Closed templary closed 2 years ago
Hi
Thanks for opening this issue. I never saw this issue starting to appear with the release of 2.6.0. Can you provide some additional information?
kube-proxy
sidecar? In the past we had some documentation pointing to this, but this is no longer needed. Compare the updated docs https://github.com/openanalytics/shinyproxy-config-examples/blob/master/03-containerized-kubernetes/sp-deployment.yaml to the old docs: https://github.com/openanalytics/shinyproxy-config-examples/blob/2d4828b46f9001eb2288b7037252fd0924b22deb/03-containerized-kubernetes/sp-deployment.yaml#L21-L25Thanks
Hello,
No, I don't use sidecar
For production version of ShinyProxy we are using G Cloud Kubernetes version 1.19.12
I happen on every app start. But just in production environment (when shinyproxy is also in Kubernetes). On my local dev version it works completely fine.
application.yml:
----
logging:
file: shinyproxy.log
proxy:
admin-groups: XXXXXX
authentication: saml
container-backend: kubernetes
container-log-path: /XXXXXX
container-wait-time: 180000
favicon-path: /opt/XXXXXX
heartbeat-rate: 10000
heartbeat-timeout: 60000
kubernetes:
internal-networking: true
node-selector:
cloud.google.com/gke-boot-disk: pd-ssd
landing-page: /
livenessProbe: {}
logo-url: XXXXXXXXXXXX
port: 8080
readinessProbe: {}
specs:
- container-cmd:
- R
- -e
- Sys.time(); options('shiny.port'=3838,shiny.host='0.0.0.0'); XXXXXX::run_app(launch.browser
= FALSE, quiet = F)
container-cpu-limit: 1000m
container-cpu-request: 200m
container-env:
SMTP_PASSWORD: XXXXXX
SMTP_USERNAME: XXXXXX
container-image: XXXXXX
container-memory-limit: 1Gi
container-memory-request: 100Mi
description: XXXXXX
display-name: XXXXXX
id: acountprofile
- access-groups:
- admins
- developer
- internal
container-cmd:
- R
- -e
- Sys.time(); options('shiny.port'=3838,shiny.host='0.0.0.0'); XXXXXX::run_app(launch.browser
= FALSE, quiet = F)
container-cpu-limit: 1000m
container-cpu-request: 200m
container-env:
R_CONFIG_ACTIVE: XXXXXX
container-image: XXXXXX
container-memory-limit: 1Gi
container-memory-request: 100Mi
description: XXXXXX
display-name: XXXXXX
id: aXXXXXX
kubernetes-pod-patches: |
- op: add
path: /spec/volumes
value:
- name: nfs-storage-api
persistentVolumeClaim:
claimName: XXXXX
- op: add
path: /spec/containers/0/volumeMounts
value:
- name: XXXXX
mountPath: XXXXXX
.
.
.
.
Hi
This seems like a strange issue. The Kubernetes client library should be able to auto discover the correct URL of the k8s API server. Are you sure you did not specify the proxy.kubernetes.url
property? (maybe using an environment variable or something?) If this isn't the case you could maybe try specifying the proxy.kubernetes.url
yourself. You could try setting it to https://kubernetes.default
or https://kubernetes.default.svc.cluster.local
. The exact domain my depend on your cluster configuration.
Hi thank you for your reply!
We solved problem just by adding this line to config file. Both of them make apps work.
url: https://kubernetes.default.svc.cluster.local
url: https://kubernetes.default
Hello, I would like to ask for info about this error.
After updating the system to shinyproxy 2.6.0 and containerproxy 0.8.10, a problem with launching applications in Kubernetes started to appear.
If I run the application locally on my machine and use the kubectl proxy and port-forward a given instance of the R application, the application runs without problems and everything works as it should.
The problem occurs when Shinyproxy is on a production version in a kubernetes cluster where this error occurs.
Thank you a lot for your time