use kube service user for pod sweeper (#4737)

* use kube service user for pod sweeper

* add pod sweeper logs

* temporarily switch to stable for testing

* temporarily remove building steps for kube testing since it can use prod images

* output date strings from date command

* load stable images

* remove loading since it can pull the images

* increase window for success storage to two hours

* revert test logging changes

Author: jrhizor

kube/resources/pod-sweeper.yaml

@@ -18,7 +18,7 @@ data:
     while :
     do
       # Shorter time window for completed pods
-      SUCCESS_DATE_STR=`date -d 'now - 30 seconds' --utc -Ins`
+      SUCCESS_DATE_STR=`date -d 'now - 2 hours' --utc -Ins`
       SUCCESS_DATE=`date -d $SUCCESS_DATE_STR +%s`
       # Longer time window for pods in error (to debug)
       NON_SUCCESS_DATE_STR=`date -d 'now - 24 hours' --utc -Ins`
@@ -59,6 +59,7 @@ spec:
       labels:
         airbyte: pod-sweeper
     spec:
+      serviceAccountName: airbyte-admin
       containers:
         - name: airbyte-pod-sweeper
           image: bitnami/kubectl

tools/bin/acceptance_test_kube.sh

@@ -28,8 +28,9 @@ sleep 120s
 
 server_logs () { echo "server logs:" && kubectl logs deployment.apps/airbyte-server; }
 scheduler_logs () { echo "scheduler logs:" && kubectl logs deployment.apps/airbyte-scheduler; }
+pod_sweeper_logs () { echo "pod sweeper logs:" && kubectl logs deployment.apps/airbyte-pod-sweeper; }
 describe_pods () { echo "describe pods:" && kubectl describe pods; }
-print_all_logs () { server_logs; scheduler_logs; describe_pods; }
+print_all_logs () { server_logs; scheduler_logs; pod_sweeper_logs; describe_pods; }
 
 trap "echo 'kube logs:' && print_all_logs" EXIT