How often should you reboot the kinetic servers and in what order? Epicor Support said we should only need to reboot the sql server but i wanted to get other opinons.
I’ll be interested to see other responses, but in my experience we don’t need to restart the actual Operating System often at all. Basically only when installing Windows Updates.
In a case like that, SQL Server should be restarted/started up before Epicor. Epicor needs the SQL Server to function, that’s where all of its data lives.
In past versions (10.1), we found that if the SQL Server restarted, Epicor would become angry and the Application Server and/or Task Agent would need to be restarted. In more recent versions (10.2 and newer), we haven’t experienced that problem.
If the Epicor Application server needs to be ‘restarted’, it is usually a matter of Recycling the Application Pool, not restarting the actual Windows server.
My opinion is that the server’s OS needs to be restarted at a minimum of bi-weekly, and that’s NOT accounting for patches and security updates. This really is more of a mistrusting Microsoft thing, than an Epicor thing. But I’ve found that under whatever circumstances they are, the appservers will consume way more memory/processor than they need and do not give it back, leading to an ‘overload’ situation where things get ugly. I’ve not dug into it, but it requires me to reboot it (or manually restart the "worker’ threads) so I might as well bounce the whole box. SQL doesn’t exhibit this behavior, but I reboot it at the same time.
I’ve turned the whole process into my bi-weekly MS Update + Reboot + VM Snapshot process on Sunday’ mornings. Only take about 15 minutes regularly (with no major MS updates) but it lets me sleep easier.
Is there any particular order that you are rebooting the servers?
We reboot once a month after applying updates. SQL first, then app servers.
I stop the Task Agent, appservers and SQL Services in that order. Then I apply MS updates and occasionally do some other hoousekeeping. Then I reboot SQL, then reboot the appservers (2 for ERP, 1 for ECM).