This message was deleted.
# citrix-vad
s
This message was deleted.
f
Storage vMotion is not supported, I’m not aware of any standard DRS limitation. Does support refer to any knowledge base article ?
r
They didnt give me a article I asked for that. Here is what they said.
Siddhi (5/1/2020, 12901 PM): we don't have any support article for vmotion there are articles related to vmware Hence an engineer will need to check if issue related to lag is caused due to ICA session or vmotion Me (5/1/2020, 13003 PM): ok Siddhi (5/1/2020, 13240 PM): As per one of the answer related to a similar query We don’t support DRS yet, VMware just provided an API, once we implement that in MCS, I think vMotion will be possible (pending test). Siddhi (5/1/2020, 13307 PM): That means there are some pending test after which we will be able to comment if it will be supported Siddhi (5/1/2020, 13337 PM): But I can raise a voice support ticket for you so that an engineer can check over a remote session Me (5/1/2020, 13340 PM): Is that from a case or data you can send to support that evindence? Siddhi (5/1/2020, 13430 PM): Yes its from a case Siddhi (5/1/2020, 13517 PM): There is no ETA on this available Me (5/1/2020, 13523 PM): Ok. I understand.
w
For any vdi product regardless of the vendor, I never recommend using drs
r
Interesting Carl. Vmotion could be what causing my lags at times. WHen I dont see any ICA issues. @François-Xavier Rigaud yes, that is what I thought as well. But it may seem to come down to each eviroment? IDK.
So you guess are saying No DRS and definitly no Storage DRS (Which I turned off)
@Leee Jeffries What is your experince man?
w
I have seen places where fully automatic drs is used for vdi and vCenter does nothing but move VMs between hosts. Turn drs to partially automatic and almost all issues go away. Partial drs will place VMs on the lesser loaded hosts on power on.
Yes @Ray Davis , I am saying for clusters dedicated to vdi to use partially automated drs
r
yes sir. Thank you I will try this.
w
If a cluster is general purpose use, I recommend partial drs until it is time to use update manager, then I turn in fully automatic drs. But turn it down a notch once all updates are installed and all hosts are out of maintenance mode
But I would hope all your vdi VMs are off when doing cluster maintenance
r
We do DRS for Cluster resources during the day. MSC Full Clones only
PVS as well.
I do value all opiniosn and I will try to correct this. Thanks.
j
Well let’s not say DRS is bad. For placement/power on/partial not an issue. Now fully automatic is entire different conversation.
r
DRS is great, I am just trying to figure out if its ideal to allow fully automated DRS on CVAD during the day. From what I am finding online, its hit and miss. Storage DRS is a no no based on my research and repsonces here. But normal DRS seems to be based on experinces.
But as you can see Citrix support said no on DRS for MCS. I guess VMware is making a API for this to be 100%? I have no idea.
But this is Full Clones, Not linked clones
j
Full clones you should be able to Storage DRS.
Moving MCS VMs around VM/compute wise between hosts shouldn’t be an issue either. Storage wise no.
r
We do it now. But I have a hunch its causing the machine to lag some
l
Full clones is the same as normal vms. Lol.
r
Yea I got it. I am just saying the user is working in a session. then DRS kicks on, I wonder how that effects the ICA and network latency
l
As it switches hosts it's possible. Really depends on your network set up to be honest. But even then it should be so minimal.
⤴️ 1
r
Yea makes sense. I am sharing my Citrix VAD on the same load as SQL and all datacenter workloads. So that could be another possible issues.
I am looking to seperate it out on Nutanix,
l
That is definitely not ideal and if say that's more of an issue. Nutanix will sort you out.
l
Im probably the lone Xenserver user here, but we have WLB enabled on VDI Pools as well. It helped a ton since most VMs are on and CVAD can't choose a VM based on host.
r
I always advise that if you use DRS don’t be aggressive with the configuration as it will cause unnecessary vmotions across hosts which equate to more network and cpu processing which ultimately hinders performance, also one of the steps of vmotion is a quick pause of the vm to complete transfer so it can be perceived by the user if doing something at the time.