VM with yellow warning icon/symbol

[AmpNode]

https :// i.gyazo .com /72aea7ed6f6a53ae692489887fc46dc8 .png
I cannot find information on this issue anywhere yet.
The RAM, CPUs, etc is all way under max load and the VM loads up from a full shutdown for about 5-10 seconds and then stops with a lockup.
I have been trying to resolve this for a long time.
If anyone can offer input, I'd appreciate it.
There are no errors in the bottom panel to copy/paste here but if there is a directory I can navigate to in order to provide more information on this specific matter, please advise.

As a reference, this is what it looks like once it locks up:
https :// i.gyazo .com /9a362ca02f37f83864958558e659305c .png

Thanks

p.s. sorry for the spaces in the image links; proxmox forum won't let me post the links I guess

 

[wolfgang]

Hi,

the yellow icon comes if qemu reports an virtual HW error.
I personally have it only seen when MS Windows install new driver, like switch the GPU to spice.

The RAM, CPUs, etc is all way under max load and the VM loads up from a full shutdown for about 5-10 seconds and then stops with a lockup.

Can you provide a bit more information about this VM. OS programme what produce the lode, ...?

 

[AmpNode]

Hi,

the yellow icon comes if qemu reports an virtual HW error.
I personally have it only seen when MS Windows install new driver, like switch the GPU to spice.


Can you provide a bit more information about this VM. OS programme what produce the lode, ...?

OS is mixed between centos and windows server 2012. The two vms experiencing the icon during the time of opening the thread were windows server 2012 - no updates for drivers had taken place and the other 3 windows server 2012 vms were unaffected.

This seems to be caused by the main node experiencing limits on the hard disk.

 

[dcsapak]

if your vm in the tree has a yellow exclamation mark, you should see the reason under summary of the vm: status running (io-error) for example

 

[AmpNode]

Well, right now there isn't an issue since I freed up the node's disk space but it may have been an io-error.
What would that mean?

 

[dcsapak]

io-error means error with io (input/outpu)
this most likely is the disk, but could be anything io related where qemu could detect an error

 

[AmpNode]

io-error means error with io (input/outpu)
this most likely is the disk, but could be anything io related where qemu could detect an error

ok, that makes sense with the reduced main disk consumption resolving that error icon
it's just odd to me that a single vm would impact other vms when its disk becomes full

 

[dcsapak]

i guess that your vm disks are over provisioned and that simply happens when the underlying storage runs full

 

[AmpNode]

i guess that your vm disks are over provisioned and that simply happens when the underlying storage runs full

by over provisioned I am assuming you mean there's more disk space technically associated to the vms than there is available on the main disk?

 

[dcsapak]

by over provisioned I am assuming you mean there's more disk space technically associated to the vms than there is available on the main disk?

exactly

 

[AmpNode]

exactly

ok, thanks
i thought that once you assigned disk space to a vm it used up that space from the total disk but it sounds more like it is assigned but not used until it's ACTUALLY used. is that accurate?
So a vm assigned with 10GB of space doesn't take 10GB away from the main node until those 10GB are actually used up space on the VM...

 

[dcsapak]

i thought that once you assigned disk space to a vm it used up that space from the total disk but it sounds more like it is assigned but not used until it's ACTUALLY used. is that accurate?

no, that is only the case with non thin-provisioned storages
directory, lvm-thin, rbd, etc. are thin-provisioned

 

[AmpNode]

no, that is only the case with non thin-provisioned storages
directory, lvm-thin, rbd, etc. are thin-provisioned

once you create a vm, is the storage space assigned and "used" up from the total disk space?
or is the total disk space not consumed until the vm starts fill that space with files?

 

[dcsapak]

with thin-provisioned storage types (lvmthin, directory,...) the space is only consumed if it is allocated inside the vm
with thick-provirision storage types (LVM, ...) the space is consumed when the vm disk is created

 

[AmpNode]

with thin-provisioned storage types (lvmthin, directory,...) the space is only consumed if it is allocated inside the vm
with thick-provirision storage types (LVM, ...) the space is consumed when the vm disk is created

As I'm considerably unfamiliar with those terms/concepts, I assume I'm using "thick"
https://i.gyazo.com/9f0330a25b458e841f5df6b44542d44f.png

 

[dcsapak]

for some detail see https://en.wikipedia.org/wiki/Thin_provisioning

 

[AmpNode]

for some detail see https://en.wikipedia.org/wiki/Thin_provisioning

thanks for the input