Experiences with 9.1.4 FP1?

[frosterrj]

Just got the Fix Pack one for our 9.1.4 installation, and curious to see in anyone has actually applied it and NOT seen any negative consequences, or, what the downsides of the FP1 install produced.

Thanks,
Robert

[shorsted]

Hi Robert

We are running on 64bit so the following may not apply to you, however...

We downloaded the 9.1.4 release that was available for download earlier this year, carried out what we thought was thorough testing and put it into production only to find that our server then started crashing (shutting itself down) at least once a week. Also we found that the client connection when browsing cubes was really slow, much slower than version 8.4.5

Cognos then advised us to update to FP1 which we did. Since then things have been much more stable. The Client is vastly quicker (although I'm sure still not as quick as 8.4.5!) and until recently our server had only crashed once. We sent files from that crash to Cognos who asked us to add additional logging to out tm1s.log.properties files to help them analyse the data if we crashed again. That was about a month ago and since then we've crashed twice more! Again we've sent the files over to Cognos and they have apparently found 2 different reasons for our crashes
1. Stack overflow errors on rules
2. Locking issue with a query

They have said that they have experienced the same issues at other clients and whilst the stack overflow has been dealt with in 9.4 they are now also working on a fix for 9.1.4 for both issues

I have also experienced a lock on the server when running processes which did not result in a crash per se, but did not release itself, and could not be killed off using TM1Top, nor could we stop the TM1 Service in anyway at all (including within the services, cmd, kill and End process) The only the service could be stopped and restarted was to physically shut the machine down

Having said all of that 9.1.4 FP1 is certainly 10 times more stable than 9.1.4. All of the above issues we also experienced in 9.1.4 but with more frequency

Regards
Sarah

[frosterrj]

Sarah, thanks for the information. We are running 64-bit 9.1.4 as well, but have experienced nothing like what you describe.

Good to hear the fix pack seems to be an improvement and doesn't _seem_ to introduce new regressions.

Robert

[Jeroen Eynikel]

Does anyone have a link to this fixpack for 9.1.4? All I can find on the IBM website are fixpacks for 9.4..

[frosterrj]

https://testcase.boulder.ibm.com

login: anonymous/email address
Click on fromibm folder
Then click on im folder

there are 64 and 32-bit versions

[Jeroen Eynikel]

Robert,

thanks a lot; however I only see the x64 version in there and I actually need the x86 version.

am I missing something?

Jeroen

[Jeroen Eynikel]

JFYI:

I contacted Mike Cirillo and he put it on the FTP for me. If anyone else needs it => fromibm/im/mike now has the x64 and x86 versions.

Jeroen

[rfielden]

Thanks for posting this information. We were given 'bad' links by support which didn't allow us to download this update. This was extremely helpful.

[Jeroen Eynikel]

Btw:

just from my first tests:

- My TI processes with dataloads show no increase in performance, in fact they are about the same speed on 9.0 sp2, 9.1sp4 and 9.1 sp4 FP1
- My TI processes with calculations in them (it is a complex model with loops of TI processes) are about 3 times faster on FP1 than on the 'un fix packed version'
- Refreshing an excel workbook with a lot of sheets is about 3 times faster on FP1 as well.

If the savedaa fix is working and they didn't introduce any new bugs I'm a happy camper