In a standalone environment TN writes the conversation info to repository (WmRepository2 folder which typically has a DirValuesHash.dat file and files starting with VH…). This conversation is on a lock while executing the invoke step in a conversation. Am i correct ? If so, Here is my doubt… In a clustered environment, which is the location of this repository where the lock of the conversation will be ?

Advance thanks

You don’t mention which version of TN you are using, but the information supplied in your question seems to indicated pre-wm6.01.

If this is the case, then there is a clustering guide that you need to read. It relates several different ways to setup the repository for clustering. Good, better, best methodologies… you are on the right track.

The approach you choose will depend on the mission critical nature of your environment and the amount of money that can be spent.

You are far better off using the database to store the repository rather than the file system. Once the files are corrupt, you are in trouble.

Read the IS Clustering guide next.

Hope this helps.

Ray

Thanks Ray…
Will be done with clustering guide shortly…
The interesting thing is i’ve two IS 4.6 servers who are mirror environments but loadbalancing functionality is managed by third party s/w and NOT wm way of clustering is done. Ofcourse TN DB is the same for both IS/TN. Bcoz of this, TN conversation details are stored in their respective IS repositories (boxes) which makes conversations corrupted since one TN doesnt have any clue of other TN’s repository. And the solution i am looking ahead is to have these two TN’s acessing only one TN Repository and not essentially doing any wm clustering here. To cut it short, with THE ABOVE SAID environment, Conversation shud be started thru one TN and continued thru another TN. Hope am not confusing…

Thanks
DG

DG

Let me know how your configuration works once it is set up. I will be setting up a similar environment for both my TN servers in the DMZ as well as my IS servers inside the firewall. I will be using 6.0.1 but other than that I will be doing the same concept of shared repositories.

Thanks
Joe

The above said configuration works fine…

am working on taking it further like multiple repositories, evaluation of jdbc comliant repository Vs file system, procedure for recycling repository servers… wud appreciate sharing any info on these lines