This message was deleted.
# _general
s
This message was deleted.
j
16GB Pagefile? How did you came up with that value? Seems a little bit to much for me How often do you reboot the RDSH?
m
cache to RAM. Give it enough RAM and you won’t cache to disk much at all.
v
vdisk RAM sive is 4 GB, workers running on 32 GB RAM and it does not look like they are struggling for RAM. We are running weekly reboot schedule as we are 24/7 and we have night shifts. So we split workers into groups, put in maint mode, drain, then reboot on Sat, Sun. No special optimization/sealing script. (looked into BISF couple of years ago but didn't implement). 16 GB page might be overkill, agree. I'll have to re-read this article.
m
do you defrag your disks?
if you have enough ram I’d try first defragging and then doubling the RAM cache. With a big enough RAM cache your iops should go down to almost nothing
v
in OS ? or mapping VHDx and defrag it?
m
map it and defrag it, it’ll be quicker
because of the difference in block sizes between the OS and the RAM cache, if you have fragmentation you end up with lots of underutilized 2mb blocks in RAM
if you really want to get into it, you can run performance analyzer against it. https://docs.citrix.com/en-us/advanced-concepts/implementation-guides/digging-into-pvs-with-poolmon-and-wpa.html
but the short of it is that if you have 4GB of cache and it’s still writing to disk, you either have serious fragmentation or users are creating huge files
or loads of individual tiny files in your case
v
looks like lot of individual files, mostly Chrome, at least that is what I see when cleaning profiles. 300-600 +MB a day.
per user
m
are you running ublock or anything like that in chrome?
v
no, no ublock or any other ad blocker. All extensions blocked, around a dozen are whitelisted
m
ublock or a proxy for vdi that stops all the ad tracking junk getting downloaded can help
v
thank you for suggestions, we wills tart with disk defrag and adding more RAM cache.
a
Your RAM with overflow to disk can easily go up to 12 GB, this way you eliminate IOPS on D drive write cache file. Review a series of blogs called PVS Internals for more info. System page file doesn’t need to be 16gb. have a look at the committed memory and adjust the settings. Also, you may use my procmon powershell tool to have a better understanding of what processes are writing to write cache file on affectes servers https://ninaronline.co.uk/2018/07/16/automated-process-monitor-captures-remote-machines-using-powershell/
v
Thank you All. After couple of weeks I could confirm that defragmenting vdisk and changing RAM cache from 4 GB to 8GB changed things drastically. All except two from 120 + servers have write cache size on disk =0.
👍 1