Is it just me?

[justinjbird]

I'm trying to get an idea of whether this is the norm, or whether we have specific issues with our environment...

We've been running a 10.2 environment since March this year, upgraded from 8.4.1 (previously upgraded from 8.4), the environment as it is has been in existence for around 3 years.  We have constantly, on all versions, had regular inexplicable report fails, issues with jobs, missing items in run history, entire jobs dropping out of run history, issues with schedules, forced into needing to rebuild the content store on many occasions. 

Our latest inexplicable issue, is a job that won't complete.  We have reported a few errors this morning but the job that has failed still says "Pending", cannot be cancelled and restarts don't clear it.  It has been rerun, but it sits there as a constant reminder of this mornings errors.  This happened a month or so ago resulting in us having to run a SQL script over our content store to clean it out after it sitting there unfinished for 7 days.

Whenever we have problems, troubleshooting is a nightmare - all error messages result in searches that return a multitude of causes, none of which are mine, and the only reasonable response I get from our support providers is to find the latest fix pack (which then naturally leads to more errors).  We seem to constantly be chasing the next erroneous fix pack!  Our support provider seems to do little more than google the responses we've already identified, so I'm unable to getting any meaningful reponse from them (which I realise is an issue in itself).

I look after a SQL / Cognos area and Cognos requires a significantly larger amount of our time to look after and babysit compared to SQL despite the fact that maybe 80% of our workload is on SQL.  For me, Cognos is an horrific product which is far too buggy and costly to maintain.  Before I commence plans to move our reporting content away from Cognos (which admittedly would be a significant piece of work), I just wanted to reach out to the wider audience and find out whether it's just us?  Does anyone else seem to have this constant battle to keep Cognos happy?

To put things in perspective, we have rebuilt the content store at least 3 times in the last year, something equivalent to rebuilding the system databases in SQL, something you might only have to do once in your LIFETIME as a DBA.

[justinjbird]

I should point out, we aren't DBAs trying to run a Cognos environment :P  I have two certified Cognos administrators looking after this evironment but due to resource constraints none are dedicated to it.

[bdbits]

I am not the admin here but work closely with them. I have been using Cognos about 4-1/2 years from 8.3->10.2.1. Like most "enterprise" software it has it quirks and oddities. But you seem to be having way more trouble than you should. For example, I am quite certain we have never had to rebuild a content store, ever.

You might want to look through http://pic.dhe.ibm.com/infocenter/cbi/v10r2m1/nav/0_17 and see if what you have is set up properly. It sounds like something is either not configured correctly, or you have insufficient software/hardware resources. Also, do you have no access to Cognos support other than through your vendor? Cognos tech support is not always the best, depending on who gets assigned to your problem. If you are persistent with them, they can usually eventually help you solve the problem. Also use cognoise.com, and developerworks.com (IBM community) can be helpful. And there are consultants out there who really can help, like our gracious host for example. :-)

So you want to replace Cognos with SQL Server Reporting Services? We have both here, though I am not very involved at all with SSRS and am not an admin/DBA. If you think SSRS is an install-and-forget-it kind of product, think again. You will have issues there as well, just different ones. My own perception is that it is a simpler interface for report authors, but somewhat less capable, not as flexible and like any product has its own limitations. You will have just as much back-end work as far as setting up the equivalent of FM or Transformer models. BI in SQL Server is pretty reliant on Excel for the heavy duty work, and you will probably want SharePoint as a front end (but don't run integrated). In the end, I think you would be trading one can of worms for another, so it is a matter of what fits your organization and staff skills better.

[TheCognosDave]

I've seen lots of odd problems, but never the ones you describe here. 

I have had problems with some databases having problems over time though, which I could see contributing to this problem.  For example, if i hammer MySQL with intense OLAP for month, and throw in a few blocking errors, cascading integrity check failures due to read block, you can end up with a hash map on the cluster index which causes you to need to re-build the table.

I've also seen this if you create SQL Server tables with varchar() fields as part of the index.  Since indexes have a hard limit on size, you can actually corrupt the index unknowingly on a seemingly simple INSERT statement ;-) ... again cause you to need to drop/re-create table and index.

If I was in your situation, I might consider a high end database.  The most success I've seen has been on Oracle.  10g, 11, Oracle RAC, Exadata... it's all good.  I realize it costs a lot more, but it's free to try and see if your problems go away. 

I guess I should mention, I stand to gain nothing by recommending this, I'm an independent consultant, just giving my 2 cents.

cheers  ;)

[Grim]

Hey TJ, I skimmed your OP and those issues are all ones that I have seen 8.4.1 and 10.1.1. All of which were issues on IBM's part with the code. To resolve them either we needed to apply a Hotsite, FP, Advanced param or in one case a newer Java version. Just as we got the 8.4.1 under control and stable it's support has ended thus the move to 10.1.1. This was all mostly because the client was on Solaris and IBM seems to be notorious for NOT properly testing thier products on Solaris (Not just Cognos).

Definitely issues that something is not right and should be tracked and traced and logged with IBM.

Can you provide the Environment details?
OS
CPU (SPARC, X86..etc)
RAM
Cognos Single server or Distributed Install
Cognos Version and FP level and any hotsites? even better if you can attach the cmplst.txt
Java Version
Gateway Vendor and Version
App server if not Tomcat
DB vendor and version (Both Content and reporting)

If I have the above I might be able to provide a bit more insight.