Hi,

We are currently running a new 2019 farm, after migrating from SP2013. We've noticed that workflows are not starting as we would expect. We believe that the issue is described in https://help.nintex.com/en-US/nintex2019/current/#sp2019/InstallGuide/EnsureSvcsRunning.htm?TocPath=Get%2520started%257CInstall%252C%2520upgrade%252C%2520and%2520migrate%257CInstall%2520Nintex%2520for%2520SharePoint%257CInstall%2520Nintex%2520for%2520SharePoint%257CPlan%2520your%2520test%2520and%2520production%2520environments%257CSet%2520up%2520topology%2520for%2520installation%257C_____1 (the first note). The linked article was removed by Microsoft, so we are no longer able to find the instructions.

With some searching, we believe to have found an article that lists the steps, https://blogs.msmvps.com/haarongonzalez/a-quick-tale-from-the-sharepoint-support-trenches/. Since we are running WFE+cache and app server minroles, we tried method 3 in the article, yet this completely goes against the spec for the services per min role (https://docs.microsoft.com/en-us/sharepoint/administration/description-of-minrole-and-associated-services-in-sharepoint-server-2016). Instead of running the Foundation Workflow Timer Service on the app server, the WFE's are running it instead, and the service has to be disabled on the App servers. This shows the servers as no longer compliant.
After some more searching we bumped into this post: https://community.nintex.com/t5/Nintex-for-SharePoint-Forum/Workflows-and-the-SharePoint-services-required-to-run-them/td-p/84052, that tells us that the configuration is actually no longer required, and the issue should have been solved?

Main question: What is the current best practice for the Foundation Workflow Timer Service configuration in SP2019-farm, seeing that the old config goes against the minrole spec?