Hey guys I m trying to deploy the server in kubernetes digit Prefect Community #prefect-server

Hey guys, I’m trying to deploy the server in kuber...

wiretrack

07/03/2021, 6:31 PM

Hey guys, I’m trying to deploy the server in kubernetes (digitalocean) with my own manifests and external db, hasura and graphql seem to work fine, but I can’t seem to manage to make apollo’s service to work, I keep getting

connection refused

in the liveness and readiness probes. does anybody have any idea on what I might be doing wrong?

Kevin Kho

07/03/2021, 6:33 PM

Hey, do you see anything in the logs? Is this when trying to spin up an agent? Or when just doing

prefect server start

wiretrack

07/03/2021, 6:35 PM

no, I was getting a connection error when trying to spin up the agent, but I wasn’t using the readiness/liveness probe in apollo, so i figured the problem must be in apollo. no errors in the pod logs, I just get

Checking GraphQL service at <http://prefect-graphql-service/health> ...

wiretrack

07/03/2021, 6:36 PM

and when i exec into the pod and try

curl localhost:4200

i also get

connection refused

i’m not very familiar with apollo, but i thought it should work

Kevin Kho

07/03/2021, 7:02 PM

This is a bit beyond me. I’d have to ask team members on tuesday (monday is a holiday)

wiretrack

07/03/2021, 7:09 PM

thanks mate. i’ll keep digging to see if I can find a way. cheers!

wiretrack

07/03/2021, 7:20 PM

so just to complement the information, i removed the probes, and exec into the pod.

npm run serve

is not working because

graphql

and

hasura

endpoints were not working, so probably something related to the

envvars

i passed (the service addresses). i’ll keep trying and keep you guys posted. it seems that the

helm chart

uses

v1alpha

for hasura, but in my console it’s just

v1

, will let you guys know what i find out. so basically even though the

health_check

graphql

works, the endpoint

<http://prefect-graphql-service:4201/graphql>

is not found

wiretrack

07/03/2021, 11:05 PM

so I’ve been playing around, and found that there were two problems: • the lack of “/” at the end of the graphql service • livenessprobe was too short, so i had to add

initialDelaySeconds: 30

to the

apollo

deployment now I have a problem with the agent, but I’ll open a new message

8 Views

Open in Slack

Previous Next