Commit Graph
153 Commits
Author SHA1 Message Date
Jose Diaz-Gonzalez 6da9be2d65 refactor: use chart values directly instead of attempting to template out cron-job and keda files 2025-07-04 04:24:09 -04:00
Jose Diaz-Gonzalez 4ae9cca6f1 feat: implement app-json-is-valid plugin trigger 2025-06-19 02:20:02 -04:00
Jose Diaz-Gonzalez 2dea993804 fix: also stop k8s cron tasks when calling ps:stop 2025-06-10 21:45:08 -04:00
Jose Diaz-Gonzalez 0f4b5d4305 fix: continue including maintenance mode cron tasks in the cron:list subcommand 2025-06-08 22:34:47 -04:00
Jose Diaz-Gonzalez 65dd4bc1e0 feat: set shm-size volume/volumeMounts on kubernetes deployments 2025-03-06 22:59:35 -05:00
Tasos Maschalidis 63182e80e0 Use common.ToBool to check DOKKU_FORCE_TTY 2024-12-17 19:43:16 +02:00
Tasos Maschalidis f786c8c0bc Fix k3s run:detached not returning pod name in a similar fashion to docker-local (as described in docs) 2024-12-17 19:11:01 +02:00
Tasos Maschalidis 838beb4db3 Take DOKKU_FORCE_TTY setting into account for run commands with k3s-scheduler 2024-12-17 19:11:01 +02:00
Jose Diaz-Gonzalez ed0ba7072f refactor: check to see if a deployment-specific image pull secret exists before utilizing it. 2024-12-11 09:23:44 -05:00
Tasos Maschalidis aa51c37774 Properly resolve imagePullSecrets from app and deploymentID if relevant property does not exist 2024-12-10 14:29:38 +02:00
Jose Diaz-Gonzalez b98863b6a7 feat: add support for the keda-http-add-on
Externally, it is treated the same as any other scaling trigger, but requires that the chart be installed in order for it to be supported.
2024-11-13 03:29:26 -05:00
Jose Diaz-Gonzalez fd2cfe8080 fix: correctly pull the deployment id from an app's helm values when executing dokku run under the k3s scheduler
- Fix the internal identifier of deployment_id -> deployment_id
- Add tests for both enter and run.
2024-11-04 14:44:25 -05:00
Jose Diaz-Gonzalez 50244bf8ea chore: drop the pod name from stdout output
This adds unexpected output to the command, making parsing more difficult.
2024-11-04 14:42:53 -05:00
Jose Diaz-Gonzalez ec41dd2e70 refactor: output logs for completed containers
Previously, a container that exited immediately would just show an error message, which is probably not what we want.

If the pod has exited, we will display the logs on stdout for the user and then exit appropriately.

This change also refactors how log streaming works so that we respect exiting when the stream has no content (such as when the container has exited). Previously, the container would just hang around forever due to not being properly canceled. Due to this refactor, the prefix is now split out into a separate writer.
2024-11-04 14:41:42 -05:00
Jose Diaz-Gonzalez adffab13d9 fix: correctly wait for the pod to be ready 2024-11-04 14:38:06 -05:00
Jose Diaz-Gonzalez c3a573229d refactor: execute one-off commands with DOKKU_APP_SHELL as the initial command
This commit changes the launched pod to use the configured DOKKU_APP_SHELL - default /bin/bash - to launch processes. Without this, if a non-interactive process was executed and did not immediately exit, it was possible for that process to launch twice - once when the pod started, and once when entering the pod. Ideally we know whether the process is interactive or not, but this isn't always possible when running under a tty (non-tty == non-interactive). As such, this also removes the ability to launch one-off containers that do not have a configured shell in the container, but this is better than attempting to run commands twice.
2024-11-04 14:29:34 -05:00
Tasos Maschalidis 774fb546ef Remove prefixing releaseName with 'dokku-' in UninstallChart and GetValues calls in scheduler-k3s/triggers.go 2024-10-30 11:47:15 +02:00
Jose Diaz-Gonzalez 90e8cac12f feat: implement nginx:show-config for k3s plugin
This allows users to see what nginx configs are in use for their applications.

Closes #7291
2024-10-21 22:09:09 -04:00
Jose Diaz-Gonzalez 51b951643b fix: respect the exit code and pass flags correctly when tailing nginx logs on k3s 2024-10-21 21:51:37 -04:00
Jose Diaz-Gonzalez 526ea43ae4 fix: tail the correct container in the ingress-nginx pod 2024-10-21 20:37:03 -04:00
Jose Diaz-Gonzalez 611006af73 feat: add support for non-local nginx implementations
If the scheduler is not docker-local, we will assume that a separate plugin trigger will pull access/error logs on our behalf. For k3s, the naive implementation is to pull logs from a single container (multiple ingress-nginx pods will be supported in a future release).

Closes #7268
2024-10-17 00:18:24 -04:00
Jose Diaz-Gonzalez 4397533fd7 fix: implement missing scheduler-app-status trigger in k3s scheduler plugin 2024-04-04 22:14:39 -04:00
Jose Diaz-Gonzalez bb5ab0a777 refactor: use CallPlugnTrigger instead of PlugnTrigger
Refs #6422
2024-03-14 03:11:37 -04:00
Jose Diaz-Gonzalez 68a280e0b6 Merge pull request #6712 from dokku/6422-remove-plugin-trigger-output
Use CallPlugnTrigger instead of PlugnTriggerOutput
2024-03-14 01:53:38 -04:00
Jose Diaz-Gonzalez cdff4ae867 refactor: use CallPlugnTrigger instead of PlugnTriggerOutput
Refs #6422
2024-03-14 01:18:28 -04:00
Jose Diaz-Gonzalez f34fd277b3 refactor: use CallPlugnTrigger instead of PlugnTriggerOutputAsString
Refs #6422
2024-03-14 00:46:55 -04:00
Jose Diaz-Gonzalez d2b167d6ad Merge pull request #6681 from dokku/k3s-ingress-per-domain
Migrate to single ingress per app/domain combination
2024-03-12 22:22:45 -04:00
Jose Diaz-Gonzalez c840fc5f9e feat: wait until helm charts are installed before marking a deploy as failed or completed
This ensures applications are actually up vs giving a false sense of security to users about the state of their applications.
2024-03-12 08:39:22 -04:00
Jose Diaz-Gonzalez 77d9594c09 fix: use annotation to check for ingress method 2024-03-12 07:57:20 -04:00
Jose Diaz-Gonzalez d4b484a33a fix: replace existing port-map ingress objects with domain ingress objects
Domains cannot be reused across ingress objects, and port maps are a function of a domain, so the inversion here makes most sense (at the cost of some downtime when migrating.
2024-03-12 03:55:40 -04:00
Jose Diaz-Gonzalez c0f01c58d6 feat: add initial support for interacting with ClusterTriggerAuthentication objects
These will be provisioned to the keda namespace and referenced by Triggers if they exist (as a fallback to app-specific values).
2024-03-06 03:39:11 -05:00
Jose Diaz-Gonzalez f23cdffe49 feat: associate matching authentication info to a given autoscaler 2024-03-06 03:39:11 -05:00
Jose Diaz-Gonzalez 042495bcd6 fix: ensure keda auth secrets are written out 2024-03-06 03:39:11 -05:00
Jose Diaz-Gonzalez 4ddd341184 feat: add initial support for keda authentication 2024-03-06 03:39:11 -05:00
Jose Diaz-Gonzalez b7ff696507 feat: add initial support for injecting keda addons 2024-03-06 03:39:11 -05:00
Jose Diaz-Gonzalez 5e53207009 fix: ensure we pass global labels 2024-03-01 17:42:24 -05:00
Jose Diaz-Gonzalez d8946edf91 feat: add support for manipulating labels
Closes #6625
2024-03-01 17:42:24 -05:00
Jose Diaz-Gonzalez 7a5bd7ee0f refactor: use a helper function for printing out both labels and annotations 2024-03-01 17:42:24 -05:00
Jose Diaz-Gonzalez 377cfc7f59 refactor: only apply the cluster issuers helm chart when setting properties
Otherwise, we're constantly re-installing the same helm chart during deploys, leading to longer deploys.
2024-02-28 17:16:32 -05:00
Jose Diaz-Gonzalez 7c376d3e6d refactor: rename image pull secrets to standardize on secret naming pattern 2024-02-27 18:25:16 -05:00
Jose Diaz-Gonzalez ab674072bc fix: use smaller reference to image pull secrets in secret naming
This fixes an issue where the secret name is too long in tests where the generated test app name includes a uuid.
2024-02-25 14:00:28 -05:00
Jose Diaz-Gonzalez 8670dbab78 fix: ignore case where k3s isn't installed anymore when stopping services 2024-02-22 07:54:32 -05:00
Jose Diaz-Gonzalez 23fe19faaa fix: bypass deletion error if k3s is the configured scheduler and it isn't installed 2024-02-22 07:42:15 -05:00
Jose Diaz-Gonzalez 72067fcfd2 feat: add ability for users to specify alternative kubeconfig and kubecontext
This will provide the possibility for users to talk to existing kubernetes clusters, thereby removing one of the biggest reasons for having the old scheduler-kubernetes plugin around.
2024-02-22 07:42:15 -05:00
Jose Diaz-Gonzalez 3c6cc68086 refactor: capture output by default
This is almost certainly the correct default for Dokku. While it's a BC break and might cause an increase in memory usage, the api is mostly internal and therefore this is safe to use.
2024-02-12 20:28:31 -05:00
Jose Diaz-Gonzalez 047a3c00f5 chore: remove default false values when calling CallExecCommand 2024-02-12 20:07:43 -05:00
Jose Diaz-Gonzalez 1ae2df52f5 fix: use image pull secrets instead of registries.yaml to reference private repositories
While this will now more or less require a local docker login - which can be overridden by setting the image-pull-secrets property - this ensures every deploy has a valid context.

One annoying thing is that the secret is per-app vs global, meaning that if the secret must be rotated, all apps need to be redeployed. A future change can fix that by deploying a global secret, or if folks really desire, this can be done by setting the image-pull-secrets property on the app/globally.
2024-02-12 17:15:37 -05:00
Jose Diaz-Gonzalez abbc13cb85 fix: ensure the ingress class is properly set for certificate usage
Without this, switching to nginx ingress fails as the cluster issuer isn't properly configured
2024-02-12 01:31:39 -05:00
Jose Diaz-Gonzalez 433eb26f3c feat: always skip the https ingress/ingress-route/deployment port entry if there is a corresponding http entry
Also inject the corresponding entries as necessary
2024-02-12 01:31:39 -05:00
Jose Diaz-Gonzalez bfccf60a37 refactor: simplify tls handling by always redirecting 2024-02-12 01:31:39 -05:00