CI: make upgrades more robust and skip 1m precautionary sleep

.github/workflows/test.yml

@@ -119,6 +119,7 @@ jobs:
           helm install jupyterhub jupyterhub/jupyterhub --values dev-config.yaml
           await_jupyterhub
           await_autohttps_tls_cert_acquisition
+          await_autohttps_tls_cert_save
 
       - name: Install JupyterHub dev chart
         run: |
@@ -127,17 +128,17 @@ jobs:
           await_jupyterhub
           await_autohttps_tls_cert_acquisition
 
-      - name: Wait 1 minute for everything to be ready
-        # TODO: find out exactly which resources are slow and wait for thoses
-        # So far errors have included:
-        # - some pods not being ready
-        # - network policies not being ready
-        run: sleep 1m
       - name: Run tests
         run: |
           . ./ci/common
           # If you have problems with the tests add '--capture=no' to show stdout
-          pytest --verbose --exitfirst ./tests || (full_namespace_report && exit 1)
+          pytest --verbose --exitfirst --color=yes ./tests
+
+      - name: Provide info on failures
+        if: failure()
+        run: |
+          . ./ci/common
+          full_namespace_report
 
 
   docs_linkcheck:

ci/common

@@ -42,7 +42,7 @@ await_jupyterhub() {
 
 await_autohttps_tls_cert_acquisition() {
     i=0; while [ $i -ne 60 ]; do
-        kubectl logs deploy/pebble -c pebble | grep "Issued certificate" \
+        kubectl logs deploy/autohttps -c traefik | grep "Adding certificate" \
         && acquired_cert=true && break \
         || acquired_cert=false && sleep 0.5 && i=$((i + 1))
     done
@@ -62,6 +62,25 @@ await_autohttps_tls_cert_acquisition() {
     sleep 1
 }
 
+await_autohttps_tls_cert_save() {
+    i=0; while [ $i -ne 60 ]; do
+        kubectl logs deploy/autohttps -c secret-sync | grep "Created secret" \
+        && saved_cert=true && break \
+        || saved_cert=false && sleep 0.5 && i=$((i + 1))
+    done
+    if [ "$saved_cert" != "true" ]; then
+        echo "Certificate acquisition failed!"
+        kubectl get service,networkpolicy,configmap,pod
+        kubectl describe pod -l app.kubernetes.io/name=pebble
+        kubectl logs deploy/pebble --all-containers # --prefix
+        kubectl describe pod -l app.kubernetes.io/name=pebble-coredns
+        kubectl logs deploy/pebble-coredns --all-containers # --prefix
+        kubectl describe pod -l component=autohttps
+        kubectl logs deploy/autohttps --all-containers # --prefix
+        exit 1
+    fi
+}
+
 setup_kubeval () {
     echo "setup kubeval ${KUBEVAL_VERSION}"
     curl -sfL https://github.com/instrumenta/kubeval/releases/download/${KUBEVAL_VERSION}/kubeval-linux-amd64.tar.gz | tar xz kubeval
@@ -106,10 +125,6 @@ full_namespace_report () {
     echo ""
     echo "Just while debugging intermittent issue, lets output the logs of the proxy pod."
     kubectl logs --all-containers deploy/proxy
-
-    echo ""
-    echo "Just while debugging intermittent issue, lets output the logs of the autohttps pod."
-    kubectl logs --all-containers deploy/autohttps
 }
 
 install_and_run_chartpress_and_pebble () {

ci/publish

@@ -40,4 +40,4 @@ fi
 
 # Let us log the changes chartpress did, it should include replacements for
 # fields in values.yaml, such as what tag for various images we are using.
-git --no-pager diff
+git --no-pager diff --color