I am taking over a k3s cluster and am trying to solve a problem with this:
apiVersion: v1
kind: Secret
metadata:
name: clouddesktop-prod-secrets
namespace: clouddesktop-prod
data:
tskey: dHNrZXkta0drTWVWNkNOVFJMLVNlRkZKVFFRalM3RDgzRllvVkxCTQ==
... is used along with with this snippet in my deployment.yaml
- name: AUTHKEY
valueFrom:
secretKeyRef:
name: clouddesktop-prod-secrets
key: tskey
If I am understanding it correctly, the value under tskey will be "decrypted" and then made available as an envirnment variable called ENV_AUTHKEY.
In other words the decryption process that k3s applies will convert the encrypted value "dHNrZXkta0drTWVWNkNOVFJMLVNlRkZKVFFRalM3RDgzRllvVkxCTQ==" into a plaintext value, e.g. "tskey-abc145f" and makes it available to the runnning container as environment variable "ENV_AUTHKEY"
I have verified that indeed an environment variable called "ENV_AUTHKEY" is created, in other words that k3s appends ENV_ to the name.
But as far as I can tell the plaintext is wrong. The environment variable "ENV_AUTHKEY" is indeed created, but, it seems to not have the expected value.
Now by the documentation left for me by my predecessor, I am to create the encrypted value with this simple step:
echo -n "tskey-abc145f" | base64
So I am using base64 as the "encryption" expecting that this is what k3s expects. But the eventually decrypted value appears to me to be incorrect.
What I am trying to determine is what k3s will use to decrypt my encrypted value.
To view how kubernetes would encode the input string, you "may" run the following:
echo -n "tskey-abc145f" | base64
dHNrZXktYWJjMTQ1Zg==
In the following example, I am creating a secret from the command line and providing the key and value in plain text format.
kubectl create secret generic clouddesktop-prod-secrets -n clouddesktop-prod --from-literal=tskey=tskey-abc145f
Verify the secret and its encoded value; here, you can notice the value is encoded automatically. However, when you create a secret using a manifest file, you would have to provided base64 encoded string.
kubectl get secret clouddesktop-prod-secrets -n clouddesktop-prod
apiVersion: v1
kind: Secret
metadata:
name: clouddesktop-prod-secrets
namespace: clouddesktop-prod
data:
tskey: dHNrZXktYWJjMTQ1Zg==
Validate that the encoded value present in the secret is decoding back to the original string(tskey-abc145f):
kubectl get secrets -n clouddesktop-prod clouddesktop-prod-secrets -o go-template='{{.data.tskey|base64decode}}'
tskey-abc145f
Tested on k3s:
k3s --version
k3s version v1.23.3+k3s1 (5fb370e5)
go version go1.17.5
I have a configmap.yaml file as below :
apiVersion: v1
kind: ConfigMap
metadata:
name: abc
namespace: monitoring
labels:
app: abc
version: 0.17.0
data:
application.yml: |-
myjava:
security:
enabled: true
abc:
server:
access-log:
enabled: ${myvar}. ## this is not working
"myvar" value is available in pod as shell environment variable from secretkeyref field in deployment file.
Now I want to replace myvar shell environment variable in configmap above i.e before application.yml file is available in pod it should have replaced myvar value. which is not working i tried ${myvar} and $(myvar) and "#{ENV['myvar']}"
Is that possible in kubernetes configmap to reference with in data section pod's environment variable if yes how or should i need to write a script to replace with sed -i application.yml etc.
Is that possible in kubernetes configmap to reference with in data section pod's environment variable
That's not possible. A ConfigMap is not associated with a particular pod, so there's no way to perform the sort of variable substitution you're asking about. You would need to implement this logic inside your containers (fetch the ConfigMap, perform variable substitution yourself, then consume the data).
I created K8S cluster using Terraform and I also created CRD for Crunchydata Postgres Operator
I obtained CRD for Postgres cluster creation from this link
Terraform script looks like below (tailored output)
resource "kubectl_manifest" "pgocluster" {
yaml_body = <<YAML
apiVersion: crunchydata.com/v1
kind: Pgcluster
metadata:
annotations:
current-primary: ${var.pgo_cluster_name}
labels:
crunchy-pgha-scope: ${var.pgo_cluster_name}
deployment-name: ${var.pgo_cluster_name}
name: ${var.pgo_cluster_name}
pg-cluster: ${var.pgo_cluster_name}
pgo-version: 4.6.2
pgouser: admin
name: ${var.pgo_cluster_name}
namespace: ${var.cluster_namespace}
YAML
}
But when I execute 'terraform apply' it errored as
Error: pgo/UserGrp failed to create kubernetes rest client for update of resource: resource [crunchydata.com/v1/Pgcluster] isn't valid for cluster, check the APIVersion and Kind fields are valid
However, according to the official link mentioned above following should work
apiVersion: crunchydata.com/v1
kind: Pgcluster
I am not sure whether it's issue with Terraform or link was not updated correctly
Kindly let me know what should be changed / done to fix this issue as I am stuck with this issue
Finally, I figured out the issue and the issue was pgo_cluster_name was not given in lowercase
I was able to get the following error only when I executed the target individually ie terraform apply --target=<target_name>
Error: pgo/UserGrp failed to run apply: error when creating "/tmp/773985147kubectl_manifest.yaml": Pgcluster.crunchydata.com "UserGrp" is invalid: metadata.name: Invalid value: "UserGrp": a lowercase RFC 1123 subdomain must consist of lower case alphanumeric characters, '-' or '.', and must start and end with an alphanumeric character (e.g. 'example.com', regex used for validation is '[a-z0-9]([-a-z0-9]*[a-z0-9])?(\.[a-z0-9]([-a-z0-9]*[a-z0-9])?)*')
I set pgo_cluster_name=UsrGrp instead of pgo_cluster_name=usrgrp
I see here a syntax like this:
kubectl create cm configmap4 --from-file=special=config4.txt
I did not find a description of what repetition of = and the special means here.
Kubernetes documentation here only denotes one time usage of = after --from-file while creating configmaps in kubectl.
It appears from generating the YAML that this middle key mean all the keys that are being loaded from the file to be nested inside the mentioned key (special keyword in the question example).
It appears like this:
apiVersion: v1
data:
special: |
var3=val3
var4=val4
kind: ConfigMap
metadata:
creationTimestamp: "2019-06-01T08:20:15Z"
name: configmap4
namespace: default
resourceVersion: "123320"
selfLink: /api/v1/namespaces/default/configmaps/configmap4
uid: 1582b155-8446-11e9-87b7-0800277f619d
kubectl create configmap my-config --from-file=path/to/bar
When creating a configmap based on a file, the key will default to the basename of the file, and the value will default to the file content. If the basename is an invalid key, you may specify an alternate key.
Create a new configmap named my-config with specified keys instead of file basenames on disk
kubectl create configmap my-config --from-file=key1=/path/to/bar/file1.txt --from-file=key2=/path/to/bar/file2.txt
I want to store files in Kubernetes Secrets but I haven't found how to do it using a yaml file.
I've been able to make it using the cli with kubectl:
kubectl create secret generic some-secret --from-file=secret1.txt=secrets/secret1.txt
But when I try something similar in a yaml:
apiVersion: v1
kind: Secret
metadata:
name: some-secret
type: Opaque
data:
secret1.txt: secrets/secret1.txt
I´ve got this error:
[pos 73]: json: error decoding base64 binary 'assets/elasticsearch.yml': illegal base64 data at input byte 20
I'm following this guide http://kubernetes.io/docs/user-guide/secrets/. It explains how to create a secret using a yaml but not how to create a secret from a file using yaml.
Is it possible? If so, how can I do it?
As answered on previous post, we need to provide the certificate/key encoded as based64 to the file.
Here is generic example for a certiticate (in this case SSL):
The secret.yml.tmpl:
apiVersion: v1
kind: Secret
metadata:
name: test-secret
namespace: default
type: Opaque
data:
server.crt: SERVER_CRT
server.key: SERVER_KEY
Pre-process the file to include the certificate/key:
sed "s/SERVER_CRT/`cat server.crt|base64 -w0`/g" secret.yml.tmpl | \
sed "s/SERVER_KEY/`cat server.key|base64 -w0`/g" | \
kubectl apply -f -
Note that the certificate/key are encoded using base64 without whitespaces (-w0).
For the TLS can be simply:
kubectl create secret tls test-secret-tls --cert=server.crt --key=server.key
You can use --dry-run flag to prepare YAML that contains data from your files.
kubectl create secret generic jwt-certificates --from-file=jwt-public.cer --from-file=jwt-private.pfx --dry-run=true --output=yaml > jwt-secrets.yaml
Edit
Thanks to #Leopd for comment about API deprecation, new kubectl uses this command:
kubectl create secret generic jwt-certificates --from-file=jwt-public.cer --from-file=jwt-private.pfx --dry-run=client --output=yaml > jwt-secrets.yaml
On my machine I still have old kubectl version
When using the CLI format basically you're using a generator of the yaml before posting it to the server-side.
Since Kubernetes is client-server app with REST API in between, and the actions need to be atomic, the posted YAML needs to contain the content of the file, and best way to do that is by embedding it as a base64 format in-line. It would be nice if the file could be otherwise embedded (indentation maybe could be used to create the boundaries of the file), but I haven't seen any example of such until now.
That being said, putting a file reference on the yaml is not possible, there is no pre-flight rendering of the yaml to include the content.
So I just learned a super useful k8s fundamental I missed, and then discovered it has a security vulnerability associated with it, and came up with a resolution.
TLDR:
You can have cleartext multiline strings/textfiles as secret.yaml's in your secret repo !!! :)
(Note I recommend storing this in Hashicorp Vault, you can store versioned config files that have secrets, and easily view/edit them through the vault webpage, and unlike a git repo, you can have fine grain access control, pipelines can use the REST API to pull updated secrets which makes password rotation mad easy too.)
cleartext-appsettings-secret.yaml
appsettings.Dummy.json is the default file name (key of the secret)
(I use the word default file name as you could override it in the yaml mount)
and the clear text json code is the file contents (value of the secret)
apiVersion: v1
kind: Secret
metadata:
name: appsettings
namespace: api
type: Opaque
stringData:
appsettings.Dummy.json: |-
{
"Dummy": {
"Placeholder": {
"Password": "blank"
}
}
}
When I
kubectl apply -f cleartext-appsettings-secret.yaml
kubectl get secret appsettings -n=api -o yaml
The secret shows up cleartext in the annotation...
apiVersion: v1
data:
appsettings.Dummy.json: ewogICJEdW1teSI6IHsKICAgICJQbGFjZWhvbGRlciI6IHsKICAgICAgIlBhc3N3b3JkIjogImJsYW5rIgogICAgfQogIH0KfQ==
kind: Secret
metadata:
annotations:
kubectl.kubernetes.io/last-applied-configuration: |
{"apiVersion":"v1","kind":"Secret","metadata":{"annotations":{},"name":"appsettings","namespace":"api"},"stringData":{"appsettings.Dummy.json":"{\n \"Dummy\": {\n \"Placeholder\": {\n \"Password\": \"blank\"\n }\n }\n}"},"type":"Opaque"}
creationTimestamp: 2019-01-31T02:50:16Z
name: appsettings
namespace: api
resourceVersion: "4909"
selfLink: /api/v1/namespaces/api/secrets/appsettings
uid: f0629027-2502-11e9-9375-6eb4e0983acc
Apparently the yaml used to create the secret showing up in the annotation is expected behavior for kubectl apply -f secret.yaml since 2016/has been posted as a bug report, but issue closed without resolution/they're ignoring it vs fixing it.
If you're original secret.yaml is base64'd the annotation will at least be base64'd but in this scenario it's straight up non-base64'd human readable clear text.
Note1: it doesn't happen with imperative secret creation
kubectl create secret generic appsettings --from-file appsettings.Dummy.json --namespace=api
Note2: Another reason for favoring the declarative appsettings-secret.yaml, is that when it's time to edit kubectl apply -f will configure the secret, but if you run that create command it'll say error already exists and you'll have to delete it, before it'll let you run the create command again.
Note3: A reason for kubectl create secret generic name --from-file file --namespace / a reason against secret.yaml is that kubectl show secret won't show you the last time the secret got edited. Where as with the create command, because you have to delete it before you can recreate it, you'll know when it was last edited based on how long it's existed for, so that's good for audit trial. (But there's better ways of auditing)
kubectl apply -f cleartext-appsettings-secret.yaml
kubectl annotate secret appsettings -n=api kubectl.kubernetes.io/last-applied-configuration-
kubectl get secret appsettings -n=api -o yaml
Counteracts the leak
apiVersion: v1
data:
appsettings.Dummy.json: ewogICJEdW1teSI6IHsKICAgICJQbGFjZWhvbGRlciI6IHsKICAgICAgIlBhc3N3b3JkIjogImJsYW5rIgogICAgfQogIH0KfQ==
kind: Secret
metadata:
creationTimestamp: 2019-01-31T03:06:55Z
name: appsettings
namespace: api
resourceVersion: "6040"
selfLink: /api/v1/namespaces/api/secrets/appsettings
uid: 43f1b81c-2505-11e9-9375-6eb4e0983acc
type: Opaque
You can use secode to replace secret values with base64 encoded strings, by simply doing:
secode secrets.yaml > secrets_base64.yaml
It encodes all data fields and works with multiple secrets (kind:Secret) per yaml file, when defined in a list (kind: List).
Disclaimer: I'm the author
For the Windows users in the room, use this for each of the .cer and .key (example shows the .key being encoded for insertion in to the YAML file):
$Content = Get-Content -Raw -Path C:\ssl-cert-decrypted.key
[Convert]::ToBase64String([System.Text.Encoding]::UTF8.GetBytes($Content)) | Out-File -FilePath C:\ssl-cert-decrypted.key.b64
Open the new .b64 file and paste the (single line) output in to your YAML file - be aware that if checking in the YAML file to a source code repo with this information in it, the key would effectively be compromised since base64 isn't encryption.