rancher-users #fleet

mammoth-postman-10874

04/12/2023, 10:23 AM

Is there a regular helm repo for fleet?

most-kite-870

04/13/2023, 8:30 AM

Does anyone know if Github Personal Access Token (PAT) for Organization work with Fleet? The classic token works.

<https://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@github.com/myorg/c-nginx.git> refs/heads/main error: exit status 128, detail: remote: Invalid username or password. fatal: Authentication failed for '<https://github.com/myorg/c-nginx.git/>'

mysterious-whale-87222

04/13/2023, 9:32 PM

Hi, get the following fleet error on installing argocd helm chart.

chart requires kubeVersion: >=1.22.0-0 which is incompatible with Kubernetes v1.20.0

but my cluster version is

Server Version: <http://version.Info|version.Info>{Major:"1", Minor:"25", GitVersion:"v1.25.6", GitCommit:"ff2c119726cc1f8926fb0585c74b25921e866a28", GitTreeState:"clean", BuildDate:"2023-01-18T19:15:26Z", GoVersion:"go1.19.5", Compiler:"gc", Platform:"linux/amd64"}

What can cause this?

✅ 1

steep-furniture-72588

04/18/2023, 7:05 AM

i have changed my server address for rancher. However on my existing clusters for fleet it is still trying to get to the old rancher address. How do i change this so that the existing cluster fleet agent is going to the new address?

powerful-florist-21503

04/20/2023, 2:32 PM

Hello! I have a problem trying to pass cluster name as a value to Helm in fleet.yaml. My fleet.yaml:

defaultNamespace: monitoringfleet

kustomize:

dir: ""

helm:

chart: <https://xxxxxx.yyyyyyy.xxxxxxxxxxx.com/artifactory/api/helm/vii-helm/_external/https/cdn.zabbix.com/zabbix/integrations/kubernetes-helm/6.0/zabbix-helm-chrt-1.1.1.tgz>

releaseName: "zabbixfleet"

values:

zabbixProxy.env[1].value: ${ .ClusterName }

valuesFiles:

- zabbix_values.yaml

So Im trying to pass my cluster name (

${ .ClusterName }

as a value to the helm chart. I get no error on Rancher UI and I was unable to find any relevant logs what could be the issue, but fleet does not install the chart with the clustername, it does not reconcile. I believe there is some error in that above yaml? How to pass cluster name to helm chart or pass cluster annotations to helm chart?

rhythmic-night-12176

04/20/2023, 4:44 PM

Hey all! I have a question about best practice that someone might have run into before: We have several Git repos containing

fleet.yaml

files. Each one of those has several corresponding

GitRepo

s in Continuous Delivery each one pointing at different clusters. This lets us change a branch on a

GitRepo

pointing to a sandbox cluster so we can POC changes without affecting production. Once the branch gets merged to

main

branch the changes are propagated to all clusters that reference the

main

branch. This works really well but it's a huge bottleneck because only one engineer can be referencing that sandbox cluster at a time with their selected development branch. Question: Has anyone else run into this and does anyone have any examples of ways they've solved this in the past? I can't really give everyone their own sandbox cluster yet, unfortunately, or I'd just do that.

powerful-florist-21503

04/21/2023, 11:18 AM

In fleet.yaml, seems to me, that it is not possible to use both

values:

and

valuesFiles:

at the same time. For example, I have fleet.yaml:

values:

zabbixProxy:

env:

- name: "ZBX_HOSTNAME"

value: <http://global.fleet.clusterLabels.management.cattle.io/cluster-display-name|global.fleet.clusterLabels.management.cattle.io/cluster-display-name>

valuesFiles:

- zabbix_values.yaml

This is resulting that

ZBX_HOSTNAME

is not passed to helm. If I disable use of that values file, then the

ZBX_HOSTNAME

is passed to helm. Is this expected behaviour?

billowy-apple-60989

04/21/2023, 12:34 PM

Since the long-standing issue of not being being able to assign clusters to Workspaces https://github.com/rancher/rancher/issues/36132 is still present and Ranchers approach seems to be to ignore it https://github.com/rancher/dashboard/issues/7734 what is Ranchers recommendation for users that make use of custom workspaces? Delete all clusters and import them again to the default one? Let the clusters remain in the custom workspaces and hope stuff doesn't break in the future?

hundreds-evening-84071

04/24/2023, 7:55 PM

On Rancher 2.7.2 and as per this doc: https://github.com/rancher/dashboard/issues/7315 Created

Fleet User

and assigned to my user. The user has Cluster Owner permission (to a downstream cluster) Was able to add a git repo into Fleet, but the list of clusters is blank (in Fleet) In Cluster Management screen the cluster is accessible. What am I missing?

hundreds-sugar-37524

04/28/2023, 4:25 PM

Hey, I'm trying to deploy a basic app using raw manifests and fleet. One manifest is the namespace and the other one is my deployment. Is there a way to tell fleet to create the namespace first ? Is it a bad idea to manage the namespace with fleet ?

refined-analyst-8898

04/28/2023, 5:27 PM

@refined-analyst-8898 has left the channel

millions-notebook-58367

04/29/2023, 5:49 PM

Hi guys, Is it possible to not install a helm chart on a specific cluster using targetCustomizations? Eg: I want to install metrics-server on every possible cluster except the ones that has the label:

provider=gcp

Copy code

defaultNamespace: kube-system
helm:
  releaseName: metrics-server
  chart: metrics-server
  repo: <https://charts.bitnami.com/bitnami>
  values:
    apiService:
      create: true
    securePort: 4443
    extraArgs:
      - --cert-dir=/tmp
      - --kubelet-insecure-tls=true
      - --kubelet-preferred-address-types=InternalIP,ExternalIP,Hostname

targetCustomizations:
  - clusterSelector:
      matchLabels:
        provider: gcp
    helm:
      disabled: true

Something like this

refined-analyst-8898

04/30/2023, 12:31 AM

I created a cluster with Rancher. It comes with Fleet. Can I register a repo with this included Fleet to apply manifests from Git, or should I create another instance of Fleet for non-Rancher purposes?

✅ 1

refined-analyst-8898

04/30/2023, 1:20 PM

I see how to create and manage cluster resources with Fleet bundles. However, deleting a bundle doesn't delete the resource, and searching "delete" in Fleet docs wasn't fruitful. What's the general technique for destroying cluster resources?

✅ 1

strong-action-21448

05/02/2023, 1:33 PM

Hello, I'm trying to deploy helm chart to a rancher-managed EKS 1.25 cluster via fleet but I'm getting this error:

Copy code

Chart requires kubeVersion: >=v1.24.0-0 which is incompatible with Kubernetes v1.20.0

It seems like this indicates rancher is using an old kubectl client when deploying helm charts, and I'm not sure how to upgrade fleet's kubectl?

bored-australia-39597

05/02/2023, 2:45 PM

Hey we upgraded our rancher cluster to 2.7.3 from 2.7.0 and it seems like fleet bundles are now failing to parse some helm values template with "Failed to parse helm values template" has anyone else seen this? It seems to be breaking all of our elasticsearch bundles including logstash,kibana,filebeat,metricbeat

bored-australia-39597

05/02/2023, 2:54 PM

looks like it is the helm prepocessing added with 0.6 break the helm charts since they have ${} values in their helm chart.

hundreds-evening-84071

05/03/2023, 11:08 PM

Hello, I have Rancher 2.7.3... and trying to setup and figure out Fleet as mentioned here https://github.com/rancher/dashboard/issues/7315 I have created "Fleet User" role and added to a user that also has StandardUser role assigned. This user has Cluster Owner privileges for a K3s and a RKE2 clusters As logged on with this user, I go to Continuous Delivery - Continuous Delivery Dashboard and added a repository Next when I go to Cluster, there are no clusters available (yet the user has cluster owner privileges and visible in Cluster Management screen). So, I am trying to figure out what I am doing wrong?

hundreds-sugar-37524

05/04/2023, 10:41 AM

Hello, I'm trying to deploy a helm chart with fleet on two downstream clusters. These clusters are located in different networks and they each have a dedicated helm private registry. I used a

targetCustomizations

in fleet.yml so based on cluster name, fleet pulls from different repo. It works only for one of my cluster, because

helmSecretName

in my

gitRepo

refers to the account from one of the helm registry. Can I override this in my fleet.yml or can I define multiple account in

helmSecretName

? My fleet.yml looks like this:

Copy code

helm:
  repo: repo-a
  chart: my-chart


targetCustomizations:
- name: cluster-b
  clusterName: cluster-b
  helm:
    repo: repo-b
    chart: my-chart

ambitious-plastic-3551

05/07/2023, 4:28 PM

Humm, I still fight this fleet thing, that it re-creates deployment every ~20 minutes

ambitious-plastic-3551

05/07/2023, 4:34 PM

Oh, yes I think I know what's going on, looks like when fleet loses connection to the clusters, then removes deployment from the server

ambitious-plastic-3551

05/07/2023, 4:36 PM

Though what is interesting that it is very selective what it deletes

ambitious-plastic-3551

05/07/2023, 4:38 PM

suspecting non-helm types

ambitious-plastic-3551

05/07/2023, 5:08 PM

Copy code

time="2023-05-07T17:06:31Z" level=info msg="Deleting unknown bundle ID gl-konstanta-ops-k8s-sites-viaint-si-viaint-si--ff6e8, release via-international/gl-konstanta-ops-k8s-sites-viaint-si-viaint-si-ff6e8, expecting release via-international/gl-konstanta-ops-k8s-sites-viaint-si-viaint-si--ff6e8"
time="2023-05-07T17:06:31Z" level=info msg="uninstall: Deleting gl-konstanta-ops-k8s-sites-viaint-si-viaint-si-ff6e8"

ambitious-plastic-3551

05/07/2023, 5:08 PM

Found it!

ambitious-plastic-3551

05/07/2023, 5:09 PM

why is this equality failing? 😄

ambitious-plastic-3551

05/07/2023, 5:40 PM

https://github.com/rancher/fleet/issues/1511

billowy-apple-60989

05/08/2023, 12:33 PM

General GitOps question.. not necessarily fleet-specific but throwing it out there. I have a number of

GitRepos

in Fleet that deploy some apps and cluster-config to 35+ downstream clusters, these repos all watch master/main branches. Now if i want to make a change to one of these deployed apps (update, config change, whatever else) i'd like to try this out on a few clusters before pushing it to all clusters, how would i do this? Create a new

GitRepo

that watches a new branch and only deploys to specific clusters? Some other way? Thanks

hundreds-evening-84071

05/08/2023, 6:03 PM

Hey folks; Rancher 2.7.3... We have 1 repository defined in Continuous Delivery for "fleet-default" and set a target for a downstream RKE2 cluster... But, the workload gets deployed on rancher-local cluster? Am trying to figure out why this is happening? Sorry am new to fleet so I need some basic guidance on how to troubleshoot this?

freezing-holiday-13112

05/09/2023, 10:43 PM

In fleet imageScan is a spec tagName how does this work? Is it supposed to change with every release? Also secretRef: I presume it's expecting some specific keys right?