This message was deleted.
# citrix-vad
s
This message was deleted.
l
Which hypervisor are you running on?
r
I see this all the time with SCCM. Have you tried clearing of the C:\Windows\SoftwareDistribution, then restarting bits, and windows update service? I know it sound very basic. Chances are you have.
What version of windows 10?
Another thing I did to fix this was running BISF, on the presistent machine to optimzie the WinSxs Folder. Sounds goofy i know. but it helped me
a
Nutanix and windows 10 1909. So it happened to me and one other person yesterday again but time didn't coincide with the wsus check in., but both times when we force restarted it did have updates to install
Hmm I think I tried Bisf earlier in the build but will definitely revisit.... So since corruption in winsxs maybe?
l
Whats version of CVAD are you using?
a
1912 cu1
l
Have you redirected your event logs to the MCS cache disk?
a
No I haven't... These are fully persistent so didn't really need to and there isn't one for this setup I don't believe
l
Very true, apologies, missed that they are persistent. Guy Leech has a script that will pull out all events in the event log between a certain point in time. It should be able to give you what you need.
a
It's not actually crashing or blue screening.... Just super high cpu. I watched it in the AHV graph for about 20 min and it just fluctuated from 92-100 cpu and only thing I could do was ping it ( can't log into the console, no control up connection, reboot services, c$, nothing) so force restart is the only option
Oh not a problem! I appreciate the help!
I did look at the app and system log after and didn't show much
Nice I'll see if I can find his script to takea look! Thank you! Will that be able to give an indication what the cpu is spiking from ya think?
interesting......i locked up again and let it sit (remoted in w/ a different machine to monitor and about 25 min it came back and cpu dropped
l
Show all events from every event log (300+!) in given time range & then filter/search: Get-WinEvent -ListLog * |?{ $_.RecordCount }|%{ Get-WinEvent -ea SilentlyContinue -FilterH @{logname=$_.logname;starttime='162915';endtime='163115'}}|select *|sort TimeCreated|Out-GridView https://t.co/Nim2TmCyDt
a
maybe found something and possibly not wsus related . once it came back digging in the logs the security log had a new process started right when cpu went up for Office 365's sdxhelper.exe ......i saw some reports of that causing high cpu and theres a scheduled task that kicked off right then too . just not sure what to do w/ that now!
l
If you run this between the start time and end time you'll get everything.
a
very nice! thanks man! i just posted above something i found that could be a lead...
l
Whats the trigger on the scheduled task?
So..... id probably change this to a specific time and see if you notice any effect. It is supposed to securely manage office update downloads apparently. If that's the case it may just be a one off thing. Its a bit rubbish that you can't control the resource usage though.
a
I did see an event a few seconds after saying that process excited so not totally sure if that's it yet
Unless that just kicks something bigger off
l
Its possible, you really need to be able to monitor the processes. Do you have any tools that do that? ControlUp? SysTrack? UberAgent?
a
We do have control up but the console is losing contact with the desktop when the cpu spikes and can't connect again until it drops
l
Its going to be tricky to find. I'd work on reproducing the issue. What happens if you kick off that task manually?
a
thats what i was curious to try too when i have some time this weekend to monitor on my laptop. So before i had to ramp up another 100 or so new desktops tonight i updated my image by first fully uninstalling/reinstalling 365 (found some suggestions for that sdxhelper high cpu w/ that sometimes helping) and ran bisf w/ the winsxs optimizations from @Ray Davis suggestion from his experience so we'll see how it goes! one test machine from this new update i manually kicked off that scheduled task and all seems ok. fingers crossed it holds! i also ran that log collection script from todays lockup so very curious to dig in to that . thanks for your help guys!
alright sorry to bombard here but one more update after looking at logs i saw this.....what i'm not sure is if this is a smoking gun pointing at the sdxhelper.exe process or if that line speaks to our WEM CPU spike protection (which would bring the question either way of why the hell didnt that feature help when this spiked the CPU for 20 min??).
l
Depends on your WEM settings. WEM doesn't actually stop something processing. All it does is moved the process down the priority list for a period of time. Also, depends how long you've told it to hold a process at a lower priority. If your system is not busy anyway a lower priority process can still eat CPU.
r
Sorry man, didn't mean to drop off. Got tied up.
👍 2
a
that does make sense.......i can't imagine having that WEM CPU settings is contributing to this issue right? if anything maybe helped add a useful log entry to confirm the lead i have here (if i'm reading that right it does seem to indicate that that sxdhelper IS the offending process wouldnt ya think? side note i'm having it happen again right now so gonna let it sit and see if things match up
l
Looks like that could be it yes. Set your control up agent to highest cpu priority to see if that will keep it ticking over through this period.
a
good call i'll give that a try!
latest lockup lasted about 2 1/2 hrs this time but did recovery. log collection shows that same sdxhelper.exe event from WEM trying to optimize it and task scheduler shows that task kicking off the exact minute things went south again. i'm going to try an uninstall/reinstall script for 356 to see if it helps first since i can easily do that for everyone if i have to and next person in my group maybe i try the controlup agent prioritize to be sure too. i wondered if i could try the cpu clamping w/ WEM for just that .exe too. more of a band aid though