How to schedule K2 workflows that run periodically

Hi Tejas
 
You can schedule a workflow to run from a K2 Process Portal. 
 
Provided you have SharePoint and have insalled the K2 for SharePoint components you can schedule a process from one of these portals. If you are not using SharePoint/Process Portals then I dont think there is an OOB scheduler. You may have to write a custom service/app to perform this.

I cant seem to insert the help files links here but go to help.k2.com and search on "Schedule Process" and look at the first couple of results.

What you can also do is loop the workflow every two hours. Basically once the workflow has completed the last activity, draw a line from that event t the start activity. then set the start activity start dealy to tow hours. Jsut make sure you put in a condition so that the first time round the 2 hour delay is not applied. THis is all in the workflow s you do not need any of the shareoint stuff. You just need access tot he workspace to start the process.

Hope that helps

I have done the looping thing, but with a parent process and a sub-process.

This way, the "scheduler process" runs and loops on itself (with the 2 hour delay on the start rule), but the sub-processes run independently. There will be a new instance every 2 hours.

Dear s0m3one,

While it may be cleaner in the database to have a single process instance that keeps looping, it is definitely not considered best practice. dgilmour's method of using IPC processes is probably a much better idea. For one thing, if you just keep looping the same instance over and over, the "state" of the instance will continue to grow, which can lead to significant performance issues.

Another approach to the problem could be to write an application using the K2 APIs, that programatically starts a workflow. This application could then be scheduled with the Windows scheduler.

Best regards,

Gail

Dear s0m3one,

The "state" of a workflow is persisted in the database. I don't know that there is any real documentation out there, but think of it this way:

Each workflow instance has a record in the Server.ProcInst table and the ServerLog.ProcInst table. The STATE column in the database (Server schema) is a binary represenation of the current state of the workflow. It is kind of like a list of every item the process has executed. The more loops a workflow goes through, the larger that STATE field grows. Since it is a binary representation, then every time the server wants to interact with that instance, it has to deserialize that field, and then reserialize it later to write it back to the DB. As such, the larger that STATE grows, the worse the performance of the K2 server can become.

Most of the time, a process instance's STATE will be less than 1 MB.

I hope this helps explain my earlier comments.

Best regards,

Gail

Hi,

How to do all these using K2 scheduling APIs?

Regards,

Asad Naeem