In WMF 4 the refresh and configuration check-ins were controlled by a scheduled task - I hadn’t before but I assumed these could be managed and changed from there.
On WMF5 how can you control the time the nodes check in? The scheduled tasks no longer exist and there seems to be no way to define in this in the LCM. Any ideas?
I’ve tried resetting the LCM over a staggered period, but it doesn’t appear to have an effect.
Reason for asking is we have a large number of VM’s on the same cluster all currently checking in at the same time, causing IO problems due to the CmdletCaching.
You can’t schedule check-in times, and never could; it’s always interval-based (every 15m by default). The most you could perhaps do - and I’ve never tried this - is manually trigger a set of machines, and then a few minutes later another set, and so on, so that if they’re running the same interval then they’ll remain somewhat offset. I mean, honestly, what you’re running into is a problem in your underlying infrastructure. In the current implementation, the LCM doesn’t provide a lot of workarounds for that.
You can control it via LCM properties ‘ConfigurationModeFrequencyMins’ (default value is 15) and ‘RefreshFrequencyMins’ (default value is 30).
Unlike in WMF 4.0, these two values are not needed to be multiple of each other in WMF 5.0.