Thanks, Wei! On Thu, 5 Dec 2024 at 18:01, Wei ZHOU <ustcweiz...@gmail.com> wrote:
> Hi, > > You can refer to > > https://docs.cloudstack.apache.org/en/latest/adminguide/virtual_machines.html#advanced-instance-settings > > > -Wei > > On Thu, Dec 5, 2024 at 7:41 AM Muhammad Hanis Irfan Mohd Zaid < > hanisirfan.w...@gmail.com> wrote: > > > Ah, so that will edit the XML, right? Did you know anywhere in the docs > > that the VM settings are mentioned? > > > > On Thu, 5 Dec 2024 at 10:06, Nischal P <nischalnisc...@gmail.com> wrote: > > > > > Hi Muhammad > > > > > > > > > You can stop the VM and in the settings you can use "nicAdapter" > And > > > value "virtio" > > > Start the VM and check, if the drivers are present it should have used > > > virtio in the Guest OS > > > > > > > > > > > > Thanks > > > Nischal > > > > > > > > > On Thu, 5 Dec, 2024, 4:21 am Muhammad Hanis Irfan Mohd Zaid, < > > > hanisirfan.w...@gmail.com> wrote: > > > > > > > Okay I can edit the VM XML to use virtio NIC model instead. > > > > > > > > > > > > > > https://askubuntu.com/questions/552036/where-is-the-list-of-supported-device-emulations-in-kvm > > > > > > > > But, is there an option on the UI/API to select NIC model > during/after > > VM > > > > creation? If not, maybe a new feature that can be added in future > > > versions. > > > > My suggestion since CloudStack UI looks more like a public cloud use, > > to > > > > allow changing the model only by Administrators. > > > > > > > > On Wed, 4 Dec 2024, 12:36 Muhammad Hanis Irfan Mohd Zaid, < > > > > hanisirfan.w...@gmail.com> wrote: > > > > > > > > > I have a VM running Windows Server 2022 on a KVM hypervisor. > > Currently, > > > > > it's using Intel PRO/1000 MT as the network adapter. I would like > for > > > it > > > > to > > > > > use VirtIO NIC as I've already installed the necessary drivers and > > QEMU > > > > > agent. How can I do it? Based on this ( > > > > > > > > > > > > > > > https://users.cloudstack.apache.narkive.com/wDIB00YQ/how-do-you-customize-cs-vms-on-kvm > > > > > ): > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > *Normally (at least with KVM):- The base disk (OS) will be IDE- All > > > extra > > > > > disks will be VirtIO- Networking depends on the 'OS Type' (not sure > > > where > > > > > this is defined atthe backend)- To use VirtIO Networking, select > the > > OS > > > > > Type of 'Other PV (xx-bit)'* > > > > > > > > > > Any way to change the defaults? > > > > > > > > > > Another question is regarding Ceph RDB performance for primary > > storage. > > > > > We've a virtualized Ceph cluster running on vSphere (I know is not > > the > > > > best > > > > > setup). The underlying ESXi is running at least 10G NICs. The Ceph > > VMs > > > > are > > > > > using VMXNET3 for its NIC, and SCSI controller for the disks. > vSphere > > > > > datastore is using iSCSI SAN from Dell with, maybe as I remembered, > > > > 7200RPM > > > > > HDDs. 5 Ceph nodes, 2 OSD (1 TiB) each. I'm getting around 11 MiB > > > Client > > > > > Throughput at most on Ceph. I felt quite significant slow > performance > > > for > > > > > operations e.g. converting snapshot/volume to template, launching > new > > > VMs > > > > > etc. > > > > > > > > > > Any thoughts? > > > > > > > > > > > > > > >
