Questions and answers about Compute Cloud

Written by

General questions
Virtual machines
Adding plugins
Disks, snapshots, and images
Disaster recovery
Instance groups
Monitoring
Licensing
General questions about licensing
- Windows Server licensing
- Licensing of other Microsoft products

General questions

What is cloud computing used for?

Yandex Cloud provides scalable computing capacity: you can quickly create and start VMs when necessary and stop them if the load drops. The usage of cloud resources reduces your IT infrastructure costs because you pay only for the resources you use.

It's easier to manage a cloud infrastructure so you can focus on your business challenges instead of server maintenance.

What can I do with Yandex Cloud VMs?

Use only those computing resources which are necessary to meet your needs. If you need expensive powerful systems, you don't have to buy them. You can create a VM in Yandex Cloud and use it whenever necessary.
Quickly scale processing capacity to meet your needs. You can start additional VMs during peak times and stop them when the load decreases.
Use your VMs to deploy applications that must always be available. You don't need to worry about guaranteeing server uptime: Yandex Cloud will keep it running smoothly. Focus on creating applications that work.
Configure backups to make it easier to restore your data in the event of loss.
Create and distribute VM disk images. You can use images to quickly deploy your software on other VMs.
Automate VM management using the API and scripts in the command-line interface.

For more information about Yandex Cloud VMs, see Virtual machines.

How do Yandex Cloud VMs differ from regular hosting?

Traditional hosting offers you resources on lease with pre-agreed performance for a fixed time. Yandex Cloud VMs let you use cloud resources as your personal data center. At the same time, you receive all the advantages of the resilient infrastructure of Yandex's data centers.

You can scale your cloud infrastructure as your performance requirements change. This way you can control how many resources you are using at any given time and how much you pay for them.

You will also have complete control over the VM status. You can start and stop VMs when you need to. You can set up your VM software configuration and change it as necessary. Disk images and snapshots let you easily transfer data between your VMs.

How do I get started with Yandex Cloud VMs?

You can create your first VM by following one of the scenarios described in Getting started with Compute Cloud.

How do I get access to a VM?

You can connect to a VM from another VM on the same cloud network. Use the VM's internal IP address or FQDN to do this.

If you assigned a VM a public IP during creation, you can use that IP to connect to the VM from the internet.

You can get the IP addresses, FQDNs, and other information in the management console in to the Network section on the VM's page.

For more information, see Network on a VM.

Use SSH to connect to a Linux VM. Use RDP to connect to a Windows VM.

How quickly can I change the performance of my information systems?

You can change the performance of your information system in one of the following ways:

Create VMs in advance with the appropriate characteristics and pre-installed software. During peak times, all you have to do is start these VMs to increase the performance of the information system. When the load decreases, you can stop some of the VMs to avoid paying for extra resources.
If you frequently need new VMs with the same configuration, you can create a boot disk image and use it when creating VMs.

Why is the VM running slowly?

Just like any other server, including physical ones, VM performance depends on the performance of:

The CPU.
RAM.
Disks.
The network.

Note that the physical core thread allocated to your VM handles, in addition to the vCPU itself, the qemu virtualization process and I/O operations (networks and disks).

General recommendations for improving VM performance:

For product solutions, use a vCPU with a guaranteed performance of at least 100%. If you use a vCPU with performance below 100%, then the vCPU performance may drop to the guaranteed level when "neighbors" appear on the physical core. For example, for a share of 5%, this is only about 100 MHz of clock frequency, which is most likely insufficient for a product server.

If the graphs show that the load on the vCPU is close to 100% for a long time or at the time the issue occurs, we recommend increasing the number of cores.
Never use the swap file. It causes a significant drop in performance. The best solution is simply to expand the vRAM if there isn't enough memory. With Linux, it's acceptable to use the zram-config utility.
Use large network SSDs. The weak point of the system may be network drives. They are subject to the limits of IOPS and bandwidth, which depend on the disk type and size. For more information, see Disk and file storage limits.

Keep in mind that disk performance also depends on the read and write request size. See Read and write operations.

You should read and write data in several threads.
Remember that network computing depends on the vCPU performance. Run synthetic tests that don't affect disk usage to diagnose network problems. If it does have to do with a slow network, we recommend adding a vCPU and retesting it.

What operating systems are supported by Yandex Cloud VMs?

Linux and Windows-based operating systems are supported.

Public boot disk images that have been tested in Yandex Cloud are available for popular distributions of these systems.

Does the service meet the requirements under Federal Law No. 152-FZ On personal data?

Yes, you can read the full security audit conclusion.

How do I contact technical support?

You can contact technical support under Support in the management console.

Can I get logs of my operations with services?

Yes, you can request log records about your resources from Yandex Cloud services. For more information, see Data requests.

How do I find out the cost?

You can find the cost of using the service in Pricing for Compute Cloud. You can also use our calculator.

Virtual machines

Technical specifications

What VM configuration (memory, vCPU) can I use?

When creating a VM, you select its vCPU performance level. This determines the required number and performance of cores (vCPUs). You can choose the computing resources that are appropriate for the expected load.

For more information, see vCPU performance levels.

How can I change the amount of RAM and the number of cores allocated to a VM?

For more information, see Changing VM computing resources.

How do I create a multi-interface VM?

You can only add a network interface when creating a VM. Currently, multiple interfaces can only be created using network images from the Marketplace.

VM parameters

How do I increase quotas

To increase quotas, submit a request on this page. Select the service (or services) and specify which quota values you need.

After creating the request, the message appears in the support section, where you can track its implementation. In some cases, it may be necessary to get approval from the service to ensure the availability of resources, and/or to increase the balance of your personal account.

Does nested virtualization work / Is it possible to host your own VMs inside a virtual machine

Nested virtualization isn't supported for security reasons. If you don't find a suitable public image for a VM, you can upload your own image and create a VM from it. See: Creating a VM from a custom image.

Is the data saved when I change a VM's parameters

Yes, the data should be saved. However, please note that it isn't recommended to restart the VM when writing to the system disk because this can lead to file system corruption.

To change a VM's parameters, we recommend taking the following steps:

Take a snapshot of the disk: see Creating a disk snapshot.
Turn off the VM as normal: see Stopping, starting, or restarting a VM.
Change the VM's parameters: see Changing VM computing resources.
Wait until the operation is completed (the Operations section in the Compute Cloud service).
Start the VM.
Make sure that the new parameters were applied.

Is it possible to reduce the vCPU/vRAM for a VM

You can. However, if the load on the vCPU and/or vRAM was already significant before changing the parameters, after lowering the parameters, the VM may not start.

If the load on the server is small and at the same time uniform, it's even recommended to reduce the parameters so as not to overpay.

What's the difference between GPU and vGPU

The main difference is that the GPU can be used for mathematical calculations and machine learning, and the vGPU can also be used for graphics tasks.

In addition, their cost, support of multiple processors, and licensing differ. For a detailed description and comparison of the GPU and vGPU, see GPU and vGPU graphics accelerators.

Is it possible to change the FQDN

The FQDN is issued to the VM at the time of its creation, and you cannot change it.

If you want to change the FQDN, you need to delete the VM and create a new one. To preserve the consistency of the data:

Stop the VM: see Stop.
Take a snapshot of the disk: see Creating a disk snapshot.
Create a new VM by selecting the snapshot as the source of the boot disk: see Create a VM with disks restored from snapshots. Specify the necessary FQDN during creation.

Can a VM's UUID change

Standard procedures for changing a VM, such as changing the number of vCPUs, the amount of vRAM, and adding disks, don't affect the UUID. However, in some cases, the UUID may change. For example, with updating the BIOS or similar procedures that affect the "hardware" part of the service.

Keep this in mind when using third-party software licenses. It's best if you can purchase a license that isn't tied to the UUID.

How do I change metadata after creating a VM

Use the API method updateMetadata.

This method allows you to update the metadata in the service, but without growing it into the guest operating system inside the VM. If you need to update a parameter inside a VM after it's created, in most cases, the best way is to recreate the VM from the snapshot or make changes manually.

Operations on VMs

Can I copy or clone an existing VM?

Yes, you can take snapshots of disks attached to a VM and use them when creating a new VM.

Can I move my VM to a different availability zone?

You cannot directly change the availability zone where the VM is hosted. However, you can create a copy of the VM in the availability zone you need.

Can I move my VM to a different folder?

You cannot directly change the folder that the VM belongs to. However, you can create a copy of the VM in the appropriate folder.

If I accidentally delete my VM, can I restore it?

No, you can't. Deleting a VM is an irreversible operation.

To avoid losing your data in case of accidental deletion, you can configure disk backups using snapshots or specify that the disks should not be automatically deleted when deleting the VM.

Adding plugins

How do I connect to a VM running Windows Server

See: Connecting to a Windows VM via RDP.

For the first connection, use the Administrator account. It's a good idea to choose a strong password for this account and create another account for everyday work and connect via the latter in the future.

For images without additional Remote Desktop Services licenses, a maximum of two simultaneous RDP connections are possible.

How do I connect to a Linux VM

See: Connecting to a Linux VM via SSH. If the computer you're connecting from has an older version of Windows (7, 8, and the first releases of Windows 10) and there is no built-in SSH console client, use PuTTY.

The Linux images from Yandex Cloud are initially supplied without a graphical shell. Connection via SSH using a username and a password is disabled by default because this method is outdated and unsafe.

What do I do if I can't connect after creating the VM

Your device should "see" the VM over the network:

If you're connecting from the outside, a public IP address must be assigned to the VM or connectivity must be configured via another VM with a public IP address (for example, via an NAT instance).
Outgoing traffic to and from the VM must be allowed on your device.

You shouldn't turn off or restart the VM immediately after creating it. For VMs with a small guaranteed vCPU performance, the initial launch process may take a long time (up to half an hour). If the VM is terminated prematurely, the initialization scripts might not have time to execute. In this case, delete the VM and create a new one.

How do I use root on a Linux VM

When creating a Linux VM, don't use usernames reserved by the system, such as root, admin, and so on, because cloud-init can't add a user with this kind of username. Try creating a VM by specifying the username user.

To upgrade your rights to root, use sudo. The user specified when creating the VM is a member of the sudo group. root doesn't have a password, so you can simply connect to the VM via SSH (see Connecting to a Linux VM via SSH) and enter the command:

sudo su

You can also execute various commands without switching to root. For example, to shut down a VM from the guest OS, just enter the command:

sudo poweroff

What do I do if I can't connect to the VM, even though the connection was working before

Your data was likely taken over and third parties gained access to the VM. Take a snapshot of the VM's disk (see Creating a disk snapshot) and create a new VM from it (see Create a VM with disks restored from snapshots). If you were able to connect to the new VM, use stronger passwords (for Windows Server) and don't share your private data.

If the VM created from the snapshot is also unavailable for connection, see How do I get important data from a broken VM.

How do I set up an RDP/VNC connection to a Linux VM

There are two options for enabling the graphical interface on Linux VMs:

Use RDP or VNC. To do this:
1. Install the desktop environment on a Linux-based VM (Ubuntu/CentOS).
2. Install the xrdp or VNC server to connect via RDP or VNC, respectively.
3. Configure the software.
  
  We recommend using a combination of a VNC server that listens only to localhost and an SSH tunnel. In this case, you eliminate the risk of attackers connecting to your VM via VNC and also encrypt the VNC traffic. You may find it useful to read this Habr article.
Configure X11 forwarding onto the local computer. For Windows, use Xming.
1. Install the desktop environment.
2. Configure X11 forwarding and connect via SSH.
3. Launch the application from the terminal.

How do I use the serial console / How do I directly connect to a VM

If you have problems connecting to a VM over a network, you can use the serial console. For more information about charts, see Getting started with the serial console.

You can enable access to the serial console when creating or modifying a VM. The serial console is available in the Yandex Cloud management console in the VM menu.

For a Linux VM, you additionally need to configure password access for the user in advance by connecting via SSH and setting a password with the sudo passwd user command (instead of user, specify the username that was set when creating the VM).

Why may the VM not work after a reboot

The VM may stop working after a reboot for one of the following reasons:

The VM was forcibly restarted while writing to the system disk. In this case, the file system could be damaged.
The firewall and/or network was incorrectly configured.
Changes were made to the cloud-init settings.
Changes were made to the python system version, which is used by cloud-init as well.
There was a problem with the service.

If you performed one of these actions, see How do I get important data from a broken VM. Otherwise, contact support.

How do I get important data from a broken VM

When a VM malfunctions:

Take a snapshot of the problem disk: see Creating a disk snapshot.
Create a new VM with an additional (non-bootable) disk that was restored from the snapshot: see Create a VM with disks restored from snapshots.
Connect to the VM: see Working on VMs.
Mount the disk: see Mounting a disk created from a snapshot or image.
Run a check of the disk's file system.
Transfer the necessary data to the boot disk of the new VM.

Disks, snapshots, and images

How much disk space can I use for a VM?

For disk limitations, see Quotas and limits in Compute Cloud.

How do I change the disk size

You can resize your disk within the set limits using the instructions Increasing a disk's size. The data on the disk is kept. Make sure to wait until the operation is complete.

Reducing the disk size is impossible: this is an architectural feature of Yandex Cloud technologies. It's also impossible to create from the snapshot a disk smaller than the parent disk.

If the partition on the Linux boot disk doesn't expand automatically after increasing the disk size, use the following commands:

sudo growpart /dev/vda 2

sudo resize2fs /dev/vda2

For non-bootable disks, the partition size doesn't automatically go up. Use standard OS tools to expand it (for example, parted).

How do I upload my image

See: Uploading a disk image to Yandex Cloud.

For the virtual machine created from your image to function properly, you must follow all the recommendations in the instructions.

If you're confident that you followed all the recommendations, but the image doesn't work, or if you have any additional questions, contact support.

Should I use swap

We strongly recommend not using swap in cloud systems because the disk subsystem can become a "bottleneck" for the entire guest system. Network disk limits are low enough to use the disk as a RAM "extension".

Instead of swap, you can use the zram-config utility. It allows you to organize a kind of swap inside the RAM itself thanks to compression, which is ten times faster than IO with a disk. Keep in mind that with a high load on IO and/or vCPUs, the use of zram-config may negatively affect the iowait and, consequently, the operation of the network, disks, and vCPUs.

The best solution for increasing the available memory is to expand the vRAM on the VM.

The snapshot size is larger than there was data on the disk

This happens because occupied sectors remain on the disk after deleting the files.

The solution is to write a file consisting of zeros to the entire unoccupied disk space, then flush the cache to the disk and delete the record about this file.

For Windows: stop disk operations and use the SDelete utility. You can find out how it works and download it in the Microsoft documentation.
For Linux: stop disk operations and enter the following commands one by one:
```
dd if=/dev/zero | pv > full.disk
```
```
sync
```
```
rm full.disk
```

The "empty" space on the disk then becomes truly empty and you can create a disk snapshot. Its size is similar to the current occupied disk space.

How are snapshot quotas counted

Snapshots are counted in the quotas based on the size of the parent disks. For example, if a snapshot was created from a 250 GB disk, the quota for the size of snapshots allows for 250 GB, even if the actual size of the snapshot is 20 GB. This is done so that both developers and customers can realistically understand possible loads on the disk snapshot service.

This approach of counting the quota doesn't affect the pricing: snapshots are charged according to their real size.

How do I move a VM to another folder/cloud

Grant rights in your cloud to a user from another cloud:
- Role for the cloud: resource-manager.clouds.member.
- Role for the folder: viewer or compute.images.user.
See the instructions Assigning roles.
Create an image from your snapshot under Disk snapshots or from the disk itself under Disks.

A user in another cloud must:

Run the CLI command below:

yc compute image create --source-image-id=<your_image_ID>

When creating a VM, specify this image as a boot disk.

How do I attach a new disk to a VM

After creating and connecting a new disk to the VM, you need to mount it or assign it a letter, depending on the operating system. See: Mounting a disk created from a snapshot or image.

How do I set up automatic backups

Currently, Compute Cloud doesn't have a ready-made solution for automatic backups, but you can use the approach described in our blog: Creating scheduled disk snapshots with Yandex Cloud Functions.

Note that the code is provided as is. Its further modification is outside the scope of Yandex Cloud.

To back up the data manually, use disk snapshots.

What happens to my data when a VM is deleted?

When selecting a disk to attach to a VM, you can specify whether the disk should be deleted along with the VM. You can choose this option when creating a VM, reconfiguring it, or attaching a new disk.

If a VM had previously created disks attached, they will be detached when you delete the VM. The disk data is preserved and this disk can be attached to another VM in the future.

If you would like to delete a disk with a VM, specify this option when creating the VM, reconfiguring it, or attaching the disk. Such disks will be deleted along with the VM.

Do I need to stop a VM to create disk snapshots? Do I have to wait until disk snapshots are created before I start a VM?

You don't have to stop the VM. However, keep in mind that a snapshot contains only the data written to disk when creating the snapshot. You need to take care of the data integrity yourself. For information about how to create disk snapshots, see Creating a disk snapshot.

A snapshot is created asynchronously. You can resume writing data to disk immediately after running the create snapshot command, without waiting for the snapshot creation to be completed.

Disaster recovery

Troubleshooting instructions in the event of VM disk failures

Warning

Don't try to restart your VM yourself. This may result in data loss.

If an attempt to access one or more disks of your VM returns an error:

Create a new disk and attach it to your VM.
Copy the critical data from damaged disks to the new disk. Make sure you don't use the O_DIRECT flag when reading files.
Once all the critical data is copied, detach the damaged disks from your VM and delete them.
If you fail to copy the data and need to recover the disk, contact us. We'll stop the VM on our own and check the disk integrity. This may take several days.

Instance groups

What is Instance Groups?

Instance Groups is a component that allows you to create, use, and scale groups of similar instances in the Yandex Compute Cloud infrastructure.

With Instance Groups, you can:

Create groups with the required number of instances and performance parameters.
Scale the computing capacity up or down, depending on the load.

You work with an instance group as a single entity in the Yandex Compute Cloud infrastructure. This allows you to manage the internal settings of the instance group to meet your application's requirements.

How is the cost of instance group use calculated?

Creation of an instance group is free of charge.

All other Yandex Cloud services, such as VMs and external IP addresses, are charged as usual.

How do I know I'm not overpaying?

To select the appropriate number of instances and minimize costs:

Estimate the amount of computing resources required for your service and see the calculation examples and pricing policy for Yandex Compute Cloud.
Try to frequently monitor the load on the service at different times of day.

Monitoring

Why does the vCPU on the graph show more than 100%

If you use cores with guaranteed, say, 5% vCPU performance, then this 5% represents 100% of the expected load for the monitoring system. If there are no "neighbors" on the physical core, you can be allocated up to 100% of vCPU performance, which is 20 times higher than the maximum expected load (×20 of 5%). Thus, the graph can show up to 2000%.

If you see the upper limit of 100% exceeded for quite a while in the graphs, we recommend increasing the guaranteed vCPU performance because "neighbors" may appear on the physical core at any time and your real utilization of physical core resources will drop to the guaranteed 5% (about 100 MHz). In this case, the guest system may not cope with the load and you'll lose access to the VM.

How do I track vRAM use using monitoring

The Compute Cloud service can't measure vRAM consumption inside the guest operating system because for the service, memory consumption by the virtual machine is always the same: the one that is allocated the moment it is started.

To track the vRAM state, use the Yandex Monitoring service. It allows you to record your own metrics. See Writing custom metrics via the API. Just schedule a task to download data about the vRAM state and Yandex Monitoring will display them.

Licensing

The terms of use for Microsoft software are governed by the terms and conditions of the Microsoft license agreement that you enter into when purchasing a Microsoft product. The user is responsible for complying with the Microsoft licensing terms. Licensing recommendations are provided below. You are in no way obligated to follow such recommendations as they are not legally binding. If you have any questions about Microsoft software licensing and product usage rights, please consult your legal department or Microsoft reseller. The information on this page complies with the current Microsoft Product Terms.

General questions about licensing

What is the relationship between Microsoft and Yandex Cloud?

Yandex Cloud is licensed to distribute Microsoft software products under the Services Provider License Agreement (SPLA) for providing software services. Yandex Cloud is also an authorized Microsoft partner under the License Mobility through Software Assurance program.

Does Yandex Cloud provide extended support for Microsoft products?

No, the Yandex Cloud team does not currently offer support for Microsoft products.

What is License Mobility?

License Mobility is a benefit provided to customers with Microsoft corporate licenses for eligible server software covered by active Microsoft Software Assurance (SA) agreements. With License Mobility, customers can use eligible Microsoft software when working with third-party cloud solution providers such as Yandex Cloud. To learn more, go to the Microsoft website.

What Microsoft software can I run on the Yandex Cloud platform?

Yandex Cloud supports server software available under the License Mobility through Software Assurance program.

How can I find out whether a Microsoft product is eligible for License Mobility?

This information is included in the terms of use for the Microsoft product. Whether or not a product is eligible for the License Mobility through Software Assurance program is indicated in the corresponding section of the Software Assurance agreement. Products that are eligible for the License Mobility through Software Assurance program include Remote Desktop Services, System Center, Exchange, and SharePoint.

Do I need to have a current Software Assurance agreement in place and participate in the License Mobility through Software Assurance program to deploy my own Microsoft licenses in Yandex Cloud?

Yes. When using licensed Microsoft software in Yandex Compute Cloud, you must have a current Software Assurance agreement in place and participate in the License Mobility through Software Assurance program. An active Software Assurance agreement is always a requirement for participating in the License Mobility through Software Assurance program.

Can I import my own Windows image to Yandex Cloud?

No. Currently, you cannot upload your own images with Windows.

Why do I need to specify the address when using Microsoft products?

As of January 24, 2022, to use Microsoft products, you need to specify user data and addresses of individuals and legal entities. These are the requirements of the Microsoft licensing policy. If the data isn't provided, you can't run Microsoft products.

Can I deploy my own licenses in Compute Cloud?

Questions about Windows Server licensing are discussed in detail in Windows licensing.

In the case of other software that you want to use inside the VM, read the license you purchased or contact the technical support of the product itself and they'll tell you if it's possible to use the product with this license in Yandex Cloud.

Windows Server licensing

Can I purchase Windows Server from Yandex Cloud?

Yes, you can purchase a VM with a pre-installed Windows Server 2012 R2, Windows Server 2016, or Windows Server 2019 image from Yandex Cloud. The Windows Server license is included in the Yandex Cloud service cost.

How do I activate Windows Server

You may have activation issues if the VM doesn't have internet access.

Make sure that the VM has internet access.

Run the following commands in PowerShell as an administrator:

Test-NetConnection kms.cloud.yandex.net -Port 1688
cscript.exe C:\Windows\System32\slmgr.vbs /dli
cscript.exe C:\Windows\System32\slmgr.vbs /skms kms.cloud.yandex.net:1688
cscript.exe C:\Windows\System32\slmgr.vbs /ato
Get-Date
Get-TimeZone

If activation was not successful, send us the output of these commands.

Does License Mobility cover Windows Server?

No. According to the terms of use for Microsoft products, License Mobility does not extend to Windows Server operating systems.

Licensing of other Microsoft products

Can I purchase other Microsoft products with licenses that can be deployed in Yandex Compute Cloud?

Yes. Currently, Yandex Cloud provides Windows Server, Microsoft SQL Server, and RDS licenses and transfers server software licenses under the License Mobility through Software Assurance program. Licensing of MSDN, Windows OS, Microsoft Office, and other products is not available.