Getting CM-REQ-4292

[sdf]

Warning CM-REQ-4292 Content Manager is unable to process the request because it is running in standby mode.

This issue pops up from time to time, a quick remedy is to restart the servers. But that is not acceptable by any means.
I am just wondering if the setup of the primary/standby Content manager has something to do with it.
I checked the current configuration :

In Report server1(CM1 and Disp1) the primary/standby CM is as :
CM1
CM2

While in Report server2 (CM2 and Disp2) :
CM2
CM1

I read somewhere that best practice is to have the order of the CM configuration the same across the servers. That means, Report servers 1 and 2 must have CM1 as primary and CM2 as standby.
But didn't find any supporting doc or details as to why is that so.

let me know if you have any experience with this issue or any info regarding the order of the content manager to shed some light in one of the dark paths of cognos administration.


Cheers!
sdf

[Invisi]

I would start indeed by making the order of the servers identical. It's possible that you get your errors when there is a request going to the server where the active Content Manager server is not first on the list. I love to hear if the errors stop when you make that change.

[sdf]

The task to set the CMs properly is already requested. This is just a sub issue that is happening from time to time.
But yes, I will monitor or even test this and will update this post.

Just to share, still on investigation, there are long running reports (scheduled) which causes the CM1 to fail.
Then fail over setup is not reliable as expected. Thus the issue above.

Cheers!

[Invisi]

You say long running reports? Will you specify that? Are you talking about reports that are viewed interactively in HTML and have multiple pages? And does the error occur when browsing to a new page?

[sdf]

mostly failed batchreport executions.

[Invisi]

That rules out my first idea. How is the memory usage of the Cognos machine(s)?

[sdf]

yeah, from time to time we notice spikes on the java memory, recently an unusual spike but we already are in the talks in increasing it in the setup.

what I'm looking at is that some of the long running reports are actually failed run per se. But the bibus incorporated with this runs are still eating up memory. Then other report runs and schedule run will share the rest of it.