Kubernetes runner timeout

GitLab CI/CD
1

I created a self-hosted GitLab instance using the helm chart (https://charts.gitlab.io/ v4.11.3), and external Let’s Encrypt and Cert-Manager for Ingress. The helm chart is installed via a parent chart, thus the gitlab section. Every CI task fails with a timeout fetching the git repo. I have tried many different tweaks to the .gitlab-ci.yml to no avail. I am able to access GitLab just fine in the Kubernetes cluster and on my workstation. What could be the issue of the runner timing out?

Error

Getting source from Git repository
Fetching changes with git depth set to 50...
Initialized empty Git repository in xxxxxx
Created fresh repository.
fatal: unable to access 'https://gitlab.xxxx.com/xxxx/xxxx.git/': Failed to connect to gitlab.xxxx.com port 443: Operation timed out
Cleaning up file based variables00:00
ERROR: Job failed: command terminated with exit code 1

values.yaml

global:
  logLevel: debug
  hosts:
    domain: xxxx.com
  ingress:
    annotations:
      cert-manager.io/cluster-issuer: letsencrypt-prod
      kubernetes.io/tls-acme: true
    class: nginx
    configureCertmanager: false

gitlab:
  certmanager:
    install: false
  gitlab:
    webservice:
      ingress:
        tls:
          secretName: gitlab-webservice-tls
  nginx-ingress:
    enabled: false
  registry:
    ingress:
      tls:
        secretName: gitlab-registry-tls
  minio:
    ingress:
      tls:
        secretName: gitlab-minio-tls

.gitlab-ci.yml

image:
  name: alpine/helm:3.5.4
  entrypoint: ["/bin/sh", "-c"]

variables:
  FF_GITLAB_REGISTRY_HELPER_IMAGE: 1

stages:
  - lint-helm-chart

lint-helm:
  stage: lint-helm-chart
  script:
    - helm lint
2

To make things even stranger, the simple lint task may successfully run when I do a helm upgrade causing the gitlab-runner to restart. Minio also seems to restart, so I can’t tell which restart is resolving the issue.

3

I’ve tried rebuilding the cluster a few times and reinstalling GitLab with different options, but still have the same issue. I was hoping to pitch using GitLab for use at my company, but this is a major roadblock that I cannot get past so I will need to start looking elsewhere.

4

I was able to get this working by switching the AWS load balancer type used in ingress-nginx. After switching from nlb load balancer to elb load balancer, the runner is now working.