there is an UNIX-OS parameter to allow the no. of processes per session
and also per user. This also may need to change as oracle spawns a new
process for every connection.
Best regards,
Mehul Zaveri
Oracle DBA
EPPCO-DUBAI
Roles and priveleges of Db
Date: Sat, 24 Jul 1999 23:30:02 +0500
From: S V MANSATA kunal@blr.vsnl.net.in
To: oraapps-dba@cpa.qc.ca
Subject: database roles
hi all
i have a role called DBA,i grant that role (DBA) to another role
DBALESSONE, now the new role will have the same privs as the DBA role. eg
if DBA has 89 privs, then DBALESSONE will also have 89 privs
i now grant the role ( DBALESSONE) to a user called MARK, now i need to
revoke a privilege from mark, that is only one priv, eg ALTER DATABASE
,without having an inpact on the main DBA role, when i try to revoke from
the DBALESSONE role i get an error
ORA 1952 - system privilege not granted to 'MARK'
is there any workaround this or do i need to revoke and create a new role
and assign privs one by one
thanks in advance
promeet s mansata
Date: Sun, 25 Jul 1999 00:46:37 PDT
From: "shankar govindan" shankargovindan@hotmail.com
To: oraapps-dba@cpa.qc.ca
Subject: Re: database roles
First take a DBA book and read what a role is and then check what the ADMIN
option does. Role for security would tell you clearly.
Shankar
Date: Sun, 25 Jul 1999 22:49:45 +0500
Date: Sun, 25 Jul 1999 20:49:22 PDT Objects - Moving Db
Date: 25 Jul 99 07:06:49 EDT Concurrent Manager running a request - identify
Date: Mon, 26 Jul 1999 14:59:48 -0400
Date: Mon, 26 Jul 1999 21:23:16 PDT SGA settings
Date: Wed, 28 Jul 1999 09:23:00 -0700
Date: Wed, 28 Jul 1999 10:21:08 -0700
Date: Wed, 28 Jul 1999 13:33:39 -0500
Date: Wed, 28 Jul 1999 13:06:04 -0700
Date: Wed, 28 Jul 1999 19:15:00 -0400
Date: Thu, 29 Jul 1999 08:55:37 +0200
Date: Thu, 29 Jul 1999 11:50:35 -0700 Webserver Configuration
Date: Thu, 25 Mar 1999 07:42:37 PST Resetting time on database server
Date: Mon, 29 Mar 1999 18:00:38 +0930
Date: Tue, 30 Mar 1999 08:29:54 +1000 Database Links and MQSeries
Date: Thu, 2 Sep 1999 08:56:50 -0700 (PDT) Server Partitioned Certifications
Date: Mon, 6 Sep 1999 13:58:50 -0500
Date: Mon, 6 Sep 1999 16:24:17 -0400
Date: Mon, 6 Sep 1999 17:02:47 -0500 Migrating db to another server
Date: Mon, 4 Oct 1999 10:26:16 +0300 Tablespace for AP,GL,FA,CE environment
From: Marzieh Hariri [mailto:mhariri@russelmetals.com] Date: Mon, 18 Oct 1999 14:29:08 -0500
From: Bharat Patel bpatel@DOMINOAMJET.com
From: S V MANSATA kunal@blr.vsnl.net.in
To: oraapps-dba@cpa.qc.ca
Subject: Re: database roles
shankar
the admin option allows the user to grant the role to anyone, what does it
have to do with my query, the dba role was just an example
okay take for example a role named ADMIN_ROLE ( create role ADMIN_ROLE
)containing 80 privs (grant
note that ADMIN_GROUP_ROLE contains another two roles in addition to
ADMIN_ROLE, say ADMIN_HR_ROLE( having 10 privs) and ADMIN_FIN_ROLE ( having
10 privs)
now the main role would have 80 + 10 + 10 (100) okay
now for example i have assigned the role ( ADMIN_GROUP_ROLE )to 40
users,one of them is CHRIS, now i want to revoke only one privilege from
CHRIS,ie i need chris to have only 99 privs, i try to revoke using the
following
revoke select any table from chris, i get the following error
ORA 1952 - system privilege not granted to 'CHRIS'
this is because i asigned him a role containing roles, so if i revoke it
from the main role ie ADMIN_ROLE, it would be revoked from all 40 users,
this is not what i want, the 39 users need to have all 100 privs and CHRIS
should have only 99 , is there any way instead of revoking everything and
creating a new role and then assigning it to him again, is there any way i
can modify the role, so that it would only affect him
From: "shankar govindan" shankargovindan@hotmail.com
To: oraapps-dba@cpa.qc.ca
Subject: Re: database roles
Mansta,
That was only for you to pick the roles definition from the and no
harm intended, okay!.
Well the roles can be passed on with ADMIN option that's true. The
definition of role itself is for making mass allocation than for every
individual user. That's why some of the objects still need exclusively to be
granted.If you revoke for that user a particular privelege as system then
which takes precedence. Should check that out.
Shankar
From: Dinesh Chandra dchandra@usa.net
To: oraapps-dba@cpa.qc.ca
Subject: Moving Database Objects
Hi All,
How do I move database objects, Tables and Indexes, of one user say INV from
one Tablespace, name OthersT and OthersI, to another Tablespace INVT and INVI
in Oracle Applications ?
Any suggestion is welcome.
Cheers
Dinesh Chandra
From: Jeffrey Beckstrom JBECKSTROM@gcrta.org
To: OraApps-DBA@cpa.qc.ca, OraApps-L@cpa.qc.ca, oranca-l@doug.org,
Subject: which concurrent manager ran a request
I am reviewing the number of concurrent managers we have running and would
like to know if it is possible to determine which concurrent managers ran a
request, i.e. inventory, standard, etc.
Jeffrey Beckstrom
Database Administrator
Information Systems
Greater Cleveland Regional Transit Authority
1240 W. 6th Street
Cleveland, Ohio 44113
(216) 781-4204
From: "shankar govindan" shankargovindan@hotmail.com
To: oraapps-dba@cpa.qc.ca
Subject: Re: which concurrent manager ran a request
From the application the requests Manager Log should show you this. But I
had a funny problem once wherein I checked a particular requests manager log
and it showed INVLIBR-the following requests are not registered. But the
requests were completed successfully and if the requests were completed by
STD then the manager log for STD should show me this right.
Shankar
From: Joe.Madden@sentrol.com (Joe Madden)
To:
Subject: SGA settings
Folks,
I am curious about what type of SGA setting people are using.
I have the following:
shared_pool_size 314572800
db_block_buffers 32000
db_block_size 8192
sort_area_size 5242880
I have adjusted my db_block_buffers setting from 12000 to 32000
incrementally over the last few weeks and my buffer cache hit rate never
hits above %86
I use the following script (I know there can be questions about 'how' it is
measured)
select (1-(sum(decode(name, 'physical reads', value,0))
/(sum(decode(name, 'db block gets', value,0)) +
sum(decode(name, 'consistent gets', value,0)))))
* 100 "Hit Ratio"
from v$sysstat
Our system is as follows:
HP k9000
HP-UX 10.20
Oracle 7.3.3.5
Oracle APPS 10.7SC.
2.5 GB RAM
5.0 GB Swap
6 CPUs
200 concurrent users
Database ~ 15GB
I am curious on the groups settings - has anyone experimented with caching
tables - and if so which ones. Also has anyone experimented with rebuilding
tables altering the freelist value from 1 to something>1 (particularily OE).
Any ideas would be greatly appreciated.
Regards,
Joe
From: Karen Blackwell KBlackwell@rockshox.com
To: "'oraapps-dba@cpa.qc.ca'"
Subject: RE: SGA settings
I (usually) have a consistent hit rate in the high 90's. My environment is
as follows:
Sun E5000
Solaris 2.6
Oracle 7.3.3.6
Oracle APPS 10.7NCA
2 GB RAM
4 GB Swap
4 CPUs
100 concurrent users
Database ~ 8 GB
SharedPool size: 104857600
db_block_buffers 12800
db_block_size 8192
sort_area_size 25600
I don't know how much this helps you. We are fairly new to Oracle and I am
still very green. Most of these values were established before "go-live".
Obviously, there are times during the day when my hit rate drops, sometimes
significantly. But typically not for extended periods.
Good luck in your analysis.
Karen Blackwell
DBA/System Administrator
RockShox, Inc.
408.570-4912 phone
408.435.7468 fax
From: Bharat Patel bpatel@DOMINOAMJET.com
To: "'oraapps-dba@cpa.qc.ca'"
Subject: RE: SGA settings
here are my sga size and I am getting consistently hit ratio about 97%.
Total System Global Area 349960940 bytes
Fixed Size 47852 bytes
Variable Size 85712896 bytes
Database Buffers 263782400 bytes
Redo Buffers 417792 bytes
my db_block_buffers = 32200
db_block_size = 8192 (common for OLTP)
sort_area_size = 4096000
we have
2GB RAM,
8GB swap space,
3 200 MHz CPU's,
database size 18GB,
100-125 concurrent users.
There are other things need to be consider related with performance not only
SGA size.
thanks.
Bharat Patel
D.B.A.
bpatel@dominoamjet.com
847-244-2501 ex 1249
From: Joe.Madden@sentrol.com (Joe Madden)
To:
Subject: RE: SGA settings
I appreciate your input and I am also quite knowledgeable about there being
more to tuning than the SGA but that is the question - are there other
things that you know about that will get my data buffer hitting higher than
bumping up the number of buffers - BTW which cache is 97%, library, data,
data dictionary??
Thanks again
Joe
From: "Mark W. Farnham" mwf@rsiz.com
To: "'oraapps-dba@cpa.qc.ca'"
Cc: "'gsharpe@terascape.com'" gsharpe@terascape.com
Subject: RE: SGA settings
Joe:
If you have a high ratio of interface data and user input to reports, your
maximum theoretical hit rate will be
lower, since new data is a miss every time. This can certainly account for
a variance from 87% to 97%.
Is response time below standards or requirements, or are you just trying to
meet an abstract performance metric of hit rate? (Please note that this is
a genuine question even though it may read rhetorically).
If your throughput is below what you need and you've already tuned the easy
stuff (and your parameters seem
okay at first glance. You might consider trading off some shared pool for
more block buffers, but your mileage
may vary), then you might consider some of the "Extreme Measures" from a
paper I gave at OAUG a while ago.
It may be too old to be on line, but I think it is still on point. An
important consideration is whether the work is
worth it, but doing things such as rebuilding gl_code_combinations so that
it is in the physical order of most
likely access is an example. Greatly oversimplifying, the notion is this:
If when a the database reads a block it also brings in the next "n" rows
that are very likely to be the next thing you need, then you magically
skipped
n-1 physical reads. If, for example, the folks who consume your FSG reports
have a consistent order they like to see things, so you've written most of
the FSG reports so they use the same order, then this can be a huge
win.
Please carefully consider whether the effort is worth it, especially if you
get good results from rebuilding a small table such as
gl_code_combinations. You may be tempted to rebuild gl_balances and
gl_je_lines. This can
pay huge dividends, but those tables can be enormous and may take longer
than you expect to load in your
desired physical order. If you do engage in rebuilding very large tables,
be sure to do a dress rehearsal to
estimate the outage, and plan to do the rebuild in two or more stages so
that older data is loaded with a tiny
pctfree and more recent data gets a normal pctfree.
Another tack is to live with the lower hit rate and tune the disk farm so
that your throughput rises in spite of the hit rate being less than you
desire. The emerging science of placement tuning near and dear to my heart,
and
if it interests you drop me a line off list.
Mark W. Farnham
President
Rightsizing, Inc.
mwf@rsiz.com
From: stephanie mandell steph@a2000.com
To: oraapps-dba@cpa.qc.ca
Subject: table rebuilding (was RE: SGA settings)
Mark (or anyone else),
Have you got a recipe for rebuilding tables?
steph
--
Stephanie Mandell +31 20 7707 857
Oracle Applications DBA A2000, Amsterdam
From: Joe.Madden@sentrol.com (Joe Madden)
To:
Subject: RE: SGA settings
Mark,
Thanks for the feedback - your question about cache ratio is a good one - I
am aiming for the high 90's because that is where it is 'supposed' to be -
but given the fact that new data is entered in our system all the time, I
expect to have it be less than ideal - I was curious what the groups
consensus values were...
Your other ideas are worth considering but I can't help but think that my
next step is to deal with chaining - which there is certainly some - and
since I have to do that why not bump up the freelist setting - again I am
curious if anyone has gone down this route.
The disks are about as balanced as I can get them - if anyone else has any
further insights, I would greatly appreciate them.
Cheers
Joe
From: "Basanta Kumar Dash"
Subject: WebServer Configuration
Hi,
We have chanaged the hostname, IP address and the domain name of our web
server. I have changed the respective files under ows directory to
reflect the new hostname, ip address and the domain name. But, when I
issue the command 'owsctl start www' and 'owxctl start admin' I get the
error message as 'Web listerner does not find 'www' and 'admin'
respectively.
Am I missing any set up?
we are on rel 11 and WinNT.
Thanks and regards,
basanta.
Date: Thu, 25 Mar 1999 09:48:44 -0600
From: "Logan, Ernie"
Subject: RE: WebServer Configuration
The directory where the svadmin.cfg and svwww.cfg files reside uses the host
name. I believe you will need to rename/copy the directory.
From: Hilary Soloff Hilary.Soloff@santos.com.au
Subject: GEN : resetting time on database server
Hi everyone,
We are running 10.7 NCA on Solaris 2.5.1.
Daylight saving ended on the weekend, but the time on our database server
(and middle tier) did not get reset.
When our unix people noticed it about midday, they set the time back by 1
hour to the correct time - while Oracle Apps was up and running. (We've
asked them to talk to us before they do this in the future!)
This caused a couple of problems - the Conflict Resolution Manager shut
down and we had a some julian date errors (APP-1270 error generating julian
system date ) during the 'repeated' hour.
Does anyone know of any other hidden problems this may have caused?
Thanks,
Hilary
From: "Bevin Watson"bevin.watson@abs.gov.au
Cc: Hilary.Soloff@santos.com.au
Subject: GEN : resetting time on database server
Our guys did a similar thing to us 2 years ago while we were on 10.6. They
brought the server up (and hence the databases and apps) and then changed
the time. We just shut down the apps and restarted and everything was OK.
We also did this sort of thing a fair bit in the Y2K test lab and saw no
deleterious impact.
The only thing I can think of which could cause problems is if you do any
processing of your security or audit logs. You may get some whacko
results. I'd get a fresh backup as well so you don't have to rely on redo
logs being rolled back.
Did you do any "setups" in the morning? For instance, AR requires you to
have the payment method with a start date before the payment source (or
something like that) - there is a similar trick with OE price lists.
Bevin Watson
Australian Bureau of Statistics
From: Binoy Raman binoypr@yahoo.com
To: oraapps-l@cpa.qc.ca
Subject: GEN:Database Links and MQSeries
Hi,
We have an Oracle application (CSS) running on AIX, this application
communicates with the mainframe database (IDMS ?) using the IBM
MQSeries and the Oracle Procedural Gateway for MQSeries. Oracle
Financials runs on a Sun machine, and we don't have the MQSeries
version for Sun.
We have created a database link beween our test instance of Oracle
Financials and CSS database, and tested calling the MQSeries procedures
from the financials instance. Our intention is to enable real-time
updates from Oracle Financials to the Mfg applications on the
Mainframe. Updates will be carried out in both directions.
Has anyone done something similar ?
Are there any risks in using database links for this ? Volumes ?
Binoy.
From: "nca" nca@ceteco.com
To: oraapps-l@cpa.qc.ca
Subject: GEN: SERVER Partitioned Certifications
Hello,
Can somebody explain me what is +ACI-Server Partitioned Certifications+ACI-.
I saw
this in Oracle Metalink when looking at the Server Certification for 10.7
NCA.
thanks in advance,
Dina Dubon Marchelli
Ceteco Information Technology
From: "Tom J Papaleo" tpapaleo@hearst.com
To: oraapps-l@cpa.qc.ca
Subject: Re: GEN: SERVER Partitioned Certifications
Server Partitioning for 10.7 enables the Oracle Applications to execute on
an
Oracle 8 database. The PC Clients connect directly to the Oracle 8 Database
while the Unix Concurrent Managers connect to the Oracle 8 database via the
an
Oracle 7 set of executables. In our case, we are Server Partitioned with
7.3.4
and 8.0.5. There are a few PL/SQL differences between an Oracle 7 database
and a
Server Partitioned Configuration
From: "nca" nca@ceteco.com
To: oraapps-l@cpa.qc.ca
Subject: GEN: Information about Server Partitioned Configuration
Hello,
Where can I find some information regarding to Server Partitioned
Configuration (7.3.4 and 8.0.x). We are moving from 10.7 character to 10.7
NCA.
Thanks in advance,
Dina Dubon
From: Mohammad Fouwaaz Mfouwaaz@savola.com
To: Multiple recipients of list oraapps-l@cpa.qc.ca
Subject: Migrating to a different server
Hello friends!
We are a group of companies running on 10.7. Our main server is an HP V
Class and the enterprise-wide Apps resides in it (HP-Unix 11.0). In
addition we have a totally separate server (HP K210 also HP-Unix 11.0) which
is servicing a isolated Apps database instance for one of our off-shore
companies.
Now, we are planning to move this separate database instance to the main V
Class machine (continuing to maintain it as a separate instance).
Do we simply copy the executables (files) or should they be re-installed?
How long is this going to take?
What are the precautions, if any, that we should take?
Any help/advice from those who have been through this before would be
greatly appreciated. Thanks in advance
With Best Regards
Sent: Monday, October 18, 1999 11:50 AM
To: Multiple recipients of list
Subject: GEN:What tablespaces are required in AP,GL,FA,CE environment?
Hello everyone:
When I select from my dba_tablespaces I get 69 tablespace names ,some of
them are needed due to table integrities (Like PO is needed since we are
using AP) but I am sure there are some which we don't need and they are
just taking space .
Is there any documentation, sites where I can find out about all table
integrities?
We are on 10.7 SC 16.1 and we have AP,GL,FA,CE.
Insights , feedback and opinions are most welcome.
Thanx.
Marzieh Hariri
Senior Oracle Support Analyst
Russel Metals Inc.
Phone:(905) 816-5152
To: "'oraapps-l@cpa.qc.ca'" oraapps-l@cpa.qc.ca
Subject: RE: What tablespaces are required in AP,GL,FA,CE environment?
Harari, in my case, I will prefer not to drop the tablespace but shrink it
as much as possible. Dropping is very dangerous as Apps modules are inter
related.
I already shrink some of to about 10-20MB.
hope this helps.
Bharat Patel
D.B.A.