We have a customer who heavily utilizes Laserfiche (200+ users working pretty heavily 8am-4pm daily). Their system consists of 4 separate servers:
- LF Service only
- Workflow Server & Subscriber
- Forms & Web Access
- Dedicated SQL Server (Serves LF, WF, and Forms)
We have recently noticed some odd Workflow Starting Rule evalutations, and I suspect it is due to latency (either from the network, or due to high usage of the Laserfiche system). Here is the scenario:
We have a Forms process that saves the form into the repository. It saves the form into a new folder, and applies a template/metadata to that folder in the process. Workflow has a starting rule that looks for this folder (a "Create" rule, that checks for the proper template to be applied), and begins a routing/approval process. Recently we have been noticing that sometimes the Workflow fails to start when the folder is created. What I am noticing in the Subscriber log, is that is seems like when Forms creates the folder, it is actually a pair of events (a Create event, followed by a Change event that applies the template). Normally, they happen so close together, that by the time the Create starting rule is evaluated, the Change event has also occurred, and thus the folder has the correct template and triggers the WF. It seems that in some cases, the LF system is under enough load that the Create event starting rule is evaluating before the template is applied. Obviously, we will have to make some changes to how the workflow triggers to work around this.
A. Can someone confirm that this is how LF works when creating a document/folder and applying metadata (is it always 2 separate events like this)?
B. Are there any best practices to design your starting rules when you expect a heavily used system?