This message was deleted.

I was always on the daily bus unless there was a damn good reason not too - the more frequent, the more stable

I would be on the more frequent bus, but management doesn't like the idea of users not being able to connect at all times, day or night. We have a weekly maintenance window, but aren't allowed to reboot weekly, only can use the maintenance window if there is actually something to change that requires are reboot. Really we end up rebooting about once a month unless a server has issues. If we were to restart more often, the servers would probably have less issues in the middle of the day...

Daily

I recommend nightly where possible to my clients, but most go with a Sunday on/after 12 AM schedule

We do every other day. 1/2 one day and 1/2 the other. Leaving it up for users to connect at all times. Also drain users for 8 hours before reboot. On avg we end up kicking 10s of users out of 1000s that connect

As often as you can

Each week (24/7 environment)

Daily.

Never

I’m sticking with daily, preventative goodness with no downsides if scheduled properly 🤷‍♂️ to each their own though, whatever works really

Daily. I've setup 2 groups of servers. One for daytime and one for nighttime using tags. I have scheduled tasks that take the daytime ones out of production so they can rebooted and the nighttime ones are made available overnight. Then around 5 or 6 the maintenance toggles and the daytime ones become available and the nighttime ones go out of production and are booted later in the morning. I've got a PowerShell module to do the maintenance toggle and I "borrowed" @Matthias Schlimm’s invoke-RebootXDMaintDevices.ps1 script to handle the user logoff and reboot.

My question to the team is, "Do you find that you NEED to reboot frequently or has it just become a BEST PRACTICE? I remember, back in the day, that we HAD to reboot frequently because apps had memory leaks and wouldn't release resources when terminated. And this was exacerbated in an multi-user system since many people ran those processes. Do you still find that that is the case or are apps behaving better in the 2020s?

My point exactly. This is why I've ended up with my first ever deployment with no reboot scheduled (admittedly we're not doing "tough" applications). Reboot schedule was preventative maintenance that seems to have become mantra

The main reason behind why we do it is keeping the PVS writecache clear. We have a couple of apps that can fill it up like WebEx or Zoom installs and OneNote. I'm going to test moving at least OneNote to a container to see how well that works. I'm not sure you can prevent WebEx always installing to the user's profile when it launches.

Fair enough when cache is in use. We don't do PVS or MCS

Non provisioned is a different story…

I think there's an increasing tendency to try and avoid "preventative" steps like this if IT teams are being held to external SLAs and audit. We have customers who run synthetic availability testing on services they use from MSPs / or silo'd different teams in their own organizations... for teams running operations where their end users or customers hold them to account for downtime / users kicked off etc... unnecessary maintenance time takes away from the time they have to resolve real issues and still meet SLAs and they can get questions why was big-bob-working-on-his-super-important-powerpoint kicked off at 2am.... because the systems a bit flaky and we turn it off and on because we don't really know why conversations etc... when there's a clear service / user delimitation the customer is more likely to expect always up...

Every 2 days with W16, 49 GB WC PVS

for those that don’t reboot nightly, when are you doing FSLogix compaction?

100% in my experience it’s been warranted across many verticals, particularly health, else you look at a farm and admins are constantly fighting fires with maintenance mode on dead hosts which typically don’t have problems if you bounce them. MCS and PVS

@Mike Streetz (O_P) didn't know you needed to do a reboot to compact? We run it on a schedule, if anything is in use it is ignored

I wonder if @JimMoyle has any official/jim best practice advice around FSLogix's needs?

if you don’t have session limits and you don’t reboot, when is it not in use? I’m running into this with a client now

You gotta get all Mike Streetz on them users!

not my monkeys

weekly, the same for compacting VHDs.

Don't forget to enable natural reboot, only available through Powershell for some unknown reason. https://developer-docs.citrix.com/projects/citrix-virtual-apps-desktops-sdk/en/latest/Broker/New-BrokerRebootScheduleV2/ If someone at Citrix is listening this must be added to the GUI.