This project is mirrored from https://github.com/siderolabs/omni.
Pull mirroring updated .
- Jun 25, 2024
-
-
Utku Ozdemir authored
This is the official v0.38.2 release. Signed-off-by: Utku Ozdemir <utku.ozdemir@siderolabs.com>
-
Utku Ozdemir authored
Requests targeting workload proxies were incorrectly detected to be in the old URL format with the `p-` prefix when the instance name had a dash (`-`) in it. Fix it by checking explicitly for the `p-` prefix in the logic. Add a test to cover this case. Signed-off-by: Utku Ozdemir <utku.ozdemir@siderolabs.com> (cherry picked from commit dc7c2b3e)
-
- Jun 21, 2024
-
-
Artem Chernyshev authored
This is the official v0.38.1 release. Signed-off-by: Artem Chernyshev <artem.chernyshev@talos-systems.com>
-
Artem Chernyshev authored
`9999` seems to break things. `auto` looks to be working properly. Signed-off-by: Artem Chernyshev <artem.chernyshev@talos-systems.com> (cherry picked from commit 8b16da39)
-
- Jun 20, 2024
-
-
Artem Chernyshev authored
This is the official v0.38.0 release. Signed-off-by: Artem Chernyshev <artem.chernyshev@talos-systems.com>
-
Artem Chernyshev authored
Sort by the creation time of the `MachineSetNode` instead. As the creation time of the `ClusterMachineTalosVersion` doesn't replicate the creation time of the `MachineSetNode` anymore. Signed-off-by: Artem Chernyshev <artem.chernyshev@talos-systems.com>
-
- Jun 19, 2024
-
-
Artem Chernyshev authored
Otherwise it might be hidden under the other elements. Signed-off-by: Artem Chernyshev <artem.chernyshev@talos-systems.com>
-
Artem Chernyshev authored
It was dropped, unfortunately linter didn't catch it. Signed-off-by: Artem Chernyshev <artem.chernyshev@talos-systems.com>
-
- Jun 18, 2024
-
-
Artem Chernyshev authored
This is the official v0.38.0-beta.0 release. Signed-off-by: Artem Chernyshev <artem.chernyshev@talos-systems.com>
-
Artem Chernyshev authored
Filter them in the same way as in manual allocation: available, connected and reporting events. Signed-off-by: Artem Chernyshev <artem.chernyshev@talos-systems.com>
-
Artem Chernyshev authored
This change fixes race during machine allocation in the machine class based machine sets, which was happening due to the eventual consistency of the in-memory cache. Signed-off-by: Artem Chernyshev <artem.chernyshev@talos-systems.com>
-
Artem Chernyshev authored
Fixes: https://github.com/siderolabs/omni/issues/380 Signed-off-by: Artem Chernyshev <artem.chernyshev@talos-systems.com>
-
Artem Chernyshev authored
Fixes: https://github.com/siderolabs/omni/issues/378 That should fix the migration from the vanilla Talos images which do not exist in the image factory. Signed-off-by: Artem Chernyshev <artem.chernyshev@talos-systems.com>
-
Grzegorz Rozniecki authored
Add support for Akamai Connected Cloud (Linode) installation media in Omni. Signed-off-by: Grzegorz Rozniecki <grozniec@akamai.com>
-
- Jun 14, 2024
-
-
Artem Chernyshev authored
Do not exit the loop, but just ignore such machines. Otherwise we don't create `ClusterMachineTalosVersion` resources for all new machines in the machine set, which blocks `ClusterMachineConfigController` and the machines are not configured. Signed-off-by: Artem Chernyshev <artem.chernyshev@talos-systems.com>
-
Artem Chernyshev authored
The return type should be input type, not the output type. Also leverage runner equality functions to not to restart the collectors if nothing changed. Signed-off-by: Artem Chernyshev <artem.chernyshev@talos-systems.com>
-
Utku Ozdemir authored
Rework the workload service proxying feature to support the new domain format in addition to the existing one. Old format: ``` p-g3a4ana-demo.omni.siderolabs.io ``` New format: ``` g3a4ana-demo.proxy-us.omni.siderolabs.io ``` The old format required a new DNS records to be added for each new workload service, causing issues with the resolution on clients. The new format addresses it by leveraging wildcard records. Additionally, build the full exposed service URL on the backend and make it a field on `ExposedService` resource, so they can be accessed using `omnictl get exposedservice`. Part of siderolabs/omni#17. Signed-off-by: Utku Ozdemir <utku.ozdemir@siderolabs.com>
-
- Jun 13, 2024
-
-
Artem Chernyshev authored
Safety net against any bugs there. Signed-off-by: Artem Chernyshev <artem.chernyshev@talos-systems.com>
-
- Jun 12, 2024
-
-
Artem Chernyshev authored
Should fix nil reference errors. Signed-off-by: Artem Chernyshev <artem.chernyshev@talos-systems.com>
-
Artem Chernyshev authored
Fixes: https://github.com/siderolabs/omni/issues/33 It is now possible to get full access `kubeconfig` and `talosconfig` (operator role), if the Omni instance has `enable-break-glass-configs` flag enabled. They can be downloaded using cli commands: `omnictl kubeconfig --admin --cluster <name>` `omnictl talosconfig --admin --cluster <name>` After you download the config the cluster will be marked with `omni.sidero.dev/tainted` annotation to keep in mind that this cluster has weaker security and might need to get secrets rotation in the future. Signed-off-by: Artem Chernyshev <artem.chernyshev@talos-systems.com>
-
- Jun 11, 2024
-
-
Artem Chernyshev authored
Accidentally hardcoded `8090` there
Fixes: https://github.com/siderolabs/omni/issues/310 Signed-off-by: Artem Chernyshev <artem.chernyshev@talos-systems.com> -
Artem Chernyshev authored
`migrateInstallImageConfigIntoGenOptions` was missing that check. Add the check and add the test. Signed-off-by: Artem Chernyshev <artem.chernyshev@talos-systems.com>
-
Utku Ozdemir authored
Validate the user email when creating a new identity. Closes siderolabs/omni#312. Signed-off-by: Utku Ozdemir <utku.ozdemir@siderolabs.com>
-
Artem Chernyshev authored
Looks like my assumption wasn't correct. Signed-off-by: Artem Chernyshev <artem.chernyshev@talos-systems.com>
-
- Jun 10, 2024
-
-
Utku Ozdemir authored
Switch from the old machine log storage approach to using the new persistence feature in the go-circular: https://github.com/siderolabs/go-circular/pull/4 Implement a migration code to migrate from old format to the new format when a machine log buffer is initialized. Move some hardcoded log buffer settings into config (CLI args). Use a default jitter of 0.1 (10+-1 minutes). Rework the CLI arg names to be consistent. Closes siderolabs/omni#118. Signed-off-by: Utku Ozdemir <utku.ozdemir@siderolabs.com>
-
Artem Chernyshev authored
Changes: https://github.com/cosi-project/runtime/pull/457 Signed-off-by: Artem Chernyshev <artem.chernyshev@talos-systems.com>
-
Artem Chernyshev authored
Display pending machine config updates count for the `MachineSet` and mark each machine which is locked and has the config update which is not applied yet. Fixes: https://github.com/siderolabs/omni/issues/15 Signed-off-by: Artem Chernyshev <artem.chernyshev@talos-systems.com>
-
Artem Chernyshev authored
Otherwise the controller might get stuck trying to destroy the resource forever. Signed-off-by: Artem Chernyshev <artem.chernyshev@talos-systems.com>
-
Dmitriy Matrenichev authored
I have run it with `–count=100` and it no longer fails. The https://github.com/golang/go/issues/67434 can't come soon enough, so we can drop all fake clocks. Signed-off-by: Dmitriy Matrenichev <dmitry.matrenichev@siderolabs.com>
-
- Jun 08, 2024
-
-
Dmitriy Matrenichev authored
Fixes [CVE](https://scout.docker.com/v/CVE-2024-3727?s=github&n=image&ns=github.com%2Fcontainers&t=golang&vr=%3C5.30.1 ) which doesn't affect us, but tools complain about it nevertheless. Signed-off-by: Dmitriy Matrenichev <dmitry.matrenichev@siderolabs.com>
-
- Jun 07, 2024
-
-
Dmitriy Matrenichev authored
With new gRPC (both gateway and modules) it uses `grpc.NewClient` call to create clients. It no longer support custom addresses without a `passthrough:` prefix. Previous fix didn't account for that in some places, so this one changes the structure of `Transport` to always return address in proper form for external users. Signed-off-by: Dmitriy Matrenichev <dmitry.matrenichev@siderolabs.com>
-
Artem Chernyshev authored
Omni now allows using `--grant-type=authcode-keyboard` option in `oidc-login`. That allows to log in without forwarding `:8000` port from the remote machine. Instead kubelogin will present a link and ask for the one time code. Clicking the link will open Omni web page with one time code. Fixes: https://github.com/siderolabs/omni/issues/111 Signed-off-by: Artem Chernyshev <artem.chernyshev@talos-systems.com>
-
Artem Chernyshev authored
The code had an issue: if the loop pulled at least one image and one of the images in the end of the loop already exist on the node, it was skipping the update, so only the outdated event was sent to the channel. Signed-off-by: Artem Chernyshev <artem.chernyshev@talos-systems.com>
-
Utku Ozdemir authored
Fix the flaky test due to the race caused by the test ending before the last log, `task finished`, is written by the task runner. Signed-off-by: Utku Ozdemir <utku.ozdemir@siderolabs.com>
-
Utku Ozdemir authored
The test `TestSchematicChanges` was flaky, as it was doing the following: 1. Trigger one or more `factory.talos.dev` build image generations & talos node upgrades by modifying resources 2. Capture these upgrade calls 3. Assert the build image in these calls (to be one or more with, always the expected image) 4. Clear captured upgrade calls 5. Trigger one or more `ghcr.io` build image generations & talos node upgrades by modifying resources 6. Capture these upgrade calls 7. Assert the build image in these new calls (to be one or more with, always the expected image) The problem was, between step 4 and 7, an old `factory.talos.dev` build image could sometimes still be captured due to the nature of controllers asynchronously reconciling resources. Replace this logic by not clearing the upgrade requests at all, instead, simply ignoring the previous image's preceding captured requests in step 7. Signed-off-by: Utku Ozdemir <utku.ozdemir@siderolabs.com>
-
Dmitriy Matrenichev authored
https://github.com/siderolabs/omni/actions/runs/9412691342/job/25928064700?pr=336 failed for some reason but output like: ``` #63 60.86 Error Trace: /src/internal/backend/runtime/omni/controllers/omni/machine_set_status_test.go:652 #63 60.86 /src/internal/backend/runtime/omni/controllers/omni/machine_set_status_test.go:661 #63 60.86 Error: unexpected event #63 60.86 Test: TestMachineSetStatusSuite/TestConfigUpdateWithMaxParallelism #63 60.86 Messages: got event {event:{Resource:0xc0023486e0 Old:0xc0016bfe40 Error:<nil> Type:1}} ``` Is not saying much, so lets format event output. Also update github.com/cosi-project/runtime in client module, so editors aside from VS Code do not go crazy. Signed-off-by: Dmitriy Matrenichev <dmitry.matrenichev@siderolabs.com>
-
- Jun 06, 2024
-
-
Utku Ozdemir authored
We were not correctly checking if the nodes passed via `talosctl --nodes` were resolved before replacing the target in the GRPC metadata with the resolved ones. We were handling a single node in the metadata correctly, but not doing the same handing for multiple nodes. This PR fixes that. This PR simplifies that logic. Signed-off-by: Utku Ozdemir <utku.ozdemir@siderolabs.com>
-
Artem Chernyshev authored
Dropped some dependencies which no longer worked: - monaco-editor-vue3 - replaced with a tiny wrapper. - polyfill for streaming classes - replaced with vite plugin for polyfills. - buffer - replaced base64 encoding with calls from the `fetch.pb.ts`. Dropped the code that sets up monaco in `main.ts` in favor of Vite monaco plugin that does that for you. Updated `monaco-yaml` as the old version didn't work with Vite. Switched from `Jest` to `bun:test`. Use TypeScript in the tailwind config, as it imports `colors.ts` now. Had to use TypeScript for `colors` as Vite can't import CommonJS modules. Add eslint linter to the `lint` dependencies, enable a bit more linters and fix all detected errors. Fixes: https://github.com/siderolabs/omni/issues/294 Signed-off-by: Artem Chernyshev <artem.chernyshev@talos-systems.com>
-
Artem Chernyshev authored
When you open `cluster/machine/patches` it should go back to `cluster/machine/patches` after you finished editing. The bug was sending you to `machine/patches` out-of-cluster version of it. Signed-off-by: Artem Chernyshev <artem.chernyshev@talos-systems.com>
-
Artem Chernyshev authored
Looks like setting headers as `Record<string, string>` doesn't work. Should use `new Headers()` instead. Signed-off-by: Artem Chernyshev <artem.chernyshev@talos-systems.com>
-