The `scheduler-k3s:charts:set` command only changed one chart override at a time, so matching a chart's stored overrides to a declared set meant reading the report, computing the difference in both directions and issuing one command per added, changed and removed property, with a failure partway through leaving a mixture of the two sets. A `--replace` flag on `scheduler-k3s:charts:set` now takes a chart name followed by `property=value` pairs and writes the declared map for that chart in a single call. Every pair is parsed before anything is written, so a rejected pair leaves the stored overrides untouched, and a pair ending in `=` stores an empty value, which the per-property form cannot express. An empty pair list is rejected rather than treated as a request to remove everything, and the new `scheduler-k3s:charts:clear` command handles that case by removing every override for a chart.
The `scheduler-k3s` metadata commands were all per-key, so matching an app's stored annotations, labels or node sysctls to a declared set meant reading the report, computing the difference in both directions and issuing one command per added, changed and removed key, with a failure partway through leaving a mixture of the two sets. A `--replace` flag on `annotations:set`, `labels:set`, `node-sysctls:set` and `autoscaling-auth:set` now writes the declared map in a single call, taking trailing `key=value` pairs everywhere except `autoscaling-auth:set`, which keeps its repeatable `--metadata` flag. Every pair is parsed before anything is written, so a rejected pair leaves the stored map untouched, and a pair ending in `=` stores an empty value, which the per-key form cannot express. An empty pair list is rejected rather than treated as a request to remove everything, so a generated list that expands to nothing cannot silently drop an app's metadata; `annotations:clear`, `labels:clear` and `node-sysctls:clear` handle that case, taking the same optional filters as their report commands. The `--resource-type` flag and the app name are now validated on the set commands as well, where a typo previously exited zero and wrote a property that no report and no deploy ever read.
Sysctls the kernel does not namespace, such as `vm.max_map_count`, cannot be set from a pod spec and previously had no answer beyond editing `/etc/sysctl.d` on each host by hand. `scheduler-k3s:node-sysctls:set` now applies them through a privileged daemonset, which reaches nodes joined later and reapplies after a reboot. Sysctls may be scoped to a node profile, with a profile scope inheriting the global values and overriding them on conflict so that every node is covered by exactly one daemonset. Clearing a sysctl stops dokku managing it but does not restore the previous value, which persists until the node reboots.
Host-crontab generation for `app.json` cron tasks now lives in the `cron` plugin, gated by a new `scheduler-uses-host-cron` trigger that the `docker-local` scheduler answers true while self-managed schedulers such as `k3s` answer false. This lets any host-cron scheduler participate in normal `app.json` cron without coupling to `scheduler-docker-local` or duplicating the crontab writer, while the `k3s` scheduler continues to manage its own in-cluster cron jobs.
Closes#8862.
The previous commit added the dispatcher case but missed appending
`subcommands/preview` to the plugin's `SUBCOMMANDS` list, so no symlink
was created at install time and dokku reported the command as unknown.
The scheduler-k3s plugin exposes `annotations:set` and `labels:set` but no matching report subcommands, so scripts could not inspect the configured state for idempotent management without reading the property files directly. The new `scheduler-k3s:annotations:report` and `scheduler-k3s:labels:report` surface the configured entries with stdout, JSON, single-flag query, and `--process-type` / `--resource-type` filtering, mirroring the recently added `scheduler-k3s:charts:report`. The literal `GlobalProcessType` value `--global` is rendered as `global` in report keys to avoid leading dashes. The `scheduler-k3s:autoscaling-auth:report` command is updated alongside to loop over every app when no app and no `--global` flag is provided, matching the convention used by `scheduler-k3s:report`.
Helm chart overrides on the scheduler-k3s plugin were previously set
through the generic `scheduler-k3s:set --global chart.<chart>.<property>`
interface, which mixed chart-level configuration into the same command
that manages scheduler properties and offered no focused way to inspect
which overrides were configured. A dedicated `scheduler-k3s:charts:set`
sets and clears chart-specific helm values, and a complementary
`scheduler-k3s:charts:report` surfaces configured overrides per chart
with optional JSON output and single-field flag queries. The legacy form
on `scheduler-k3s:set` continues to work but now emits a deprecation
warning pointing users at the new subcommand.
Moves the actual exec out of the storage plugin and into a new scheduler-storage-exec plugn trigger. scheduler-docker-local does docker run with TTY-aware -it/-i selection and --user derived from entry.Chown; scheduler-k3s creates a throwaway Pod via the kubernetes API, waits for it to reach Running with structured error reporting (ImagePullBackOff and friends are surfaced from the container status verbatim, no kubectl involvement), execs the user command via the existing SPDY plumbing in k8s.go, and deletes the Pod on the way out. (Entry).Validate now accepts either an absolute path or a docker named-volume token for docker-local entries so the migration synthesizer's named-volume legacy entries work cleanly. storage:exec gains --as-user for one-off uid overrides, propagates the underlying tool's exit code via os.Exit, and detects TTY/interactive mode from os.Stdin so non-interactive scripted use no longer trips over docker's input-device-is-not-a-tty error.
Adds a per-entry storage helm chart that owns the PVC and (when host_path is set) the PV, and three triggers for the storage plugin to drive: storage-create installs/upgrades the chart with storage-class existence validation, storage-destroy uninstalls it, storage-status reports the PVC phase. App deployment and cron-job charts now consume storage-app-mounts and reference each PVC by name; the // todo: implement volumes blocks are gone. ProcessVolume gained sub_path, read_only, and persistent_claim fields and the deployment/cron-job templates render them.
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
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.
Different options get applied to each depending on what is needed for the role.
This also updates how we initialize the cluster to better support how metrics exposure and cross-server networking.
This utilizes a Kubernetes Job to start a pod in order to take advantage of completed Job cleanup in Kubernetes. It also properly handles interactivity by ensuring the container has a TTY.
Note that there is currently no provision for parsing docker arguments and translating them into their Kubernetes equivalents.
The big feature missing from deployments today is healthchecks, which require extra parsing of the app.json format since the healthchecks there aren't immediately serializable to kubernetes deployment healthchecks.
Other smaller missing functionality from the existing OSS kubernetes scheduler are:
- The ability to set custom pod annotations
- The ability to set custom deployment annotations
- The ability to mount volumes
Exposing these pieces of functionality will come at a later date, if at all.