We currently have the Laserfiche and Workflow servers on the same VM. We are looking at possibly moving Workflow to its own server as we start to publish more and more workflows.
What is the recommended process for making that transition?
We currently have the Laserfiche and Workflow servers on the same VM. We are looking at possibly moving Workflow to its own server as we start to publish more and more workflows.
What is the recommended process for making that transition?
The move itself is not complicated, but making sure no events are lost is a bit involved.
So I would check Message Queuing (Control Panel, Administrative Tools, Services, Message Queuing, Private Messages). If all the "lfwf" queues show zero messages, then you don't have any current activity. If any of them show messages, I'd stop the Subscriber so it stops processing new events from Laserfiche and wait for the queues to empty out as the WF Server picks up the messages and starts new workflows.
Then, to make sure the new subscriber picks up where it left off on the old machine, I'd export HKLM\Software\Laserfiche\Workflow\monitored Repositories from the registry.
Thanks for your help
So glad I asked this question 4 years ago. Just referenced it again!
Miruna
I just had a client contact me that I moved their Workflow server to its own server by following your steps above, but they are having issues with published Business Processes. Here is the error the end user is seeing from the Client:
If we go into Workflow and try to republish that Business Process, we receive the following message:
If they say yes it fixes the previous mentioned error for the end user. Is there another step that we should have done to prevent this from happening?
Good point. Yes, you do need to republish business processes.
Or you can create a DNS alias for the new server to the old one's name. (I've seen customers who always DNS alias the servers specifically so they can be replaced without disturbing whatever applications are using them. All users know is they go to "lfs" for the repository and "lfwf" for Workflow. The actual machine underneath can be replaced as needed and its actual name is irrelevant)
Good to know. Can a note be added to the help file that you created to include that please?
will do.
Much appreciated.
On step 6 of the help file it says to export a registry key. I noticed in the registry key it has values for the server. If the repository server is also being moved, do we need to edit that value? Are there any other values that would need to be edited as well?
You can, but that will only reconnect the subscriber. The Workflow Server needs redirects as well. It's best if you put in redirects, detach the repository from WF, turns off Workflow services, move LFS and add the repository back in to Workflow and confirm it works before moving WF.
Even if you choose to move Workflow and LFS at the same time, you should still add the redirects first.
Best practice would be to use DNS aliases for your servers so that the machines underneath can be changed as needed without requiring reconfiguration of any services that connect to them.
Can the details of checking the Message Queuing also be added to the help file please?
Updating the Workflow Web Services URLs should probably also be added.