enable kubectl del --force for rogue pods

Change-Id: Ic60262dd5b471f64bee2b1c9581edf54ab1eddd6
Issue-ID: LOG-914
Signed-off-by: Michael O'Brien <michael@obrienlabs.org>

1 files changed, 16 insertions, 15 deletions

diff --git a/deploy/cd.sh b/deploy/cd.sh
index ebb1241..ae8e45a 100755
--- a/deploy/cd.sh
+++ b/deploy/cd.sh
@@ -17,7 +17,7 @@
 #############################################################################
 # v20190115
 # https://wiki.onap.org/display/DW/ONAP+on+Kubernetes
-# source from https://jira.onap.org/browse/OOM-320, 326, 321, 898, 925
+# source from https://jira.onap.org/browse/OOM-320, 326, 321, 898, 925, 914
 # Michael O'Brien
 #
 
@@ -90,9 +90,7 @@ deploy_onap() {
     if [ "$BRANCH" == "amsterdam" ]; then
       oom/kubernetes/oneclick/deleteAll.bash -n $ENVIRON
     else
-      # run undeploy for completeness of the deploy/undeploy cycle - note that pv/pvcs are not deleted in all cases
-      # this will fail as expected on a clean first run of the deployment - the plugin will be installed for run n+1
-      sudo helm undeploy $ENVIRON --purge
+      echo "kubectl delete namespace $ENVIRON"
       # workaround for secondary orchestration in dcae
       kubectl delete namespace $ENVIRON
       echo "sleep for 4 min to allow the delete to finish pod terminations before trying a helm delete"
@@ -101,21 +99,23 @@ deploy_onap() {
     fi
 
     # verify
-    DELETED=$(kubectl get pods --all-namespaces | grep -E '0/|1/2|1/3|2/3' | wc -l)
+    DELETED=$(kubectl get pods --namespace $ENVIRON | grep -E '0/|1/2|1/3|2/3' | wc -l)
     echo "showing $DELETED undeleted pods"
+    kubectl get pods --namespace $ENVIRON | grep -E '0/|1/2|1/3|2/3'
     echo "verify deletion is finished."
+
+    # this block: for very infrequent rogue pods that are out-of-band of kubernetes - they dont delete without a force delete
     # max number of cycles exits to --force block next
-    local MAX_DELETION_WAIT_PERIODS_BEFORE_RUNNING_FORCE=120 # 30 min
-    local DELETE_COUNTER=0
-    while [  $(kubectl get pods --all-namespaces | grep -E '0/|1/2|1/3|2/3' | wc -l) -gt 0 ]; do
+    local MAX_DELETION_WAIT_PERIODS_BEFORE_RUNNING_FORCE=40 # 10 min
+    while [  $(kubectl get pods --namespace $ENVIRON  | grep -E '0/|1/2|1/3|2/3' | wc -l) -gt 0 ]; do
       sleep 15
-      echo "waiting for deletions to complete, iteration $DELETE_COUNTER of $MAX_DELETION_WAIT_PERIODS_BEFORE_RUNNING_FORCE"
+      echo "waiting for deletions to complete, iterations left: $MAX_DELETION_WAIT_PERIODS_BEFORE_RUNNING_FORCE"
       # addressing rare occurrence on Terminating instances requiring scripted --force in next merge for LOG-914
-      COUNTER=$((COUNTER + 1 ))
       MAX_DELETION_WAIT_PERIODS_BEFORE_RUNNING_FORCE=$((MAX_DELETION_WAIT_PERIODS_BEFORE_RUNNING_FORCE - 1))
       if [ "$MAX_DELETION_WAIT_PERIODS_BEFORE_RUNNING_FORCE" -eq 0 ]; then
         #https://wiki.onap.org/display/DW/ONAP+Development#ONAPDevelopment-WorkingwithJSONPath
-        export POD_NAMES=$(kubectl get pods --field-selector=status.phase!=Running --all-namespaces -o jsonpath="{.items[*].metadata.name}")
+        #export POD_NAMES=$(kubectl get pods --field-selector=status.phase!=Running --all-namespaces -o jsonpath="{.items[*].metadata.name}")
+        export POD_NAMES=$(kubectl get pods --namespace $ENVIRON -o jsonpath="{.items[*].metadata.name}")
         echo "--force delete on pods: $POD_NAMES"
         for pod in $POD_NAMES; do
           echo "running: kubectl delete pods $pod --grace-period=0 --force -n $ENVIRON"
@@ -123,8 +123,9 @@ deploy_onap() {
         done
       fi
     done
-    echo "Pod deletions completed"
-    
+
+    echo "Pod deletions completed - running helm undeploy then kubectl delete pv,pvc,secrets,cluserrolebindings"
+    sudo helm undeploy $ENVIRON --purge   
     # specific to when there is no helm release
     kubectl delete pv --all
     kubectl delete pvc --all
@@ -222,7 +223,7 @@ deploy_onap() {
           # account for podd that have varying deploy times or replicaset sizes
           # don't count the 0/1 completed pods - and skip most of the ResultSet instances except 1
           # dcae boostrap is problematic
-          DEPLOY_NUMBER_PODS_PARTIAL_ARRAY=(2 5 13 9 11 1 11 2 6 16 10 12 11 2 8 6 3 18 2 5 5 5 1 9 11 3 1)
+          DEPLOY_NUMBER_PODS_PARTIAL_ARRAY=(2 5 11 9 11 1 11 2 6 16 10 12 11 2 8 6 3 18 2 5 5 5 1 9 11 3 1)
         else
           # casablanca branches
           DEPLOY_ORDER_POD_NAME_ARRAY=('consul msb dmaap dcaegen2 aaf robot aai esr multicloud oof so sdc sdnc vid policy portal log vfc uui vnfsdk appc clamp cli pomba vvp contrib sniro-emulator')
@@ -231,7 +232,7 @@ deploy_onap() {
           # account for podd that have varying deploy times or replicaset sizes
           # don't count the 0/1 completed pods - and skip most of the ResultSet instances except 1
           # dcae boostrap is problematic
-          DEPLOY_NUMBER_PODS_PARTIAL_ARRAY=(2 5 13 9 11 1 11 2 6 16 10 12 11 2 8 6 3 18 2 5 5 5 1 9 11 3 1)
+          DEPLOY_NUMBER_PODS_PARTIAL_ARRAY=(2 5 11 9 11 1 11 2 6 16 10 12 11 2 8 6 3 18 2 5 5 5 1 9 11 3 1)
         fi
         echo "deploying for $BRANCH using profile $DEPLOY_ORDER_POD_NAME_ARRAY"
       else