This allows using the environment values defined in the environments: section of helmfile.yaml to be used from other sections of the file.
This works by having two template renderers, the first-pass and the second-pass renderer.
The first-pass render renders a helmfile.yaml template with replacing template functions has side-effects with noop. So, use only funcs that don't have side-effects to compose your environment values.
Then the second-pass renderer renders the same helmfile.yaml template, but with the environment values loaded by the first-pass renderer.
The implementation uses a buffer instead of re-reading the file twice.
Resolves#297
This enables `helmfile lint` and `helmfile template` commands to fetch and untar all the required charts concurrently. The concurrency is configurable via the `--concurrency` flag, that defaults to `0`.
Ref #292
helmfile as of today ensures that all the targeted helmfile.yaml to have the specified environment defined in it.
That is, `helmfile --environment prod -f helmfile.d/ sync` fails if any helmfile under `helmfile.d/` is missing the `production` environment.
This changes the validation logic, so that helmfile fails only when all the helmfiles miss the environment.
Resolves#279
Enhance the `diff` functionality to be able to return affected releases that has any changes,
so that the succeeding `sync` can be run against only the affected releases.
This provides us extra idempotency.
Resolves#277
This removes --concurrenty flag from `helmfile template` and `helmfile lint`. YAGNI.
Also separates out the downloadCharts into a function so it can be used by lint too.
Fixes#288
`helmfile template` runs `helm template` over releases within the helmfiles, and provide you a stream of generated yaml documents of Kubernetes resources via stdout.
Resolves#283
Adds `--suppress-secrets` to `helmfile apply` and `helmfile diff`, so that the diff command omits the contents of secrets from its output. This is a security feature that should always be turned on for CI/CD use-cases.
With `--suppress-secrets`, the output when there is any change looks like:
```
Comparing bar stable/grafana
default, baz-grafana, Secret (v1) has changed:
+ Changes suppressed on sensitive content of type Secret
```
Resolves#269
This command syncs releases only if there is any difference between the desired and the current state. It asks for an confirmation by default. Provide `--auto-approve` flag after the `apply` command to skip it.
Resolves#205
`helmfile lint` works with relative chart reference (#252)
The tempalte function `readFile` accepts the path relative to helmfile.yaml
Resolves#246Fixes#252
This is a follow-up for #235.
We setup the default logger only after the flags are parsed successfully.
That's because we init the logger according to the --log-level flag value.
Fixes#233
Output on values render error:
```
err: failed to render [/Users/sstarcher/xxx/xxx/values.yaml], because of template: stringTemplate:10:18: executing "stringTemplate" at <requiredEnv "HELM_AC...>: error calling requiredEnv: required env var `HELM_ACCOUNT` is not set
```
Also removes panic and sets the output as `apps.Run()` can and will return errors. Panic makes no sense.
Changelog:
* improve error handling for value rendering
* only output if error exists
* add exit status
Adds the `--detailed-exitcode` to the `helmfile diff` command to return `1` on failure, and `2` when no error but diff is seen.
This feature requires the latest `helm-diff` containing https://github.com/databus23/helm-diff/pull/78, and `helm` containing https://github.com/helm/helm/pull/4367.
This is verified to work by manually running commands like the followings:
```bash
./helmfile --helm-binary helm211dev -f ./examples/helmfile.d diff --detailed-exitcode; echo $?
./helmfile --helm-binary helm211dev -f ./examples/helmfile.d diff; echo $?
```
Note that, in above example commands, `helm211dev` is a custom `helm` binary that is built from helm's master branch containing [the necessary enhancement to allow propagate non-zero plugin exit code](https://github.com/helm/helm/pull/4367).
Fixes#187
Changelog:
* fix: `helmfile -f dir/ -l k=v sync` fails on first helmfile w/ no match
* fix(ux): present the filename if the selector did not match any releases
```console
$ helmfile -f helmfile.d/ -l foo=bar sync
2018/07/12 23:37:29 specified selector did not match any releases in helmfile.d/00-backend.yaml
2018/07/12 23:37:29 specified selector did not match any releases in helmfile.d/01-frontend.yaml
2018/07/12 23:37:29 err: specified selector did not match any releases in any helmfile
```
Add tillerNamespace and kubeContext to helmDefaults. Also add merging of args capability.
People can start adding additional fields under helmDefaults on top of this. The additional fields are merged to args as long as they are implemented using the framework introduced by this commit.
Changelog:
* adding merging for helm defaults args and --args, adding tillernamspace, kubecontext to helmdefaults, adding getargs to lint command
* adding docs for helmdefaults
* checking if empty default values
Closes#93
Summary of changes:
- Adds the [uber-go-/zap](https://github.com/uber-go/zap) library for leveled logging (69040996e0bcbd2f211d0d8d576147027e928e3b)
- A new CLI arg `--log-level` which can take the values `debug` `info` `warn` `error` `fatal` `panic`. The default is `info`, and the existing `--quiet` flag is treated as `--log-level warn`
- All `helm.exec` calls are preceded by a one-line summary. The current information on the full helm command line arguments `exec: helm exec ... ` is only output at `debug` level. This means sensitive command line arguments which may include passwords and `--set` arguments should be hidden by default.
`helm` may not be version compatible. Add an option to specify an alternative `helm` command on runtime, possibly according to the version of tiller running on your cluster.
The use case is to have a list of helmfile releases version controlled together with all settings and have a CI pipeline that will lint all releases with settings before running sync. The new functionality was mostly copy pasted from the Diff implementation with some extra handling for fetching remote charts.
Notes:
* Added release name to chart path to avoid potential race condition when fetching the chart
**Feature**
An additional sub-command to the helmfile binary; helmfile test
**Why**
Helm provides helm test (https://github.com/kubernetes/helm/blob/master/docs/chart_tests.md) as a method to run automated tests post chart install to ensure things are working as they should be.
It would be nice to be able to run something like helmfile test against a particular helmfile in order to run helm tests against all charts/releases defined in the file. Either as part of the sync (i.e. helmfile sync --test) to be ran immediately after the corresponding chart is installed, or as a separate command ran after a sync (i.e. helmfile test).
A chart without tests will exit with a 0 status, so it can be safely ran against any charts.
**Notes**
`--cleanup` (bool) & `--timeout` (default: 300) are available as first class arguments. Additional arguments can be passed to the helm binary as with other sub commands using `--args=`
Resolves#144
`helmfile delete` has been implying `--purge` but it is not the case since this change.
The new behavior is `helmfile delete --purge` to actually purge releases.
Run just `helmfile delete` to delete releases but not purge them.
Resolves#71
This adds `releases[].labels` in which you can set arbitrary number of key-value pairs, so that commands like `helmfile sync --selector key=value` can be used to run the helmfile subcommand against a subnet of declared releases.
`labels` and `selector` are named as such on purpose of being consistent with terminology of Kubernetes and other tools in the K8S ecosystem, including kubectl, stern, helm, and so on.
Resolves#8
* Allow specifying a default namespace used across all the releases
Omit `namespace` in `charts[].namespace` of charts.yaml and then run `helmfile sync -n yourns` to install your charts into `yourns`.
Resolves#31
* Add option to throttle concurrent `helm repo update` calls
Have added a new flag `--concurrency N` to `helmfile sync charts` that can be used to set a
limit on the number of concurrent calls to helm.
Implementation details:
Switched `SyncCharts` from using a WaitGroup to using a pool
of workers and a queue of jobs. To ensure that this is thread safe and
an attempt is made to sync each chart at the end.
Fixes#23
* Fix formatting and update CI to catch these errors
Have fixed the formatting so that `make pristine` now passes. Have also
added this to the Circle CI config to catch these errors in the future.
- Simple copy-paste for the most part from sync job. I had started down
the path of adding in a meta PluginCommand directive and trying to make it
more modular, but in the end there are some small differences between
the execution that were a bit difficult to model and it just got ugly.
Figured keeping it simple would be easier to manage
KUOKA Yusuke
Dan Helfman
David Genest
Shane Starcher
Takuma Hashimoto
Karl Stoney
rmartinez3
Maciej Bryński
Simon Li
hori-ryota
Johan Lyheden
Mike Eves
Cedric Meury
Daniel
Fabien Silberstein
Alex Withrow
Maximilian Gaß
Thomas O'Donnell
jayste
Justin Nauman
AMeng
rob boll
Nathan Vecchiarelli
rob boll