If you are unable to create a new account, please email support@bspsoftware.com

 

News:

MetaManager - Administrative Tools for IBM Cognos
Pricing starting at $2,100
Download Now    Learn More

Main Menu

Reconciliation Cancellation at 30 seconds per item

Started by jv_oz, 20 Nov 2008 06:32:20 AM

Previous topic - Next topic

jv_oz

Hello,

I have an app that is only 100k cells per elist slice, but we have 6,000 e-list items (don't ask why).  We have two 8 CPU boxes with 8Gb Ram and Win Ent 2003.  I am running a GTP (which worked) and subsequent Reconcile.  The reconcile keeps failing at 30 seconds, while other e-list items successfully reconcile faster than that.  I am gradually making my way towards zero items to go by manually restarting the reconcile, but would like to know of any time out settings that may influence the failure?

Thanks


sascha


StuartS

Not sure if this helps, but, with 8.1 there is a setting in the cognos configuration for the max size of an elist.  The default is 3000.

arun08

I think you can try the epreconciliationprogress.exe and repair the reconiciliation. You will find it in the c8/bin folder. You may also try epdebugtoolviewer.exe and choose the detailed debug option. Remember to turn it off when you are done.
But Fisherman is dead right when he says the default elst size is 3000. Also it depends upon how many aggregation levels you have in the e-list if I have not mistaken.

Arun

jv_oz

Thanks for your posts.  I eventually overcame the issue by restoring from backup and reimplementing my changes.  Cheers, jv_oz

ExpertPlanner

I've seen several applications exhibit this behavior.  Manually forcing a reconcile or using epreconcile allows the completion of several elist items before failing again.  Usually, after successfully getting through all elist items, the problem returns on the next gtp. 

I once saw this problem manifest following an application database restore.  In that case, stopping the Cognos Planning service and restarting it resolved the issue (on the next reconcile all items completed successfully and the problem did not return.

In all other cases I've ended up going to backup.

craig_karr

you might also want to consider adding more RAM, alternatively not using all 8 cpu in the job server settings. each cpu might use up to 2 GB RAM so with 8 GB RAM to 8 CPUs you might run out of memory