Hi...
I wish to run a stateless process as repository owner even the process initiated by any users.
Note: For the time being, from state-full process i am calling this process with run as repository owner option.
If there are Accepted Answers, those will be shown by default. You can switch to 'All Replies' by selecting the tab below.
If it needs to be synchronous/stateless, and you can obtain the requested behavior by putting the user in a group (i.e.: it's a lack of sufficient permit that requires you to run as repository owner), you can use the Privileged DFC (and privileged module) logic to dynamically include the process user in a (dynamic) group, so he/she can execute the required action.
More specifically, this would require you to register your specific DFC as privileged (done in DA) and write a Java service and make it a privileged module (allowing it to perform actions by dynamically putting the executing user in a group with correct permits).
Another option is to write a server method. You can configure server method to "run as repository owner"
Can we use the service with privileged code and run in stateless process? Wont it have impact on the stateless process as the whole stateless process will be running under single transaction?
If you are talking about privileged module, I don't get the remark, as indeed everything runs in a single transaction for the same user. You do stay in the same session, dynamically adding the user in a group to allow doing certain actions. Everything is indeed one transaction, which is also the purpose of a stateless process.
Hi Johnny, I was wondering about server methods, as this is typically what I did before discovering the concept of privileged DFC. From a security perspective, I always considered it dangerous, as you would at least need a mechanism (using login ticket I guess) to validate the credentials and permissions of the user, before even allowing an action as repository owner. What do you think? Any opinion on the pros and cons of each approach?
My con for methods was (at least a feel of) them being unsecure. That's why I moved to privileged DFC.
Hi Gino,
Your statements are correct in that server method is less secure; however, it also simple way to work around the stringent Documentum security model. The privilege DFC client is design to address this; however, even getting this to work with EMC products (eg RPS) is still a hassle to getting it work without any issues. Using server methods is straight-forward.
-Johnny
Absolutely agreed . Took me a little time to have it working. Ones it works, it's pretty powerful though. I'll write up what I found and post it later on, to have a full "how to create privileged Java Services from xCP" available to at least simplify the process of making a privileged client.
Thanks for your feedback!
Cheers.
