I have the following in main.tf
data "kubectl_path_documents" "yaml-files" {
pattern = "${path.module}/manifests/*.yaml"
}
resource "kubectl_manifest" "yaml-manifests" {
for_each = toset(data.kubectl_path_documents.yaml-files.documents)
yaml_body = each.value
}
I would need to parameterize certain fields in the yaml files to be able to deploy different set of resources for dev vs prod. I knew there was a way to do this if it was one yaml. How should this be done for many yaml files
├── manifests
│ ├── gdp-configmap.yaml
│ ├── gdp-agent-deamonset.yaml
│ ├── gdp-collector-configmap.yaml
│ ├── gdp-collector-deployment.yaml
Any help is appreciated.
I was able to resolve this using a null_resource that executes a sh script with values to be sed'ed.
Related
I have the next tree:
root_project/
├── app
│ ├── default_photo_profile.jpg
│ ├── config.py
│ ├── __main__.py # My app are python package, I'm runnig it via "python -m"
│ └── ...
├── tests
│ ├── test_unit.py # import config.py inside
│ ├── functional # import config.py inside
│ ├── pytest.ini
│ └── ...
...
Currently default_photo_profile causing error because tests doesn't have this file.
Reading file in config.py:
DEFAULT_PHOTO_FILE_PATH = Path('default_photo.jpg')
with open(file=DEFAULT_PHOTO_FILE_PATH, mode='rb') as file_obj:
DEFAULT_PHOTO_BYTES = file_obj.read()
How I can solve this?
I tried:
Patch access to default_photo.jpg with fixture - not helped, error during import stage, not executiion.
set flag to pytest comamnd line: --rootdir app - not helped (don't know why).
try/except for reading the file in app.config.py - may help but it's not my intention, I really want raise error if file not found
Put default_photo.jpg inside EVERY test directory - will help bit dirty.
Patch os.path like suggested in https://stackoverflow.com/a/43003192/11277611 - dirty
Include tests into package (move __main__.py into root_project - not sure that it's a good idea (have not enough experience to decide).
Set absolut path to default_photo.jpg - will fail on the production server.
Probably adoptable solutions (What I want):
Set root dir to root_project.app somehow inside pytest.ini to immitate regular execution.
Set root dir to root_project.tests somehow to place file in root of tests and access from any of tests folder.
Try to use following code in config.py:
DEFAULT_PHOTO_FILE_PATH = Path(__file__).parent / 'default_photo.jpg'
with open(file=DEFAULT_PHOTO_FILE_PATH, mode='rb') as file_obj:
DEFAULT_PHOTO_BYTES = file_obj.read()
Is it what you are trying to achieve?
I am creating a helm chart for the job I want to run in our k8 cluster. When you execute helm create it creates templates that I do not need.
$ helm create new-job
Creating new-job
$ tree new-job/
new-job/
├── Chart.yaml
├── charts
├── templates
│ ├── NOTES.txt
│ ├── _helpers.tpl
│ ├── deployment.yaml
│ ├── hpa.yaml
│ ├── ingress.yaml
│ ├── service.yaml
│ ├── serviceaccount.yaml
│ └── tests
│ └── test-connection.yaml
└── values.yaml
3 directories, 10 files
Is there a way to only create a template containing only job.yaml?
Helm has the option to use a custom "starter":
helm create --starter /path/to/starter/chart my-chart
There's nothing magical about helm create. Just so long as the chart has a Chart.yaml file and a templates subdirectory, Helm can install it. You can mkdir an empty directory and create the Chart.yaml file from scratch using your favorite text editor, if you'd prefer.
# Chart.yaml, with only the required fields
apiVersion: v2
name: my-chart
version: '0.0.1'
mkdir my-chart; cd my-chart
$EDITOR Chart.yaml
mkdir templates
$EDITOR templates/job.yaml
helm install --generate-name .
You can also go the other direction; start from the helm create template and delete all of the parts you don't need. Of note, there are a couple of generically helpful things in _helpers.tpl you might want to retain. (Keeping the ServiceAccount turns out to be useful in a couple of situations, and if you use Istio, it also likes you to have a Service even if you don't accept inbound connections.)
my directory structure is
├── xxx
│ ├── 01.md
| └── 02.md
├── auth
│ ├── j1.json
│ ├── j2.json
│ └── j3.json
└── default.template.html
And I link jsons from markdowns like Auth. It makes sense as we use there files as test scenarios and in json files we have credentials and roles. But if I try to generate html it fails on unresolved internal reference: ../auth/aspect_admin.json. I tried to exclude the link checking but without any help. The best would be to leave it as a link in md file but somehow follow the link and include the json as code block in generated html. Is it possible?
It was a bug and will be fixed in next version https://github.com/planet42/Laika/issues/148
Having trouble figuring out how the workflow for using Github as a VuePress site source control and deploying it to Github Pages.
When I ran deploy.sh the first time, it gave me a Github certificate error around the init command and did not initialize a new repo (I already have a repo setup so not sure if the init command in deploy.sh is required. Subsequent runs of deploy.sh resulted in no error.
**Problem:**Unfortunately, when I visit my Github Pages site, its not using VuePress templates.
I feel like I have either:
- The folder structure wrong
- The base set incorrectly in config.js
- The relative folders incorrect in deploy.sh
Can someone put eyes on this and give some feedback? Thank you.
For your reference
Local machine's folder structure:
user#system:~/powerDocs$ tree
.
├── deploy.sh
├── docs
│ └── README.md
├── node_modules
│ └── yarn
│ ├── bin
│ │ ├── yarn
│ │ ├── yarn.cmd
│ │ ├── yarn.js
│ │ ├── yarnpkg
│ │ └── yarnpkg.cmd
│ ├── lib
│ │ ├── cli.js
│ │ └── v8-compile-cache.js
│ ├── LICENSE
│ ├── package.json
│ └── README.md
├── package.json
├── package-lock.json
└── README.md
5 directories, 15 files
Content of deploy.sh:
#!/usr/bin/env sh
# abort on errors
set -e
# build
vuepress build
# navigate into the build output directory
cd docs/.vuepress/dist
# if you are deploying to a custom domain
# echo 'www.example.com' > CNAME
git init
git add -A
git commit -m 'deploy'
# if you are deploying to https://<USERNAME>.github.io
# git push -f git#github.com:SeaDude/SeaDude.github.io.git master
# if you are deploying to https://<USERNAME>.github.io/<REPO>
git push -f git#github.com:SeaDude/powerDocs.git master:gh-pages
cd -
I made deploy.sh executable with chmod +x deploy.sh. Running ./deploy.sh gives me the following output:
user#system:~/powerDocs$ ./deploy.sh
WAIT Extracting site metadata...
[12:05:53 PM] Compiling Client
[12:05:53 PM] Compiling Server
(node:15590) DeprecationWarning: Tapable.plugin is deprecated. Use new API on `.hooks` instead
[12:05:57 PM] Compiled Server in 3s
[12:05:59 PM] Compiled Client in 6s
WAIT Rendering static HTML...
DONE Success! Generated static files in .vuepress/dist.
Reinitialized existing Git repository in /home/powerDocs/docs/.vuepress/dist/.git/
On branch master
nothing to commit, working directory clean
Here is the contents of config.js:
module.exports = {
title: "PowerDocs",
description: "Where functions go to frolic.",
base: "/powerDocs/",
themeConfig: {
nav: [
{ text: "Home", link: "/" }
],
sidebar: [
'/'
]
}
};
Have you checked your dist folder to see what is actually being output? The error makes it seem like there are no files present to commit after the build.
I have the almost identical setup locally and haven't run into this problem with it, the only difference being the command I run to build is yarn docs:build
I would like to fake the LED (character device) of an embedded linux board (BeagleBone Black) in a user space library written in D.
Via the command line the led driver is represented to the user space as "device file" (e.g. for led "USER LEDS D2 0"):
debian#beaglebone:/sys/class/leds/beaglebone:green:usr0$ tree
.
├── brightness
├── device -> ../../../leds
├── invert
├── max_brightness
├── power
│ ├── async
│ ├── autosuspend_delay_ms
│ ├── control
│ ├── runtime_active_kids
│ ├── runtime_active_time
│ ├── runtime_enabled
│ ├── runtime_status
│ ├── runtime_suspended_time
│ └── runtime_usage
├── subsystem -> ../../../../../class/leds
├── trigger
└── uevent
Via the command line single entities like e.g. the brightness (turn on and off) can be written to and read from like files:
cd /sys/class/leds/beaglebone:green:usr0
echo none > trigger
echo 1 > brightness
echo 0 > brightness
I consider it reasonable to abstract the whole device as interface and implement a production class and a test class (hand-made fake object) which implements the interface. The client code of the library could then use either the production code or the fake class which could be injected using dependency injection.
Could a mock framework like https://github.com/QAston/DMocks-revived provide functionality to use a fake driver in client code out of the box instead of using the handmade fake led driver described above? (I guess to fake more complicated character devices it would be necessary to "hand-implement" anyway.)
Or are there better approaches to fake a character device (e.g. make the production class require a root path of the driver when creating an instance via "constructor injection")?