Matching Scenarios

Hi guys,
I like to know the features and improvement holes in SAP MDM regarding data quality management. For example, If we have Customer name like George and other record is also the same Customer but written differently like Gorge. I think here SAP MDM fails to find duplication records by their Soundex.
Second improvement hole is like if I have a repository where I have maintained 100 of users and user roles. Is it not like awful I have to repeat them again in one other repository.
Questions.
1. Any recommended reading where I can find such improvement holes in SAP MDM 5.5
2. User management problem for several repositories.
3. Matching with Soundex feature like we find normally in RDBMS "Oracle"
Regards,
Naeem

Hi Naeem,
The Problem of Soundex can be solved within MDM by using the Transformation feature.
- In the MDM data manger under the matching mode,you have a tab for Transformation.
- Here you can create Transformations which will identify George and Gorge as identical duplicates.
- You need to transform Geoge To George,This will make the transformation only virtual and not in reality,but however MDM will recognize them in the Data manager and select them as possible dupliacates.
- By given a high score.
- The only disadvantage in using this feature is that if you have many records with this kind of transformations,you will have to manually do it for each,before running the Strategy.
Kindly refer the below Link,to understand how this works.
MDM Matching Strategies for Master Data Consolidation
Regarding reusing the created Roles and Users in one rep into another.
- There are 3 different ways to transport your Roles and Users from one rep to another.
- Archieve/Unarchieve, Duplicate Rep, and Export/Import Schema.
- The Export/Import schema option is less preferred as the roles and the users will get transported however their pwd if set any may get resetted.
- However with the Archieve and Duplicate option you will get all the Roles and users replicated.
- The drawback being ,there is no option in MDM to transport just the Role and Users,you entire Repository structure will get transported.
- So if you want to reuse the same roles on to a differnt rep,then manul work will involve.
Regarding Data Quality, MDM's latest release 7.1 has additional facilities to handle data quality aspects in MDM.
Have a look at the Presentation which explains all the new upcoming features to be included in MDM 7.1:
https://www.sdn.sap.com/irj/scn/go/portal/prtroot/docs/library/uuid/101fdef1-7776-2b10-0ab9-8d67c1f3e1d7
Hope It Helped
Thanks & Regards
Simona Pinto

Similar Messages

GR/IR & Material GL accounts for 3 way match

Hi All Experts,
I have requirement where I have to configure the 3 way match scenario. For that I have to create material ( Stock) account, GR/IR account.
I have some doubts regarding this if I get experts guidance it will be great help for me.
Material Account
1 ) Material as a P & L OR Balance Sheet account ?
2 ) Reconciliation with material ? ( On control data tab in FS00 ! )
3 ) Line item ticked ? and Open Item ticked ?
4 ) Whether to define as cost element or not ?
GR/IR Account
1 ) GR/ IR account as a P & L OR Balance Sheet account ?
2 ) Reconciliation with material ? ( On control data tab in FS00 ! )
3 ) Line item ticked ? and Open Item ticked ?
4 ) Whether to define as cost element or not ?
and anything configuration point which I have missed can have guidance over that will be helpful.
Regards,
Sharvari Joshi.

Hi,
Material Account - Hope this is a stock A/c
1 ) Material as a P & L OR Balance Sheet account - B/S A/c
2 ) Reconciliation with material ? ( On control data tab in FS00 ! ) - No reconciliation type
3 ) Line item ticked ? and Open Item ticked ? - Only line item display
4 ) Whether to define as cost element or not ? - Not a cost element
GR/IR Account
1 ) GR/ IR account as a P & L OR Balance Sheet account ? - B/S
2 ) Reconciliation with material ? ( On control data tab in FS00 ! ) - No reconciliation type
3 ) Line item ticked ? and Open Item ticked ? - activate both
4 ) Whether to define as cost element or not ? - Not a cost element
You might need to check the config in OMR6.
Regards
SM

Implement 2-way matching for SRM PO

Hi Gurus,
We are in SRM5.0 Extended Classic and currently running 3-way match scenarios. Client wants to implement 2-way match for certain Vendors. My question is how to accomplish this. Below are my thoughts. Please let me know if I missed any.
for 2 way match:
1. In ECC Maintain Vendor MAster - remove GR/IR indicator
2. In ECC Maintain Info records - remove GR/IR indicator
In SRM
1. use create Limit shop which defaults to 2 way (also gives option to select 3 way)
Please let me know if there are any config changes or new objects to be developed for 2-way match.
Thank you all
Rao

Hi Friends,
any suggestions is appreciated.
Thanks
Rao

GR/IR 3 way Match

Hi ppl,
I have a question about the price difference in 3 way match,
scenario:
I have my PO: 20 Kg - $40      @ $2/ Kg
               GR: 20 Kg
I have received invoice for 30Kg - $60
Then I have posted the invoice, it gets blocked ...!!! When I check in MRBR, it is showing that the invoice is blocked for quantity difference 10kg
Now, I will go back and change the PO to 30Kg and I have received another 10Kg   i.e., GR for 10Kg
so If i check in ME23N under status button, I have my PO for 30Kg, goods delivered for 30Kg and qty invoiced is 30Kg.
As, everything is fine, this invoice should be unblocked automatically using MRBR. But when I check in the MRBR, the invoice is still showing, it is blocked as the quantity difference is 10Kg.....
How to unblock the invoice automatically????
Please help me!
Thanks!

hi,
Three way match means Sap system internally matches Po and Gr values while posting IR..
Now when you post IR before GR, system will not be able to find the GR values for comparing and will block the IR for qty reasons..
But this doesn't make sense because, you can't release the IR untill you post the GR...so better you go for GR first and then go for IR and complete the process on time....
If you post the IR before GR, then the cycle completes only after you post the GR and so GR is must...
Regards
Priyanka.P

Can iTunes share purchased content across Apple IDs within one household, OR

QUESTION ONE
Will iTunes share purchased content using TWOApple IDs, across different devices within one household?
QUESTION TWO
Should each device with a separate user have its own iCloud ID? CONCURRENTLY, should each device belonging to the SAME user use the SAME cloud ID?
QUESTION THREE
I think I "can" use one Apple ID, and just usedifferent iCloud IDs for EACH device. Am I fundamentally wrong? Am I legallywrong?
UNDERSTOOD FACTS:
1) Apple states, each person should have their own AppleID: http://support.apple.com/kb/HE37
2) iCloud servicesare: mail, contacts,calendar etc...
3) iTunes servicesare: music, videos, Apps, etc...
You can sync iTunes across up to5 different computers using your Apple ID: http://support.apple.com/kb/ht1420, however there does not seem to be alimit of how many Apple devices youcan sync within using a single iTunes account (Apple ID). *NOTE: beware automaticsynchronization of new apps..
4) iTunesin the Cloud is a different animal altogether! This should allow sharing of all iTunes content across all “your” devices, aslong as it was purchased thru iTunes (If not, you need to shell out an additional $25/yr for iTunes Match)
SCENARIO: I got a new iPhone, and am giving theold iPhone to my spouse. ALL iTunes content has been managed and purchased using oneApple ID that matches one iCloud ID. Ideally, just like anyother family using multiple Apple products and usingiCloud; we would like to SHARE songs & Apps, but NOT share calendars, mail etc...(sharing contacts would be nice).
QUESTION FOUR
Under the above scenario, must I have a “.me email” account?
I have been reading threadsabout managing content across multiple devices, using different IDs, both AppleIDs and Cloud IDs, and there is a LOT of confusion. I'm visual, so I've attached the graphic illustrating how I "think" I should proceed. **I have not set up the new iPhone yet.
We do not have a Mac or a".me email" account, and we use Gmail.
Any help is appreciated!

I have the computer authorized to use the Apple ID content (been using iTunes since 2007).
My issue is I cannot create sperate iCloud IDs for each device. I get prompted/directed to create a new Apple ID, which is what I want to avoid. I want to use one "FAMILY" AppleID as it is, and use iCloud IDs for each device.
My logic is: ALL my purchases are tied to an AppleID, and I just want to share on devices that leave the house, like another iPhone.
So- what to do? How to get a seperate iCloud ID without a seperate AppleID? I have an iPhone & iPad, husband has an iTouch & iPhone. How to share selectively? All devices were originally setup using one Cloud ID and one Apple ID. Now I need new IDs to represent the new user on the iTouch & iPhone.

XMLfile in archieve folder but no record in MDM-DM

Hi All,
I work in MDM 5.5. automatic
XML File is getting processed in MDIS (i could see in Archeive folder)and with no Exception, but iam not able to see the record in Data manager .
In Log folder-log file i see as below (Iam pasting the last few lines) with a line 'Thread ordered to stop'. I am not sure why this happens and whats the issue... confusing...
<Trace ts="2011/09/19 11:11:23.946 GMT" tid="2057" entry-no="123708">Save Value Key Mapping.</Trace>
          <Timer ts="2011/09/19 11:11:23.952 GMT" tid="2057" entry-no="123709" name="Import Chunk" total="49.520218"/>
          <Trace ts="2011/09/19 11:11:25.828 GMT" tid="2057" entry-no="123713">[MDS=tsdux13 Repos=SRM_MDM_CATALOG ClientSystem=SRMMDM Port=TP1_Contract]: xImporter: Thread ordered to stop: RC = 0xffaa6000</Trace>
          <Trace ts="2011/09/19 11:11:25.830 GMT" tid="2057" entry-no="123715">[MDS=tsdux13 Repos=SRM_MDM_CATALOG ClientSystem=SRMMDM Port=TP1_Contract]: xImporter: Thread finished; Start -> End: 3023.000000000 seconds.</Trace>
          <Trace ts="2011/09/19 11:11:26.387 GMT" tid="2571" entry-no="123718">[SRM_MDM_CATALOG]Import Task Finished. Chunk size[50000], No. parallel chunks[5]</Trace>
     </Open>
</MDM_Log>
Any one can throw u suggestions please,
how to correct the issue.
Thanks,
Pradeep.

Hi,
Go to that respective inbound port where the archive folder is present and check the Import log from the Log folder for that port. Open the log and find the section where it talks about the number of records created or updated or skipped and paste the log.
Also in Import map make sure that you have the proper action specified in case of match/no match scenarios. You have to specify Create/Update/Skip actions. Be default it is mostly Skip.
Also check that you are logging into the same MDM repository for which you are checking the input/archive files.
Thanks,
Shambhu.

Bind Variable : required - no message, not required - error

hi
Please consider the example application created using JDeveloper 11.1.1.6.0
at http://www.consideringred.com/files/oracle/2012/RequiredOrNotBVarApp-v0.01.zip
It has these View Objects defined
- EmployeesReqNoValueBVarVO which has a required Bind Variable, and where clause "last_name like '%' || :LastNameBVar || '%'"
- EmployeesNotReqNoValueBVarVO which has a NOT required Bind Variable, and where clause "(:LastNameBVar is null or last_name like '%' || :LastNameBVar || '%')"
The Bind Variable in both View Objects does NOT have a value (no default value, not programmatically set, not entered by the user, ...).
This "no value set" aspect could be the result of a "development error", but the resulting observed behaviour is at least "peculiar".
- scenario (sc1) : run "try tryEmployeesReqNoValueBVarVO", see a table filled with rows and NO message about the required Bind Variable
- scenario (sc2) : run "try tryEmployeesNotReqNoValueBVarVO", see "No data to display." in the table and the "Missing IN or OUT parameter at index:: 1" error for the NOT required Bind Variable for a where clause that can deal with null values
- question (q1) How can be explained that scenario (sc1) does not result in an error message about the required Bind Variable (and scenario (sc2) does result in an error message about the NOT required Bind Variable)?
- question (q2) What does the "Required" checkbox for a Bind Variable really mean for Bind Variables used in a where clause of a View Object?
If the described behaviour can be reproduced (using RequiredOrNotBVarApp-v0.01.zip), confirmations are welcome.
(Aspects of these scenario's could be related to forum thread "Bind variable required/not required: strange behaviour".)
many thanks
Jan Vervecken

Thanks for your reply Frank.
Frank Nimphius wrote:
... your query actually is dependent on the existence of the bind variable because the bind variable is used in the where clause part. ...Sure, and the Bind Variable does exist (in both (sc1) and (sc2)).
... This has nothing to do with whether the bind variable value is checked for NULL and if the value it contains is NULL performs an "all you can eat" type of query. Because the existing, NOT required Bind Variable has not been given a value in scenario (sc2), the check for NULL only makes more explicit that NULL is a valid value for the SQL statement.
... Non required bind variables are used in the context of view criteria to avoid missing IN or OUT parameters. ...The "Required" checkbox for a Bind Variable is also available/enabled when the Bind Variable is not used in a View Criteria (in both (sc1) and (sc2)).
... In the case of a view criteria, the VO query may be executed without the view criteria applied and for this reason should not fail only because the defined bind variable is not provided. ...No View Criteria in the scenario's (sc1) and (sc2) I descirbe.
... If you run the query in the tester, you get
(java.sql.SQLException) Missing IN or OUT parameter at index:: 1
and no query is executed at all for EmployeesNotReqNoValueBVarVOSure, that matches scenario (sc2), but the Bind Variable can be configured as NOT required, how can it be missing?
The JDeveloper help page "Create or Edit Bind Variables Dialog - Variable Page" says:
"... Alternatively, you can use the SQL Statement page of the Edit View Objects dialog to enter a parameterized WHERE clause. Note that the bind variables you enter in a parameterized SQL WHERE clause will require a valid value at runtime or a runtime exception error will be thrown. ..."
So, null as a valid value, seems to explain the behaviour in scenario (sc1).
"... In contrast, when you create a view criteria filter condition that references a named bind variable, you can specify whether the value is required or optional. ..."
But, the "Required" checkbox is also available (to uncheck) for Bind Variables that are not used in a View Criteria.
Specifically for the "Required" checkbox the help page says : "... Select if you want to make the value of a name bind variable required for any usage the references the named bind variable. For example, when the value is required (default), all view criteria items that reference the named bind variable will fail to execute unless a valid value is supplied at runtime. Alternatively, you can leave the value not required and use the Create View Criteria dialog to specify whether or not individual view criteria items require the value. ..."
Again, the focus is on View Criteria, although it can also be unchecked (configuring as NOT Required) for Bind Variables that are not used in a View Criteria (which does not seem to be something you should do).
But, still, there is a perspective here that makes this possibly very confusing:
- a View Object where clause with a required Bind Variable, no value set --> results in rows, and NO message about a missing value
- a View Object where clause with a NOT required Bind Variable, no value set --> results in no rows, and a message about a missing value
regards
Jan

Valuated purchase order??

Hi MM experts,
Please help me understand better about cost flow from purchase order.
1) Where in the purchase order document entered via SRM that identify if the purchase order is valuated or non valuated??
2) In 3 way match scenario, when good is received, costs hit a WBS/cost center and GRIR account got credited. This transaction shows up on WBS/cost center cost report as well as on FI document. When invoice is received, GRIR account got debited and vendor got credited. Even this second pair of transactions does not show up on WBS or cost center report; in some cases, it shows account assignment to a WBS/cost center on an FI document.
Why this is so?? The GRIR is cost element category 90. I thought that FI document should not show the account assignment since cost already hit a cost object at the good receipt?
Your help is much appreciated.
Monica
Edited by: Monica Kong on May 5, 2009 5:08 AM

Hi
In case of PR/PO with account assignment category as P (WBS element) following postings will takes place on WBS element report.
1) PO created
There will be only commitment on WBS element.
2) Good receipt
- Dr. Meterial acccount assigned to WBS = 100
- Cr. GR/IR = 100
at this point of time commitment will become actual because of goods receipt (FI posting)
In a FI document, we saw the above entries. In a WBS cost report, we see a good receipt transaction hitting a WBS = 100. So both places, we saw the same information which is above.
3) Invoice receipt. In an FI document, we saw
- Dr. GR/IR acccount assigned to WBS = 100
- Cr. Vendor =100
here only GR/IR clearing account which is an intermediate account will be hit but no posting will be made on WBS element.
The amount on PO will become actual on WBS element only at time of either Goods receipt or Invoice receipt but not in both transaction unless there is some price change.
Say PO value is 100 and GR value is 100 and IV value is 110. Then at time of MIGO, 100 will be the actual value on WBS element and at time of MIRO extra 10 will become actual on WBS element.
You can simulate this and check WBS ACTP (S_alr_87013542) report.
Thanks

JNDI lookup help please-Bean to Bean callout locally in same oc4j container

Hi - I'm fairly new to oc4j/ejb coding - I'm trying to get a simple application running to show that our code can be run in local mode - here is the scenario:
- Bean A (our application bean) within Application A
- Bean B (customer-developed bean) within Application B
- Bean B calls Bean A
- Bean A + Bean B are different application instances in the same oc4j container
- Jsp (residing in ApplicationB) within the same oc4j container calling Bean B, which invokes Bean A
====
I get the following exception (classes renamed to match scenario):
javax.naming.NameNotFoundException: oracle.apps.applicationa.common.BeanA not found in anothercompany.applicationb.common.BeanB
[java] at com.oracle.naming.J2EEContext.getSubContext(J2EEContext.java:93)
[java] at com.oracle.naming.J2EEContext.lookup(J2EEContext.java:76)
[java] at com.evermind.server.ApplicationContext.lookupJavaCompEnv(ApplicationContext.java:265)
[java] at com.evermind.server.ApplicationContext.lookupJavaComp(ApplicationContext.java:132)
[java] at com.evermind.server.ApplicationContext.lookup(ApplicationContext.java:306)
[java] at com.evermind.server.ApplicationContext.lookup(ApplicationContext.java:120)
[java] at javax.naming.InitialContext.lookup(InitialContext.java:347)
- BeanA is part of application named "AppA" in server.xml
- BeanB is part of application named "AppB" in server.xml
- AppB has attribute 'parent="AppA"' in server.xml
- If I change the Jsp file (residing in AppB) to call BeanA directly in the jsp code it works with no JNDI error, but changing the jsp code back to calling BeanB (which then calls BeanA) will throw the NameNotFoundException
My Stateless Session Bean is very simple and I have tried not setting the context and setting the context. I'm thinking this is a very simple configuration issue, but have not been able to find the solution yet... I've tried playing with the setSessionContext method in the session bean but have had no luck and am not sure what to set the variables to or even if I'm in the correct method...
public void setSessionContext(SessionContext ctx) {
if (verbose)
System.out.println("setSessionContext called");
Properties env = new Properties();
// tried all three context factories...
//env.put(Context.INITIAL_CONTEXT_FACTORY,"com.evermind.server.ApplicationInitalContextFactory");
env.put(Context.INITIAL_CONTEXT_FACTORY,"com.evermind.server.ApplicationClientInitialContextFactory");
//env.put(Context.INITIAL_CONTEXT_FACTORY,"com.evermind.server.rmi.RMIInitialContextFactory");
// not sure what url to put, if any
//env.put(Context.PROVIDER_URL, "ormi://localhost:23791/AppB");
env.put(Context.PROVIDER_URL, "ormi://localhost/AppB");
// tried with and without security principals... also tried sysadmin/sysadmin
env.put(Context.SECURITY_PRINCIPAL, "guest");
env.put(Context.SECURITY_CREDENTIALS, "welcome");
// mProps is a local transient member var of type Properties
mProps = env;
// mCtx is a local transient member var of type SessionContext
mCtx = ctx;
(edit) additional info:
- all beans/classes are in the same classes directory; there are no jar files, so in application.xml the specified classpath is "classes" which includes both appA and appB classes
- I verified that if I point (appB) jsp directly to BeanA it works with the "parent" attribute set on appB in server.xml, and when I remove the "parent" attribute on appB in server.xml the jsp throws a NameNotFoundException
edit #2:
another interesting note is that the exception message says "oracle.apps.applicationa.common.BeanA not found in anothercompany.applicationb.common.BeanB"
and usually the naming error I see is:
"oracle.apps.applicationa.common.BeanA not found in appB"
(as with the case when I use a jsp to access appA beans without setting the parent application of appB to appA)
- also my j2ee/home/applications/appB/WEB-INF/web.xml includes the oracle.apps.applicationa.common.BeanA ejb-ref tag
any help is appreciated - thanks!
Message was edited by:
oleon

Hi Steve
thanks for the reply - I currently have the parent attribute set in server.xml which is why the jsp connection works, but for some strange reason the bean cannot connect to other bean...
my server.xml has something like...
<application name="appA" location="appA" auto-start="true"/>
<application name="appB" location="appB" parent="appA" auto-start="true"/>
(ok, the tags might be off because I'm not in front of my workstation looking at the xml file...)
on another interesting note, in the session bean for AppB I've tried searching for the ejb bean for app B (itself basically) with the default initial context and it cannot find its own bean within the same application... the message would be something like "NameNotFoundException: anothercompany.applicationb.common.BeanB not found in anothercompany.applicationb.common.BeanB"

Import Manager Issue

Hi,
I have a bunch of records which have similarities in the description field of the source file and I need to consolidate all such records and add a separate ID to all such records in a different field. Example data is like this.
1 Rajani Kumar
2 Roman Rytov
3 Rytov Roman
4 Rytov R
5 Kumar Rajani
Here,1&5 represent the same Record for Rajani Kumar
and 2,3 & 4 represent same record for Roman Rytov.
So, I need to give an Id into a different destination field - A for Rajani Kumar(that is for 1&5 records)and B for Roman Rytov (that is for 2,3 &4 records).
Any solutions for this scenario while importing the records using Import Manager? Which is the best method ??
Regards,
RK

The process your are describing is a de-duplication process. For this exercise you need first to define a matching strategy and then implement this in MDM 5.5. This concepts has been demonstrated in a similar scenario identifying duplicate customers. Since this is rather extensive example I will only describe the highlights of what was done.
Based on the customer definition, a set of matching strategies was defined, each representing a specific matching scenario (e.g. one such strategy is: a matching consisting of : VAT number + Street + Name + Postal Code fields is assigned a Match Rage of 100, Match Group = Vat Number and Match Type = Matched.) Each element was initially assumed a weight (e.g. Vat Number = 34 ,street = 14 etc..) and the criteria was set to match results > 60 to be Match, result between 30 and 60 to Dubious and result <30 as Not matched.
For each of the above strategies a specific map was constructed based on the corresponding matching criteria per each strategy. Once these maps configured, the records for the de-duplication process are imported with this import only updating pre-defined fields for each of suspect matched records recording the Match Rate, Match Group and Match Type.
Once completed (all imports run ) you can see the MDM client all the matched records and group/sort them by Match type and grouping finally if you wish , merging the duplicate records while maintain their remote key information.
Kevin

Add hyperlink to open from falsh movie

Hi,
In Captivate 4: I am looking for a way to add a link to one of the slides that will open a URL.
Is there a way to do it?
Thanks,
Merav.

Hi Lilybiri
The Timeline appears to me that the Button object is pausing the slide before the Click Boxes, no? So if the slide is paused and the Click Boxes don't release the pause when they click, by clearing the "Continue" option, wouldn't that keep the playhead paused?
Note that I didn't configure a matching scenario to test fully, but it would appear to me that this is how it would work.
Cheers... Rick
Helpful and Handy Links
Captivate Wish Form/Bug Reporting Form
Adobe Certified Captivate Training
SorcerStone Blog
Captivate eBooks

Cash management of JRE6

Hello,
I wish to get some suggestions regarding our applet.
Our team is developing an applet for Windows IE.
We have encountered the following problem on JRE6.
(The applet works well on JRE1.5.)
The applet tries to get a JPEG image like,
Object o = curl.getContent ();
Where �curl� is an URL like �http://xxx.xxx.xxx.xxx/camera1�
This sentence fails and we found the following message in the log file.
Network: can�t find the cash entry [URL://http:// xxx.xxx.xxx.xxx/camera1, version: null]
(The actual message is in Japanese and I translated it to English)
If we set the cash management of the Java plug-in to OFF (disable), our applet works well on JRE6 as well.
I�d appreciate it if you give us any suggestions regarding the problem.
We especially want to know the difference of cash management between JRE1.5 and JRE6.
Thanks in advance for your help.
This is my first post and I'd apologize for my strange English expression.
Tetsusan,

Sure It will work, just be very careful when it comes to performance of the autoreconciliation process. I've been at a site where the autoreconciliation process takes roughly 2-4 hours per day for 300 transactions. Im sure the poor performance relates to the fact that they have in excess of 1 million journal lines and that GL journal Line description isn't indexed.
To compound the problem, they then wanted another bank a/c in Cash Management (approx 2000 txns per day). Reconciling to je_line_descriptions just wasn't going to work so we took the option of using the cash management reconciliation open interface. From the feeder system, we post bank entries in summary to GL, load detail (cheques & deposits) into open interface and reconcile to it. The Autoreconciliation takes roughy 40 minutes to run, a significant improvement on the journal line matching scenario.
Refer to the cash management documentation for more detail on the open interface and depending on which way you go my my other tip is TEST TEST TEST.

Attempt to access expired blob handle

We get a random error in our test environment after upgrading from SQL server 2008 R2 to 2012.
System.Data.SqlClient.SqlException (0x80131904): A system assertion check has failed. Check the SQL Server error log for details. Typically, an assertion failure is caused by a software bug or data corruption. To check for database corruption, consider running
DBCC CHECKDB. If you agreed to send dumps to Microsoft during setup, a mini dump will be sent to Microsoft. An update might be available from Microsoft in the latest Service Pack or in a QFE from Technical Support.
A severe error occurred on the current command. The results, if any, should be discarded.
Location: tmpilb.cpp:2576
Expression: fFalse
SPID:   55
Process ID: 1552
Description: Attempt to access expired blob handle (3)
The weird thing is that is looks almost exactly like the below kb article, we are using a matching scenario as the symptons described in the below kb.
support.microsoft.com/kb/2644794/
It did not make this issue in 2008 R2, we then took the database from 2008 R2 and attached to the 2012 db server, and then
Event viewer (throws a lot of the 4 below errors):
Level Date   Source   Event Id Task category
Error 01-04-2014 17:42:00 MSSQL$DGOFFICEMAINSQL 17310  Server
A user request from the session with SPID 58 generated a fatal exception. SQL Server is terminating this session. Contact Product Support Services with the dump produced in the log directory.
Error 01-04-2014 17:41:59 MSSQL$DGOFFICEMAINSQL 17065  Server
SQL Server Assertion: File: <tmpilb.cpp>, line = 2576 Failed Assertion = 'fFalse' Attempt to access expired blob handle (3). This error may be timing-related. If the error persists after rerunning the statement, use DBCC CHECKDB to check the database
for structural integrity, or restart the server to ensure in-memory data structures are not corrupted.
Error 01-04-2014 17:41:58 MSSQL$DGOFFICEMAINSQL 3624  Server
A system assertion check has failed. Check the SQL Server error log for details. Typically, an assertion failure is caused by a software bug or data corruption. To check for database corruption, consider running DBCC CHECKDB. If you agreed to send dumps
to Microsoft during setup, a mini dump will be sent to Microsoft. An update might be available from Microsoft in the latest Service Pack or in a QFE from Technical Support.
Error 01-04-2014 14:47:41 MSSQL$DGOFFICEMAINSQL 17066  Server
SQL Server Assertion: File: <qxcntxt.cpp>, line=1137 Failed Assertion = '!"No exceptions should be raised by this code"'. This error may be timing-related. If the error persists after rerunning the statement, use DBCC CHECKDB to check
the database for structural integrity, or restart the server to ensure in-memory data structures are not corrupted.
(DBCC CHECKDB) has been run, and no errors seems to occur
All usefull ideas are welcome :)

OMG weirdest thing ever:
I have installed SQL server 2012 SP1 on my Windows 7 machine, running the program against the same database, then I tried to replicate the error:
Using IE 11 going to the page that causes the issue, validating in event viewer at the same time, nothing happens, everything looks good.
Using Newest FF or Chrome, CONSISTANTLY throws 3 event logs, everytime I go to the page, 2x event id 17065 and 1 x 17066, I assume it throws the other errors upon too many, or too frequent errors.
But WHY, I simply cannot wrap my brain around why such a feature inside the SQL server, should be able to act differently based on the browser.
Error 01-04-2014 17:41:59 MSSQL$DGOFFICEMAINSQL 17065  Server
SQL Server Assertion: File: <tmpilb.cpp>, line = 2576 Failed Assertion = 'fFalse' Attempt to access expired blob handle (3). This error may be timing-related. If the error persists after rerunning the statement, use DBCC CHECKDB to check the database
for structural integrity, or restart the server to ensure in-memory data structures are not corrupted.
Error 01-04-2014 14:47:41 MSSQL$DGOFFICEMAINSQL 17066  Server
SQL Server Assertion: File: <qxcntxt.cpp>, line=1137 Failed Assertion = '!"No exceptions should be raised by this code"'. This error may be timing-related. If the error persists after rerunning the statement, use DBCC CHECKDB to check the database
for structural integrity, or restart the server to ensure in-memory data structures are not corrupted.
I am still very interested in good ideas!!!

Re: Informatica 9.5 client and 9.6 server

Duly noted, thank you - and thank you for sharing the real-world experience. I will avoid server/client mis-matches at all costs.

Nico -Addressing your customers as "Kids"? "Don't give a damn"? "no use in discussing this any further"? Really? That's rather harsh and unprofessional - especially coming from an employee of Informatica, don't you think?... Had you been drinking? There's nothing wrong with asking the questions. Sure, I think it's pretty easy to assume Informatica wouldn't officially support such a mismatch, but you know what else Informatica doesn't "support"? running their products on overprovisioned virtual environments - which is ridiculous considering over-provisioning is the reason many companies go to virtualized hardware in the first place. So sometimes the official statement of support is nothing more than a technicality. Of course a customer should not expect the vendor to officially list support for myriad combinations of client/server versions, but that doesn't mean it can't work - you even state that in your reply regarding v8. And you know, in the real world outside of a lab environment, there are some very legitmate reasons why such a mis-match scenario might be the only option, at least for a short period of time. In my case it's because my company requires all workstation software to go through a lenghty certification process before it can be packaged for distribution, but I need to get my upgrade going on the server side asap. We've got the 960HF1 client already certified, and the 960HF2 client certification is underway, but I need to upgrade to 960HF2 server now. The reason I asked is to try to find out if anyone in the real world is doing this and if there are risks. I wasn't really interested in the "official" response considering that's what I get from Informatica Global Support after paying who knows how many hundreds of thousands of dollars in annual support fees. What about another hypothetical scenario: an installation with hundreds of users goes through an upgrade but not all those users get the client install completed immediately. A developer "really needs to get some work done now" and doesn't have time to perform the client install so they use the old client with the new server. I'm sure this happens multiple times per day, every day among Informatica customers in the real world so maybe Informatica should consider developing clients that can more easily tolerate a minor version or hotfix mis-match. Makes sense, doesn't it? Backwards compatibility is certainly not a new concept... I know our team in India routinely has major problems trying to get their clients updated in a timely fashion due to the fact that the client tools package is a whopping 2.8 GB - that can take hours to download in some places, assuming the transfer isn't interrupted... Maybe next time check your labcoat and attitude at the door before coming in, thanks.

Recovery scenario - Voting disk does not match with the cluster guid

Hi all,
Think of you can not start your guest VMs just because it has a corrupted system.img root image. And assume it contains 5 physical disk( which are all created by the RAC template) hence ASM on them.
What is the simplest recovery scneario of the guest vms (RAC)?
Can it be a feasible scenario for recover of the availablity? (Assume both of the RAC system images are corrupted and we prefer not a system level recovery rather than backup / restore)
1. Create 2 RAC instances using the same networking and hostname details as the ones that are corrupted. - Use 5 different new disks.
2 Shutdown the newly created instances. Drop the disks from the newly created instances using VM manager.
3. Add the old disks whose system image is failing to be recoverd but ASM disks are still in use (from the newly created instances using VM manager.) to the newly created instances.
4. Open the newly created instances
Can we expect the ASM and CRS could be initialized and be opened without a problem?
When I try this scenario I get the folllowing error from the cssd/crsd .
- Cluster guid 9112ddc0824fefd5ff2b7f9f7be8f048 found in voting disk does not match with the cluster guid a3eec66a2854ff0bffe784260856f92a obtained from the GPnP profile.
- Found 0 configured voting files but 1 voting files are required, terminating to ensure data integrity.
What could be the simplest way of recovery of a virtual machine that has healthy ASM disks but corrupted system image?
Thank you

Hi,
you have a similar problem, when trying to clone databases with 11.2.
The problem is that a cluster is uniquely identified, and this information is hold in the OCR and the Voting disks. So exactly these 2 are not to be cloned.
To achieve what you want, simply setup your system in that way, that you have a separate diskgroup for OCR and Voting (and ASM spfile), which is not to be restored in this case of szeanrio.
Only all database files in ASM will then be exchanged later.
Then what you want can be achieved.
However I am not sure that the RAC templates have the option to install OCR and Voting into a separated diskgroup.
Regards
Sebastian

Matching Scenarios

Similar Messages

Maybe you are looking for