1 - What is Talos?

Talos is a container optimized Linux distro; a reimagining of Linux for distributed systems such as Kubernetes. Designed to be as minimal as possible while still maintaining practicality. For these reasons, Talos has a number of features unique to it:

  • it is immutable
  • it is atomic
  • it is ephemeral
  • it is minimal
  • it is secure by default
  • it is managed via a single declarative configuration file and gRPC API

Talos can be deployed on container, cloud, virtualized, and bare metal platforms.

Why Talos

In having less, Talos offers more. Security. Efficiency. Resiliency. Consistency.

All of these areas are improved simply by having less.

2 - Quickstart

Local Docker Cluster

The easiest way to try Talos is by using the CLI (talosctl) to create a cluster on a machine with docker installed.



Download talosctl:

curl -Lo /usr/local/bin/talosctl https://github.com/talos-systems/talos/releases/latest/download/talosctl-$(uname -s | tr "[:upper:]" "[:lower:]")-amd64
chmod +x /usr/local/bin/talosctl

For linux and darwin operating systems talosctl is also available for the arm64 processor architecture.

curl -Lo /usr/local/bin/talosctl https://github.com/talos-systems/talos/releases/latest/download/talosctl-$(uname -s | tr "[:upper:]" "[:lower:]")-arm64
chmod +x /usr/local/bin/talosctl


Download kubectl via one of methods outlined in the documentation.

Create the Cluster

Now run the following:

talosctl cluster create

Verify that you can reach Kubernetes:

$ kubectl get nodes -o wide
talos-default-master-1   Ready    master   115s   v1.20.2      <none>        Talos (v0.14.0)   <host kernel>    containerd://1.5.5
talos-default-worker-1   Ready    <none>   115s   v1.20.2      <none>        Talos (v0.14.0)   <host kernel>    containerd://1.5.5

Destroy the Cluster

When you are all done, remove the cluster:

talosctl cluster destroy

3 - Getting Started

This document will walk you through installing a full Talos Cluster. You may wish to read through the Quickstart first, to quickly create a local virtual cluster on your workstation.

Regardless of where you run Talos, you will find that there is a pattern to deploying it.

In general you will need to:

  • acquire the installation image
  • decide on the endpoint for Kubernetes
    • optionally create a load balancer
  • configure Talos
  • configure talosctl
  • bootstrap Kubernetes



The talosctl tool provides a CLI tool which interfaces with the Talos API in an easy manner. It also includes a number of useful tools for creating and managing your clusters.

You should install talosctl before continuing:


curl -Lo /usr/local/bin/talosctl https://github.com/talos-systems/talos/releases/latest/download/talosctl-$(uname -s | tr "[:upper:]" "[:lower:]")-amd64
chmod +x /usr/local/bin/talosctl


For linux and darwin operating systems talosctl is also available for the arm64 processor architecture.

curl -Lo /usr/local/bin/talosctl https://github.com/talos-systems/talos/releases/latest/download/talosctl-$(uname -s | tr "[:upper:]" "[:lower:]")-arm64
chmod +x /usr/local/bin/talosctl

Acquire the installation image

The easiest way to install Talos is to use the ISO image.

The latest ISO image can be found on the Github Releases page:

For self-built media and network booting, you can use the kernel and initramfs:

When booted from the ISO, Talos will run in RAM, and it will not install itself until it is provided a configuration. Thus, it is safe to boot the ISO onto any machine.

Alternative Booting

If you wish to use a different boot mechanism (such as network boot or a custom ISO), there are a number of required kernel parameters.

Please see the kernel docs for more information.

Decide the Kubernetes Endpoint

In order to configure Kubernetes and bootstrap the cluster, Talos needs to know what the endpoint (DNS name or IP address) of the Kubernetes API Server will be.

The endpoint should be the fully-qualified HTTP(S) URL for the Kubernetes API Server, which (by default) runs on port 6443 using HTTPS.

Thus, the format of the endpoint may be something like:

  • https://kube.mycluster.mydomain.com:6443
  • https://[2001:db8:1234::80]:6443

Because the Kubernetes controlplane is meant to be supplied in a high availability manner, we must also choose how to bind it to the servers themselves. There are three common ways to do this.

Dedicated Load-balancer

If you are using a cloud provider or have your own load-balancer available (such as HAProxy, nginx reverse proxy, or an F5 load-balancer), using a dedicated load balancer is a natural choice. Just create an appropriate frontend matching the endpoint, and point the backends at each of the addresses of the Talos controlplane nodes.

This is convenient if a load-balancer is available, but don’t worry if that is not the case.

Layer 2 Shared IP

Talos has integrated support for serving Kubernetes from a shared (sometimes called “virtual”) IP address. This method relies on OSI Layer 2 connectivity between controlplane Talos nodes.

In this case, we may choose an IP address on the same subnet as the Talos controlplane nodes which is not otherwise assigned to any machine. For instance, if your controlplane node IPs are:


You could choose the ip as your shared IP address. Just make sure that is not used by any other machine and that your DHCP will not serve it to any other machine.

Once chosen, form the full HTTPS URL from this IP:

You are also free to set a DNS record to this IP address instead, but you will still need to use the IP address to set up the shared IP (machine.network.interfaces[].vip.ip) inside the Talos configuration.

For more information about using a shared IP, see the related Guide

DNS records

If neither of the other methods work for you, you can instead use DNS records to provide a measure of redundancy. In this case, you would add multiple A or AAAA records for a DNS name.

For instance, you could add:

kube.cluster1.mydomain.com  IN  A
kube.cluster1.mydomain.com  IN  A
kube.cluster1.mydomain.com  IN  A

Then, your endpoint would be:


Decide how to access the Talos API

Since Talos is entirely API-driven, it is important to know how you are going to access that API. Talos comes with a number of mechanisms to make that easier.

Controlplane nodes can proxy requests for worker nodes. This means that you only need access to the controlplane nodes in order to access the rest of the network. This is useful for security (your worker nodes do not need to have public IPs or be otherwise connected to the Internet), and it also makes working with highly-variable clusters easier, since you only need to know the controlplane nodes in advance.

Even better, the talosctl tool will automatically load balance and fail over between all of your controlplane nodes, so long as it is informed of each of the controlplane node IPs.

That does, of course, present the problem that you need to know how to talk to the controlplane nodes. In some environments, it is easy to be able to forecast, prescribe, or discover the controlplane node IP addresses. For others, though, even the controlplane nodes are dynamic, unpredictable, and undiscoverable.

The dynamic options above for the Kubernetes API endpoint also apply to the Talos API endpoints. The difference is that the Talos API runs on port 50000/tcp.

Whichever way you wish to access the Talos API, be sure to note the IP(s) or hostname(s) so that you can configure your talosctl tool’s endpoints below.

Configure Talos

When Talos boots without a configuration, such as when using the Talos ISO, it enters a limited maintenance mode and waits for a configuration to be provided.

Alternatively, the Talos installer can be booted with the talos.config kernel commandline argument set to an HTTP(s) URL from which it should receive its configuration. In cases where a PXE server can be available, this is much more efficient than manually configuring each node. If you do use this method, just note that Talos does require a number of other kernel commandline parameters. See the required kernel parameters for more information.

In either case, we need to generate the configuration which is to be provided. Luckily, the talosctl tool comes with a configuration generator for exactly this purpose.

  talosctl gen config "cluster-name" "cluster-endpoint"

Here, cluster-name is an arbitrary name for the cluster which will be used in your local client configuration as a label. It does not affect anything in the cluster itself. It is arbitrary, but it should be unique in the configuration on your local workstation.

The cluster-endpoint is where you insert the Kubernetes Endpoint you selected from above. This is the Kubernetes API URL, and it should be a complete URL, with https:// and port, if not 443. The default port is 6443, so the port is almost always required.

When you run this command, you will receive a number of files in your current directory:

  • controlplane.yaml
  • worker.yaml
  • talosconfig

The three .yaml files are what we call Machine Configs. They are installed onto the Talos servers to act as their complete configuration, describing everything from what disk Talos should be installed to, to what sysctls to set, to what network settings it should have. In the case of the controlplane.yaml, it even describes how Talos should form its Kubernetes cluster.

The talosconfig file (which is also YAML) is your local client configuration file.

Controlplane, Init, and Worker

The three types of Machine Configs correspond to the three roles of Talos nodes. For our purposes, you can ignore the Init type. It is a legacy type which will go away eventually. Its purpose was to self-bootstrap. Instead, we now use an API call to bootstrap the cluster, which is much more robust.

That leaves us with Controlplane and Worker.

The Controlplane Machine Config describes the configuration of a Talos server on which the Kubernetes Controlplane should run. The Worker Machine Config describes everything else: workload servers.

The main difference between Controlplane Machine Config files and Worker Machine Config files is that the former contains information about how to form the Kubernetes cluster.


The generated files can be thought of as templates. Individual machines may need specific settings (for instance, each may have a different static IP address). When different files are needed for machines of the same type, simply copy the source template (controlplane.yaml or worker.yaml) and make whatever modifications need to be done.

For instance, if you had three controlplane nodes and three worker nodes, you may do something like this:

  for i in $(seq 0 2); do
    cp controlplane.yaml cp$i.yaml
  for i in $(seq 0 2); do
    cp worker.yaml w$i.yaml

In cases where there is no special configuration needed, you may use the same file for each machine of the same type.

Apply Configuration

After you have generated each machine’s Machine Config, you need to load them into the mahines themselves. For that, you need to know their IP addresses.

If you have access to the console or console logs of the machines, you can read them to find the IP address(es). Talos will print them out during the boot process:

[    4.605369] [talos] task loadConfig (1/1): this machine is reachable at:
[    4.607358] [talos] task loadConfig (1/1):
[    4.608766] [talos] task loadConfig (1/1): server certificate fingerprint:
[    4.611106] [talos] task loadConfig (1/1):   xA9a1t2dMxB0NJ0qH1pDzilWbA3+DK/DjVbFaJBYheE=
[    4.613822] [talos] task loadConfig (1/1):
[    4.614985] [talos] task loadConfig (1/1): upload configuration using talosctl:
[    4.616978] [talos] task loadConfig (1/1):   talosctl apply-config --insecure --nodes --file <config.yaml>
[    4.620168] [talos] task loadConfig (1/1): or apply configuration using talosctl interactive installer:
[    4.623046] [talos] task loadConfig (1/1):   talosctl apply-config --insecure --nodes --interactive
[    4.626365] [talos] task loadConfig (1/1): optionally with node fingerprint check:
[    4.628692] [talos] task loadConfig (1/1):   talosctl apply-config --insecure --nodes --cert-fingerprint 'xA9a1t2dMxB0NJ0qH1pDzilWbA3+DK/DjVbFaJBYheE=' --file <config.yaml>

If you do not have console access, the IP address may also be discoverable from your DHCP server.

Once you have the IP address, you can then apply the correct configuration.

  talosctl apply-config --insecure \
    --nodes \
    --file cp0.yaml

The insecure flag is necessary at this point because the PKI infrastructure has not yet been made available to the node. Note that the connection will be encrypted, it is just unauthenticated.

If you have console access, though, you can extract the server certificate fingerprint and use it for an additional layer of validation:

  talosctl apply-config --insecure \
    --nodes \
    --cert-fingerprint xA9a1t2dMxB0NJ0qH1pDzilWbA3+DK/DjVbFaJBYheE= \
    --file cp0.yaml

Using the fingerprint allows you to be sure you are sending the configuration to the right machine, but it is completely optional.

After the configuration is applied to a node, it will reboot.

You may repeat this process for each of the nodes in your cluster.

Configure your talosctl client

Now that the nodes are running Talos with its full PKI security suite, you need to use that PKI to talk to the machines. That means configuring your client, and that is what that talosconfig file is for.


Endpoints are the communication endpoints to which the client directly talks. These can be load balancers, DNS hostnames, a list of IPs, etc. In general, it is recommended that these point to the set of control plane nodes, either directly or through a reverse proxy or load balancer.

Each endpoint will automatically proxy requests destined to another node through it, so it is not necessary to change the endpoint configuration just because you wish to talk to a different node within the cluster.

Endpoints do, however, need to be members of the same Talos cluster as the target node, because these proxied connections reply on certificate-based authentication.

We need to set the endpoints in your talosconfig. talosctl will automatically load balance and fail over among the endpoints, so no external load balancer or DNS abstraction is required (though you are free to use them, if desired).

As an example, if the IP addresses of our controlplane nodes are:


We would set those in the talosconfig with:

  talosctl --talosconfig=./talosconfig \
    config endpoint


The node is the target node on which you wish to perform the API call.

Keep in mind, when specifying nodes that their IPs and/or hostnames are as seen by the endpoint servers, not as from the client. This is because all connections are proxied first through the endpoints.

Some people also like to set a default set of nodes in the talosconfig. This can be done in the same manner, replacing endpoint with node. If you do this, however, know that you could easily reboot the wrong machine by forgetting to declare the right one explicitly. Worse, if you set several nodes as defaults, you could, with one talosctl upgrade command upgrade your whole cluster all at the same time. It’s a powerful tool, and with that comes great responsibility.

The author of this document generally sets a single controlplane node to be the default node, which provides the most flexible default operation while limiting the scope of the disaster should a command be entered erroneously:

  talosctl --talosconfig=./talosconfig \
    config node

You may simply provide -n or --nodes to any talosctl command to supply the node or (comma-delimited) nodes on which you wish to perform the operation. Supplying the commandline parameter will override any default nodes in the configuration file.

To verify default node(s) you’re currently configured to use, you can run:

$ talosctl version
        NODE:        <node>

For a more in-depth discussion of Endpoints and Nodes, please see talosctl.

Default configuration file

You can reference which configuration file to use directly with the --talosconfig parameter:

  talosctl --talosconfig=./talosconfig \
    --nodes version

However, talosctl comes with tooling to help you integrate and merge this configuration into the default talosctl configuration file. This is done with the merge option.

  talosctl config merge ./talosconfig

This will merge your new talosconfig into the default configuration file ($XDG_CONFIG_HOME/talos/config.yaml), creating it if necessary. Like Kubernetes, the talosconfig configuration files has multiple “contexts” which correspond to multiple clusters. The <cluster-name> you chose above will be used as the context name.

Kubernetes Bootstrap

All of your machines are configured, and your talosctl client is set up. Now, you are ready to bootstrap your Kubernetes cluster. If that sounds daunting, you haven’t used Talos before.

Bootstrapping your Kubernetes cluster with Talos is as simple as:

  talosctl bootstrap --nodes

IMPORTANT: the bootstrap operation should only be called ONCE and only on a SINGLE controlplane node!

The IP there can be any of your controlplanes (or the loadbalancer, if you have one). It should only be issued once.

At this point, Talos will form an etcd cluster, generate all of the core Kubernetes assets, and start the Kubernetes controlplane components.

After a few moments, you will be able to download your Kubernetes client configuration and get started:

  talosctl kubeconfig

Running this command will add (merge) you new cluster into you local Kubernetes configuration in the same way as talosctl config merge merged the Talos client configuration into your local Talos client configuration file.

If you would prefer for the configuration to not be merged into your default Kubernetes configuration file, simple tell it a filename:

  talosctl kubeconfig alternative-kubeconfig

If all goes well, you should now be able to connect to Kubernetes and see your nodes:

  kubectl get nodes

4 - System Requirements

Minimum Requirements

Init/Control Plane2GB2
Init/Control Plane4GB4

These requirements are similar to that of kubernetes.

5 - What's New in Talos 0.14


Kubelet configuration can be updated without node restart (.machine.kubelet section of machine configuration) with commands talosctl edit mc --immediate, talosctl apply-config --immediate, talosctl patch mc --immediate.

Kubelet service can now be restarted with talosctl service kubelet restart.

Kubelet node IP configuration (.machine.kubelet.nodeIP.validSubnets) can now include negative subnet matches (prefixed with !).

Kubernetes Upgrade Enhancements

talosctl upgrade-k8s was improved to:

  • sync all boostrap manifest resources in the Kubernetes cluster with versions bundled with current version Talos
  • upgrade kubelet to the version of the control plane components (without node reboot)

So there is no need to update CoreDNS, Flannel container manually after running upgrade-k8s anymore.

Log Shipping

Talos can now ship system logs to the configured destination using either JSON-over-UDP or JSON-over-TCP: see .machine.logging machine configuration option.

NTP Sync

Talos NTP sync process was improved to align better with kernel time adjustment periods and to filter out spikes.

talosctl support

talosctl CLI tool now has a new subcommand support that gathers all cluster information that could help with debugging in.

Output of the command is a zip archive with all Talos service logs, Kubernetes pod logs and manifests, Talos resources manifests and so on. Generated archive does not contain any secret information, so it is safe to send it for analysis to a third party.

Component Updates

  • Linux: 5.15.6
  • etcd: 3.5.1
  • containerd: 1.5.8
  • runc: 1.0.3
  • Kubernetes: 1.23.1
  • CoreDNS: 1.8.6
  • Flannel (default CNI): 0.15.1

Talos is built with Go 1.17.5

Cluster Discovery

Cluster Discovery is enabled by default for Talos 0.14. Cluster Discovery can be disabled with talosctl gen config --with-cluster-discovery=false.

Kexec and capabilities

When kexec support is disabled Talos no longer drops Linux capabilities (CAP_SYS_BOOT and CAP_SYS_MODULES) for child processes. That is helpful for advanced use-cases like Docker-in-Docker.

If you want to permanently disable kexec and capabilities dropping, pass kexec_load_disabled=1 argument to the kernel.

For example:

    - sysctl.kernel.kexec_load_disabled=1

Please note that capabilities are dropped before machine configuration is loaded, so disabling kexec via machine.sysctls will not be enough.

installer and imager images

Talos supports two target architectures: amd64 and arm64, so all Talos images are built for both amd64 and arm64.

New image imager was added which contains Talos assets for both architectures which allows to generate Talos disk images cross-arch: e.g. generate Talos Raspberry PI disk image on amd64 machine.

As installer image is used only to do initial install and upgrades, it now contains Talos assets for a specific architecture. This reduces size of the installer image leading to faster upgrades and less memory usage.

There are no user-visible changes except that now imager container image should be used to produce Talos disk images.

A set of Talos ehancements is going to unlock a number of exciting features in the upcoming release of Sidero:

  • SideroLink: a point-to-point Wireguard tunnel connecting Talos node back to the provisioning platform (Sidero).
  • event sink (kernel arg talos.event.sink= delivers Talos internal events to the specified destination.
  • kmsg log delivery (kernel arg talos.logging.kernel=tcp:// sends kernel logs as JSON lines over TCP or UDP.

VLAN Enhancements

Talos now supports setting MTU and Virtual IPs on VLAN interfaces.

6 - Support Matrix

Talos Version0.140.13
Release Date2021-12-212021-10-11 (0.13.0)
End of Community Support1.0.0 release (2022-03-27, TBD)0.14.0 release (2021-12-21)
Enterprise Supportoffered by Sidero Labs Inc.
Kubernetes1.23, 1.22, 1.211.22, 1.21, 1.20
Architectureamd64, arm64
- cloudAWS, GCP, Azure, Digital Ocean, Hetzner, OpenStack, Scaleway, Vultr, Upcloud
- bare metalx86: BIOS, UEFI; arm64: UEFI; boot: ISO, PXE, disk image
- virtualizedVMware, Hyper-V, KVM, Proxmox, Xen
- SBCsRaspberry Pi4, Banana Pi M64, Pine64, and other
- localDocker, QEMU
Cluster API
CAPI Bootstrap Provider Talos>= 0.4.3>= 0.3.0
CAPI Control Plane Provider Talos>= 0.4.1>= 0.1.1
Sidero>= 0.4.1>= 0.3.0

Platform Tiers

Tier 1: Automated tests, high-priority fixes. Tier 2: Tested from time to time, medium-priority bugfixes. Tier 3: Not tested by core Talos team, community tested.

Tier 1

  • Metal
  • AWS
  • GCP

Tier 2

  • Azure
  • Digital Ocean
  • OpenStack
  • VMWare

Tier 3

  • Hetzner
  • nocloud
  • Scaleway
  • Vultr
  • Upcloud