Extensions are posted the following way:
`extensions.talos.dev/<name>=<version>`
The name should be valid as a label (annotation) key.
If the value is valid as a label value, use labels, otherwise use
annotations.
Also implements node annotations in the machine config as a side-effect.
Fixes#9089Fixes#8971
See #9070
Signed-off-by: Andrey Smirnov <andrey.smirnov@siderolabs.com>
Add test for Gvisor extensions when kvm platform is used.
The test is marked as skipped until pod termination issue is resolved.
Signed-off-by: Noel Georgi <git@frezbo.dev>
- replace `interface{}` with `any` using `gofmt -r 'interface{} -> any -w'`
- replace `a = []T{}` with `var a []T` where possible.
- replace `a = []T{}` with `a = make([]T, 0, len(b))` where possible.
Signed-off-by: Dmitriy Matrenichev <dmitry.matrenichev@siderolabs.com>
The iscsi test broke when the new disks api was introduced making the
test pass always, now filter other only `iscsi` disk types using the new
disks API.
Signed-off-by: Noel Georgi <git@frezbo.dev>
When META has never been written (e.g. booted from a disk image), it
won't be detected as `talosmeta`.
Signed-off-by: Andrey Smirnov <andrey.smirnov@siderolabs.com>
Detect CD devices, and set size to 0 for CD without media.
In user disk wipe tests, skip device mapper devices and CD-ROM.
Signed-off-by: Andrey Smirnov <andrey.smirnov@siderolabs.com>
Uses go-siderolabs/go-blockdevice/v2 for all the hard parts,
provides new resource `Disk` which describes all disks in the system.
Additional resource `SystemDisk` always point to the system disk (based
on the location of `META` partition).
The `Disks` API (and `talosctl disks`) provides a view now into the
`talosctl get disks` to keep backwards compatibility.
QEMU provisioner can now create extra disks of various types: IDE, AHCI,
SCSI, NVME, this allows to test detection properly.
The new resource will be the foundation for volume provisioning (to pick
up the disk to provision the volume on).
Example:
```
talosctl -n 172.20.0.5 get disks
NODE NAMESPACE TYPE ID VERSION SIZE READ ONLY TRANSPORT ROTATIONAL WWID MODEL SERIAL
172.20.0.5 runtime Disk loop0 1 65568768 true
172.20.0.5 runtime Disk nvme0n1 1 10485760000 false nvme nvme.1b36-6465616462656566-51454d55204e564d65204374726c-00000001 QEMU NVMe Ctrl deadbeef
172.20.0.5 runtime Disk sda 1 10485760000 false virtio true QEMU HARDDISK
172.20.0.5 runtime Disk sdb 1 10485760000 false sata true t10.ATA QEMU HARDDISK QM00013 QEMU HARDDISK
172.20.0.5 runtime Disk sdc 1 10485760000 false sata true t10.ATA QEMU HARDDISK QM00001 QEMU HARDDISK
172.20.0.5 runtime Disk vda 1 12884901888 false virtio true
```
Signed-off-by: Andrey Smirnov <andrey.smirnov@siderolabs.com>
Use shared locks, discover more partitions, some other small changes.
Re-enable the flaky test.
Signed-off-by: Andrey Smirnov <andrey.smirnov@siderolabs.com>
Try to activate/deactivate watchdogs, change timeout, run only on QEMU.
Signed-off-by: Andrey Smirnov <andrey.smirnov@siderolabs.com>
Signed-off-by: Dmitry Sharshakov <dmitry.sharshakov@siderolabs.com>
- By default, github.com/miekg/dns uses `dns.MinMsgSize` for UDP messages, which is 512 bytes. This is too small for some
DNS request/responses, and can cause truncation and errors. This change sets the buffer size to `dns.DefaultMsgSize`
4096 bytes, which is the maximum size of a dns packet payload per RFC 6891.
- We also retry the request if the response is truncated or previous connection was closed.
- And finally we properly handle the case where the response is larger than the client buffer size,
and we return a truncated correct response.
Closes#8763
Signed-off-by: Dmitriy Matrenichev <dmitry.matrenichev@siderolabs.com>
Note: this issue never happens with default Talos worker configuration
(generated by Omni, `talosctl gen config` or CABPT).
Before change https://github.com/siderolabs/talos/pull/4294 3 years ago,
worker nodes connected to trustd in "insecure" mode (without validating
the trustd server certificate). The change kept backwards compatibility,
so it still allowed insecure mode on upgrades.
Now it's time to break this compatibility promise, and require
accepted CAs to be always present. Adds validation for machine
configuration, so if upgrade is attempeted, it would not validate the
machine config without accepted CAs.
Now lack of accepted CAs would lead to failure to connect to trustd.
Signed-off-by: Andrey Smirnov <andrey.smirnov@siderolabs.com>
Run a health check before the test, as the test depends on CoreDNS being
healthy, and previous tests might disturb the cluster.
Also refactor by using watch instead of retries, make pods terminate
fast.
Signed-off-by: Andrey Smirnov <andrey.smirnov@siderolabs.com>
Provide a trace for each step of the reset sequence taken, so if one of
those fails, integration test produces a meaningful message instead of
proceeding and failing somewhere else.
More cleanup/refactor, should be functionally equivalent.
Fixes#8635
Signed-off-by: Andrey Smirnov <andrey.smirnov@siderolabs.com>
More specifically, pick up `/etc/resolv.conf` contents by default when
in container mode, and use that as a base resolver for the host DNS.
Fixes#8303
Signed-off-by: Andrey Smirnov <andrey.smirnov@siderolabs.com>
This fixes an issue with a single controlplane cluster.
Properly present all accepted CAs to the apiserver, in the test let the
cluster fully recovery between two CA rotations performed.
Signed-off-by: Andrey Smirnov <andrey.smirnov@siderolabs.com>
Dynamically map Kubernetes and Talos API ports to an available port on
the host, so every cluster gets its own unique set of parts.
As part of the changes, refactor the provision library and interfaces,
dropping old weird interfaces replacing with (hopefully) much more
descriprive names.
Signed-off-by: Dmitry Sharshakov <dmitry.sharshakov@siderolabs.com>
Signed-off-by: Andrey Smirnov <andrey.smirnov@siderolabs.com>
Fixes#8361
Talos requires v2 (circa 2008), but VMs are often configured to limit
the exposed features to the baseline (v1).
```
[ 0.779218] [talos] [initramfs] booting Talos v1.7.0-alpha.1-35-gef5bbe728-dirty
[ 0.779806] [talos] [initramfs] CPU: QEMU Virtual CPU version 2.5+, 4 core(s), 1 thread(s) per core
[ 0.780529] [talos] [initramfs] x86_64 microarchitecture level: 1
[ 0.781018] [talos] [initramfs] it might be that the VM is configured with an older CPU model, please check the VM configuration
[ 0.782346] [talos] [initramfs] x86_64 microarchitecture level 2 or higher is required, halting
```
Signed-off-by: Andrey Smirnov <andrey.smirnov@siderolabs.com>
This allows to roll all nodes to use a new CA, to refresh it, or e.g.
when the `talosconfig` was exposed accidentally.
Signed-off-by: Andrey Smirnov <andrey.smirnov@siderolabs.com>
This controller combines kobject events, and scan of `/sys/block` to
build a consistent list of available block devices, updating resources
as the blockdevice changes.
Based on these resources the next step can run probe on the blockdevices
as they change to present a consistent view of filesystems/partitions.
Signed-off-by: Andrey Smirnov <andrey.smirnov@siderolabs.com>
The current code was stipping non-`v1alpha1.Config` documents. Provide a
proper method in the config provider, and update places using it.
Signed-off-by: Andrey Smirnov <andrey.smirnov@siderolabs.com>
Let's add a very basic test for the Kata Containers extension, mimicing
what's already in place for gVisor.
This depends on the work being done in:
https://github.com/siderolabs/extensions/pull/279
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
Signed-off-by: Noel Georgi <git@frezbo.dev>
Fixes#8057
I went back and forth on the way to fix it exactly, and ended up with a
pretty simple version of a fix.
The problem was that discovery service was removing the member at the
initial phase of reset, which actually still requires KubeSpan to be up:
* leaving `etcd` (need to talk to other members)
* stopping pods (might need to talk to Kubernetes API with some CNIs)
Now leaving discovery service happens way later, when network
interactions are no longer required.
Signed-off-by: Andrey Smirnov <andrey.smirnov@siderolabs.com>
Fixes#4421
See documentation for details on how to use the feature.
With `talosctl cluster create`, firewall can be easily test with
`--with-firewall=accept|block` (default mode).
Signed-off-by: Andrey Smirnov <andrey.smirnov@siderolabs.com>
This PR adds support for custom node taints. Refer to `nodeTaints` in the `configuration` for more information.
Closes#7581
Signed-off-by: Dmitriy Matrenichev <dmitry.matrenichev@siderolabs.com>
Fixes#7873
Some services which perform mounts inside the container which require
mounts to propagate back to the host (e.g. `stargz-snapshotter`) require
this configuration setting.
Signed-off-by: Andrey Smirnov <andrey.smirnov@siderolabs.com>
As Talos doesn't consume `.machine.install` if already installed, there
is no point in validating it once already installed.
This fixes a problem users often run into: after a reboot/upgrade the
system disk blockdevice name changes, due to the kernel upgrade, or just
unpredictable behavior of device discovery. Talos fails to boot as it
can't validate the machine config, while it's already installed, so
actual blockdevice name doesn't matter.
Signed-off-by: Andrey Smirnov <andrey.smirnov@siderolabs.com>
