New Content Database. Framework Manager configuration

[locus]

Hi Folks,

As part of my deployment and upgrade to 10.2.1 from 10.1.1, I have switched from the tomcat database to SQL server.
Everything connects fine in Cognos config and startup is OK, but the Framework manger configuration no longer finds the content manager.

Error:

[CSK Request Test]
[ ERROR ] CAM-CRP-1278 An active Content Manager was not found.

If I import my content store I can open and publish packages in FM, but regardless, the above error persists.

Is there something that needs to be done in the FM config when the content database is changed?

Thanks,
Locus

[locus]

Update:

While I can open, verify and publish FM models, testing of data causes an immediate Appcrash.

This is common to both QA and Prod environments

In addition, I am getting a mixture of dpr-err 2056 and dpr-err 2077 errors when I run reports.
All data sources test successfully in administration AND reports validate 100% in report studio.

This is relevant only for the Prod environment.

The only difference between QA and prod is QA = 10.2.1 FP1 and Prod = 10.2.1 FP2
Dev which is running 10.1.1 FP2 with exactly the same setup across the board as the others runs like a dream.
What may be worth mentioning is that the 10.2.1 content store is an import from 10.1.1.
Also that the 10.1.1 configuration was imported into 10.2.1 and the relevant settings for gateway, despatcher and content store altered accordingly

Completely stumped. I'm one step away from uninstalling everything and starting from scratch

[MFGF]

Hi,

Did you update your FM to be the same version as your server? You're not trying to use FM 10.1.1 to connect to your 10.2.1 BI server, by any chance? If you do have FM 10.2.1, did you apply the fix pack to this as well as to yor BI server?

MF.

[locus]

Hi MF,

Yes both running 10.2.1 FP2 on Prod
QA running 10.2.1 FP1

Both have the same error

Locus

[MFGF]

Hmmm. Ok. The DPR-ERR 2056 messages indicate that your BIbus processes are dropping out, so you have a deeper problem than FM configuration. Have you checked you have enough temp disk space free on the server? The DPR-ERR 2077 messages might also be a facet of this?

MF.

[locus]

HI,

Yes plenty of space.

What's strange is that on QA, none of those errors are present. It runs fine despite the FM config error.
I'm dealing with that as a 'prod specific' issue.

For now I am trying to understand the error in FM config.
I'm not finding much literature for the error from Framework manager configuration:

[CSK Request Test]
[ ERROR ] CAM-CRP-1278 An active Content Manager was not found.

Cognos runs so the main settings are fine.
The keys seem to test OK in FM. Just the last step of three when looking for the content manager fails.

Locus

[locus]

As an aside, I did recreate all the keys so that isn't it.

I have added some logging to BMT.ini and will see if it spits something out

[MFGF]

When you go into the 32-bit Cognos Configuration for Framework Manager, are the dispatcher URI and gateway URI pointing to the same instance? Is it the instance you want it to point at? What happens when you save the config - does it pass the "generating cryptographic information" stage ok?

MF.

[locus]

Yes, yes and yes.

[MFGF]

Oh. Ok. At least that means the config is ok. Is it a single-server instance or is CM on a different server to the application server?

I think if I was in your situation, I would try the following:

1. Create a deployment of the entire content store
2. Stop the Cognos service
3. Create a new, empty database in SQL Server
4. Modify your configuration on the server to point to this as the content store
5. Save and start the configuration and let it initialise the new content store
6. Restore the deployment you created

This should (in theory) make sure your content store is properly configured and consistent, and rule out any corruption or hiccups that might have occured when it was upgraded.

If you still get the same issues, make a note of all your configuration settings in Cognos Configuration, rename cogstartup.xml to cogstartup.old, copy csk, signkeypair and encryptkeypair folders to a location somewhere safe then delete them, then go into Cognos Configuration and set up your configuration from scratch, save and start. You will need to re-configure FM after doing this as the encryption keys will have changed.

Good luck!

MF.

[locus]

MF,

Tried everything, config from scratch, regenerate key, reinstalled FM. No change.

Now I have a new issue that when I try to log in to connection, it keeps coming back to the logon page after I enter details.
This happened to me long ago and I reset the binding credential which sorted it. Not this time.
Everything tests 100%, redid the IIS setup. Still no login.
No error just a loop back to the login page.

Also imported working configs for both Cognos and FM, changing the relevant fields as before and regenerating keys. No joy
I wonder if it might be best to uninstall the lot and do it again  >:(
I am quite literally out of ideas, I have never seen anything like it

Locus

[MFGF]

Oh dear! Sounds like you're having a rough time!

One thing to try - clear your browser cache to make sure nothing old is lurking around in there...

Good luck!!

MF.

[sdf]

good point from MFGF, most of my usual issues with cognos resides with the cache, also with the browser security setting and compatibility.

[locus]

Right, a fresh install of everything solved the FM problem.

One issue remains and this one doesn't want to be resolved.
Every time a report is run I get the error

DPR-ERR-2056
The Report Server is not responding
....In addition to the associated PID process is shutting down.

The PID process being referenced (In every case) is the BIBUSTKSERVERmain process, which spawns and dies after a couple of seconds, with the report error following shortly after that.
My understanding is that the process should never drop.

QA and PREPROD are identical in every way and QA is fine.
Cognos has be reinstalled twice again using both FP1 and FP2, brand new content stores, IIS from scratch.
Always the server time out error or the PID error (Both of which are in the logs every time)

As a trial I went to Report studio to validate a report and I get a HAL error, referencing an HTTP 404 issue amongst other things.
This I have attached.

IBM and our IBM Cognos partner support ran out of ideas some time ago. So far I have tried more than has been suggested (Including clearing the browser cache)
What is even more odd, is that maybe one execute out of 30, the reports will run.

Maybe you guys can recognise something from the attachment or have seen this before?

I think I see the end of my wit! :)

[sdf]

as far as i can remember, i had one issue similar to this. And i did the same thing you did, reinstall cognos.
What i did was a clean uninstall and reinstall. Also i remember my team leader pointed out to me to change the content store upon re-installation and re-create the iis settings (isapi and cgi).

[locus]

Yep, pretty much all from scratch, but no joy.

It's logged with IBM.

I know some folks are having no majors with 10.2.1 but it still seems a little flakey to me.
Possibly environmental but in a company the size of mine, with endless VM's lined up and very tightly managed, highly unlikely.

[locus]

A buggy VM was the cause of the problems.

New VM. First time install with not s single issue.

Major waste of time