on 11-13-2013 4:03 PM
BusinessObjects Enterprise XI 3.1 SP3
Windows 2008 R2 64-bit (2 servers, clustered BOXI install)
Shared FRS
I run the reposcan like this (on either server, both give the same results):
reposcan -dbdriver oracledatabasesubsystem -connect "UID=xxxxx;PWD=xxxxx;DSN=xxxxx" -inputfrsdir f:\Input -outputfrsdir f:\Output -count 20000
This is with the f: drive mapped to the shared FRS on the network. It works, I suppose, I mean it doesn't crash, but the result reports every single file in the FRS (all 6,312 of them!) is missing in the repository.
"Scanning 988 repository objects and 6312 files........................."
"Found 3 application reparable items and 6312 manually reparable items."
An example of one of the 6312 items in the report:
"File f:\Input\a_000\006\000\1536\~ce1388671c4cfc4f12.rpt exists in the Input or Output FRS, but there is no corresponding InfoObject in the repository."
This is pretty useless, and doesn't seem accurate, since for the most part, BOXI/Infoview is working fine. There's no way every single file has lost the connection to the repository... or is it? Can BOE continue to function without the FRS being synced with the repository on the database? Is reposcan reporting accurately? What's going on here? Am I not running reposcan correctly?
Can you ensure that note 1267338 - How to use Repository Diagnostic Tool (reposcan) in XI
3.1 has been followed to dot. (Do not run the RDT against a live CMS database and FRS)
Also I suspect whether note 1927706 - Repository Scanner fails with a segmentation fault error
after upgrade would be applicable to your case or not. Can you check the eventviewer for logs mentioned in this note?
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Thanks for the reply. The links you provided don't work for me-- are they public facing (https://support.wdf.sap.corp)?
No, but he includes the correct Note ID's, so just use the prefix http://service.sap.com/sap/support/notes/ followed by the Note ID number.You will nned an S-ID to access the SAP Notes.
It appears though, that the reposcanner can see the files in the FRS but cannot see them in the repsoitory. You can check this in the query builder and query for a file with SI_ID = 1536 which is the example you gave. If the query builder can find it, then there is something wrong in the tool, or just in the commandline you are using.
I have used the tool in the past a number of times and have not seen that behaviour though.
Thanks for the help finding the docs, I was able to find them. Yes, I have shut down the CCM services to run the reposcan, and no, there is no error in the event viewer. The reposcan program works, it just, like you mention, doesn't seem to find anything in the database for any of the files in the FRS.
When I went into the database to try to track down one of the file's entries, I wasn't able to find an SI_ID column. I did find SI_GUID, SI_CUID, and SI_RUID, but they are all encrypted in some way-- their values all look like "AZ0NIQgXBZNKvR3wk0KBjH4". This is in the table cms_infoobjects6. I don't know enough about the CMS database and/or BOE to get any further with a visual search.
If you have any other ideas, I'd love to hear them.
Hi Randy,
Just to make you clear
For every BI Content (Universe, Report,Connection) you should have meta data in two different locations.
1. CMS Database
2. File Store (Input & Output FRS)
CMS database holds the pointers to the Objects of interest where as FRS holds the actual physical file of the Object. By running RDT you can always find the inconsistancies of objects in both the cases.
I would recommend you to try with Querybuilder for metadata search as CMS database is encrypted and cannot be query it using traditional SQL. To start with Query builder you can refer here
http://scn.sap.com/docs/DOC-42952
Regards
Mani
Wow, another tool I had no idea existed! Query Builder! That's awesome. I am admittedly a total noob with this stuff-- Typical scenario, someone retires, and I inherit ownership with no training offered. So I appreciate the information you probably regard as really basic info. It really helps.
So, now that I can query the CMS, I'm back to this reposcan just not working right. For example, the scan reported this file:
Inconsistent Filename: | ~ce109c5195695d211.wid | |
Directory: | f:\Input\a_092\003\000\860\ | |
Total Inconsistencies: | 1 | |
Inconsistency | Warning | Suggestion |
1. | File f:\Input\a_092\003\000\860\~ce109c5195695d211.wid exists in the Input or Output FRS, but there is no corresponding InfoObject in the repository. | Please republish the file. The application will not publish this file for you. |
But this file does indeed exist in the CMS-- here's the dump, sorry it's so huge.
Properties | |||||||||
SI_HIDDEN_OBJECT | false | ||||||||
SI_CREATION_TIME | 1/13/11 10:10:23 AM | ||||||||
SI_SOURCEDOC_ARTIFACTS |
| ||||||||
SI_CONTENT_LOCALE | fr_FR | ||||||||
SI_APPLICATION_OBJECT | false | ||||||||
SI_QAAWS |
| ||||||||
SI_KIND | Webi | ||||||||
SI_SHARED_DOC |
| ||||||||
SI_RUID | ASy7cJAa8rBAt.qEod_MuyI | ||||||||
SI_HAS_CHILDREN | false | ||||||||
SI_ROLES_ON_OBJECT |
| ||||||||
SI_SCOPEBATCHDOC_ARTIFACTS |
| ||||||||
SI_CHILDREN | 0 | ||||||||
SI_IS_SCHEDULABLE | true | ||||||||
SI_INSTANCE | false | ||||||||
SI_SIZE | 343150 | ||||||||
SI_PLUGIN_OBJECT | false | ||||||||
SI_SENDABLE | true | ||||||||
SI_TEMPLATE_CUID | ASy7cJAa8rBAt.qEod_MuyI | ||||||||
SI_FILES |
| ||||||||
SI_PARENTID | 859 | ||||||||
SI_OBJECT_IS_CONTAINER | false | ||||||||
SI_TABLE | 0 | ||||||||
SI_KEYWORD | WebI Document Sample Web Intelligence | ||||||||
SI_OBTYPE | 357 | ||||||||
SI_CUID | ASy7cJAa8rBAt.qEod_MuyI | ||||||||
SI_FLAGS | 2050 | ||||||||
SI_OWNERID | 12 | ||||||||
SI_FULL_CLIENT_DOC |
| ||||||||
SI_GUID | AUjSlH4hzJRDpFtr22zX8Mo | ||||||||
SI_ARTIFACTS_SCOPEBATCH |
| ||||||||
SI_READ_ONLY | false | ||||||||
SI_SYSTEM_OBJECT | false | ||||||||
SI_REFERRING_DOCS |
| ||||||||
SI_RUNNABLE_OBJECT | false | ||||||||
SI_INSTANCE_OBJECT | false | ||||||||
SI_PROGID | CrystalEnterprise.Webi | ||||||||
SI_UPDATE_TS | 1/13/11 10:10:23 AM | ||||||||
SI_NAME | Formatted Report for Drill | ||||||||
SI_CORPORATE_CATEGORIES |
| ||||||||
SI_PUBLICATIONS |
| ||||||||
SI_ID | 860 | ||||||||
SI_WEBI_DOC_PROPERTIES | <?xml version="1.0" encoding="UTF-8"?><WEBI_DOC_PROPS><WEBI_PROP NAME="VERSION" VALUE="1.0."/></WEBI_DOC_PROPS> | ||||||||
SI_ARTIFACTS_SCOPEBATCHDOC |
| ||||||||
SI_OWNER | Administrator | ||||||||
SI_PERSONAL_CATEGORIES |
| ||||||||
SI_TOPIC_TOPIC_FOR_SUBJECT_GENERICENTITY |
| ||||||||
SI_AUTHOR | Administrator | ||||||||
SI_PARENT_FOLDER_CUID | AbN5UQGdkS5IoqWiL6cbbcc | ||||||||
SI_DESCRIPTION | This sample document demonstrates a well-formatted report that is ready for drill analysis. User must have appropriate DRILL rights to work with this sample. | ||||||||
SI_ARTIFACTS_SOURCEDOC |
| ||||||||
SI_UNIVERSE |
| ||||||||
SI_SHORTCUTS |
| ||||||||
SI_PARENT_FOLDER | 859 | ||||||||
SI_PARENT_CUID | AbN5UQGdkS5IoqWiL6cbbcc | ||||||||
SI_COMPONENT | false | ||||||||
Scheduling Info | |||||||||
SI_TYPE | 357 | ||||||||
SI_SCHEDULE_INTERVAL_MONTHS | 0 | ||||||||
SI_SCHEDULE_INTERVAL_NDAYS | 0 | ||||||||
SI_SCHEDULE_TYPE | 0 | ||||||||
SI_STARTTIME | 12/12/07 12:33:15 AM | ||||||||
SI_OBJID | 860 | ||||||||
SI_SCHEDULE_INTERVAL_NTHDAY | 0 | ||||||||
SI_ENDTIME | 12/12/17 12:33:15 AM | ||||||||
SI_NAME | Formatted Report for Drill | ||||||||
SI_TIMEZONE_ID | 0 | ||||||||
SI_SCHEDULE_INTERVAL_MINUTES | 0 | ||||||||
SI_SCHED_NOW | true | ||||||||
SI_SCHEDULE_INTERVAL_HOURS | 0 | ||||||||
Processing Info | |||||||||
SI_PROMPT_USER_VIEWING | false | ||||||||
SI_PRECACHE_XLS | false | ||||||||
SI_WEBI_PROMPTS |
| ||||||||
SI_PRECACHE_HTML | false | ||||||||
SI_FILES |
| ||||||||
SI_PRECACHE_PDF | false | ||||||||
SI_HAS_PROMPTS | false | ||||||||
SI_OBJID | 860 |
Maybe it's the wrong type of object? I'm stumped.
Have you tried refreshing the appropriate report in BI Launchpad? There could be a possibility of communication failure while exporting the report to repository. If you can able to open &refresh the report try export the same report by overriding old. Then try to run RDT and check whether the issue still persists. Hopefully it might not occur again. Looking forward your response. Regards Mani
HI Mani,
We will start looking into a very different direction. This was just to identify whether reposcan is working as expected provided by berend.
@Randy : Comparing the two we can check for the availaability of the report, and understand that the metadata and the physical file is in sync or not.,
If the metadata and physical file are not in sync, it means reposcan has actually identified there is a possible corruption occuring at CMS Db level.
This query gives the path I would expect:
SI_FILES
SI_FILE1 | ~ce109c5195695d211.wid |
SI_PATH | frs://Input/a_092/003/000/860/ |
SI_VALUE1 | 343554 |
SI_NUM_FILES | 1 |
and the file is indeed there... or, it's in a location "\\UNC\to\location\Input\a_092\003\000\860". Now I'm wondering about the alignment of SI_PATH and the actual file location. Above, SI_PATH is set to
frs://Input/a_092/003/000/860
and the InputFileRepository setting in the CMC is set to
\\UNC\to\location\Input
So.. as long as the "frs://" portion of SI_PATH is "\\UNC\to\location\", WITHOUT the Input portion of the path, then they're equal. But, SI_PATH includes "Input". And, the InputFileRepository setting includes "Input". So... I'm wondering, does BOE think the file should be in location "\\UNC\to\location\Input\Input\a_092\003\000\860" (notice the doubled up Input folders)?
I'm going to have to test this notion.
Thanks to everyone for trying to help to me, my issue was resolved.
I created one Dimension object in the CMS Universe by using existing “Webi Prosperities” Dimension Object and it’s worked.
Note: New Dimension Object As: text(Properties.si_files,"500")
Query:
SELECT
Properties.si_name,
Properties.si_kind,
text(Properties.si_files,"500")
FROM
Properties
WHERE
Properties.si_kind = ('Webi',’Crystal Report’,’Excel’)
Once again thanks to every one!!
Ramesh
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Randy Whitaker wrote:
BusinessObjects Enterprise XI 3.1 SP3
Windows 2008 R2 64-bit (2 servers, clustered BOXI install)
Shared FRS
I run the reposcan like this (on either server, both give the same results):
reposcan -dbdriver oracledatabasesubsystem -connect "UID=xxxxx;PWD=xxxxx;DSN=xxxxx" -inputfrsdir f:\Input -outputfrsdir f:\Output -count 20000
It may not help much but try running reposcan with the -scancms and -scanfrs switches. this will require two runs of reposcan. That is:
Run this:
reposcan -dbdriver oracledatabasesubsystem -connect "UID=xxxxx;PWD=xxxxx;DSN=xxxxx" -inputfrsdir f:\Input -outputfrsdir f:\Output -scancms
And then run this:
reposcan -dbdriver oracledatabasesubsystem -connect "UID=xxxxx;PWD=xxxxx;DSN=xxxxx" -inputfrsdir f:\Input -outputfrsdir f:\Output -scanfrs
Compare the results. I've found that you get different information between the two switches. It may or may not help.
Like everyone else, I haven't run into this problem before.
I'll ask a dumb question also, are you sure that your DSN is pointed at the corresponding database for your filestore?
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Thanks for the reply John,
I tried running two separate times as you suggested (one with -scanfrs, one with -scancms) and the results are the same.
The DSN is pointed to the Oracle ODBC DSN that points to the database containing the CMS schema. I can only assume reposcan finds the schema, since it a) doesn't complain about not finding the tables it uses, and b) it reports it is "Scanning 988 repository objects and 6410 files......". I don't know how it would get a cound of repository objects if it doesn't find the repository.
Best thing to do in that case is to run the reposcan parameter for the FRS with the exact path as you see it in the properties of the CMC > Servers > iFRS / oFRS, so \\UNC\to\location\Input.
When you use it that way, what does reposcan say?
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
I'm finally getting back to this, had a busy couple of days!
I ran reposcan using the full UNC (not a mapped drive) and the results are the same. Every file is reported as not found in the DB. Using Query Builder I've verified there are rows in the CI_INFOOBJECTS table that match the path and filename exactly. Even file sizes match. There must be additional values in the row that are checked and found to be inconsistent. It's not a simple matter of files missing or misplaced-- everything seems to be in the right place.
I also tested the input\input theory by copying a file with its full path to the input\input location and all that did was add another file to the reposcan report saying this file can't find a match in the CMS. So it's not that...
At this point I'm stumped. Since everything's humming along and working, I can only assume the reposcan is faulty. Without knowing exactly how reposcan makes the decision if an object is missing from the CMS (which columns are checked, exactly), I can't track down the reason for the results.
Thanks everyone for trying to help. I learned a lot.
User | Count |
---|---|
88 | |
23 | |
11 | |
9 | |
8 | |
5 | |
5 | |
5 | |
5 | |
4 |
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.