Lifecycle (#111)

[Bugfixes]
- Suppress errors
- Better verbose message for time to become active in update function
- Abort "scale" jobs when starting applications
-  - Fixes errors when starting various applications

functions/backup_restore/database/cnpg_dump.sh

@@ -147,7 +147,7 @@ wait_for_postgres_pod() {
 
     # shellcheck disable=SC2034
     for i in {1..30}; do
-        pod_status=$(k3s kubectl get pods "${app_name}-cnpg-main-1" -n "ix-${app_name}" -o jsonpath="{.status.phase}")
+        pod_status=$(k3s kubectl get pods "${app_name}-cnpg-main-1" -n "ix-${app_name}" -o jsonpath="{.status.phase}" 2>/dev/null)
 
         if [[ "$pod_status" == "Running" ]]; then
             return 0
@@ -158,7 +158,6 @@ wait_for_postgres_pod() {
     return 1
 }
 
-
 backup_cnpg_databases() {
     retention=$1
     timestamp=$2

functions/enable/apt.sh

@@ -23,12 +23,10 @@ if [ "$success" = true ]; then
     echo -e "\n${bold}Example usage:${reset}"
     echo -e "For most use cases, use ${blue}apt${reset}:"
     echo -e "  ${blue}apt update${reset}"
-    echo -e "  ${blue}apt upgrade${reset}"
     echo -e "  ${blue}apt install package_name${reset}"
 
     echo -e "\nFor advanced users or specific cases, you can use ${blue}apt-get${reset}:"
     echo -e "  ${blue}apt-get update${reset}"
-    echo -e "  ${blue}apt-get upgrade${reset}"
     echo -e "  ${blue}apt-get install package_name${reset}"
 fi
 }

functions/update/commander.sh

@@ -88,7 +88,6 @@ get_apps_with_status() {
             apps_with_status+=("$app_name,$status")
         fi
     done < <(check_filtered_apps "${array[@]/,*}")
-
 }
 
 # Skip if the image update should be ignored

functions/update/post_process.sh

@@ -27,13 +27,6 @@ rollbacks_disabled(){
     echo_array+=("Manual intervention is required\nStopping, then Abandoning")
 }
 
-handle_wait() {
-    if [[ "$verbose" == true ]]; then
-        echo_array+=("Waiting $((timeout-SECONDS)) more seconds for $app_name to be ACTIVE")
-    fi
-    sleep 5
-}
-
 handle_rollback() {
     echo_array+=("Error: Run Time($SECONDS) for $app_name has exceeded Timeout($timeout)")
     echo_array+=("If this is a slow starting application, set a higher timeout with -t")
@@ -81,6 +74,9 @@ post_process(){
         status=$(update_status)
 
         if [[ "$status"  ==  "ACTIVE" ]]; then
+            if [[ "$verbose" == true ]]; then
+                echo_array+=("Became Active after $SECONDS seconds")
+            fi
             if [[ "$startstatus"  ==  "STOPPED" ]]; then
                 update_stop_handler 'Returing to STOPPED state...'
             else
@@ -107,10 +103,11 @@ post_process(){
                 break
             fi
         else
-            handle_wait
+            sleep 5
         fi
     done
 
     echo_array
 }
 export -f post_process
+

functions/update/pre_process.sh

@@ -106,6 +106,13 @@ pre_process() {
         return
     fi
 
+    if [[ $stop_before_update == true && "$startstatus" != "STOPPED" ]]; then
+        if ! start_app "$app_name"; then
+            echo_array
+            return
+        fi
+    fi
+
     if [[ $rollback == true || "$startstatus"  ==  "STOPPED" ]]; then
         if printf '%s\0' "${apps_with_status[@]}" | grep -iFxqz "${app_name},stopAll-on"; then
             echo_array+=("Stopped")

utils/start_app.sh

@@ -1,22 +1,77 @@
 #!/bin/bash
 
+abort_job(){
+    local app_name=$1
+    job_ids=""
+
+    # shellcheck disable=SC2034
+    for i in {1..60}; do
+        job_ids=$(get_running_job_id "$app_name")
+
+        if [[ -n "$job_ids" ]]; then
+            while IFS= read -r job_id; do
+                midclt call core.job_abort "$job_id" > /dev/null 2>&1
+            done <<< "$job_ids"
+            return 0
+        fi
+
+        sleep 1
+    done
+    return 1
+}
+
+wait_for_redeploy_methods(){
+    local app_name=$1
+    local methods=""
+
+    # shellcheck disable=SC2034
+    for i in {1..60}; do
+        methods=$(get_running_methods "$app_name")
+
+        if [[ "$methods" == *"chart.release.redeploy"* && "$methods" == *"chart.release.redeploy_internal"* ]]; then
+            return 0
+        fi
+
+        sleep 1
+    done
+    return 1
+}
+
+get_running_methods(){
+    local app_name=$1
+    midclt call core.get_jobs | jq -r --arg app_name "$app_name" \
+        '.[] | select( .time_finished == null and .state == "RUNNING" and (.arguments[0] == $app_name)) | .method'
+}
+
+get_running_job_id(){
+    local app_name=$1
+    midclt call core.get_jobs | jq -r --arg app_name "$app_name" \
+        '.[] | select( .time_finished == null and .state == "RUNNING" and (.progress.description | test("Waiting for pods to be scaled to [0-9]+ replica\\(s\\)$")) and (.arguments[0] == $app_name and .method == "chart.release.scale") ) | .id'
+}
+
 start_app(){
     local app_name=$1
-    local replica_count=$2
+    local replica_count=${2:-$(pull_replicas "$app_name")}
+    local job_id
 
     # Check if app is a cnpg instance, or an operator instance
     output=$(check_filtered_apps "$app_name")
-    if echo "$output" | grep -q "${app_name},stopAll-on"; then
-        if ! cli -c 'app chart_release scale release_name='\""$app_name"\"\ 'scale_options={"replica_count": '"$replica_count}" > /dev/null; then
-            return 1
-        fi
-        if ! cli -c "app chart_release update chart_release=\"$app_name\" values={\"global\": {\"stopAll\": false}}" > /dev/null; then
+    if [[ $output == *"${app_name},stopAll-on"* ]]; then
+        if ! cli -c "app chart_release update chart_release=\"$app_name\" values={\"global\": {\"stopAll\": false}}" > /dev/null 2>&1; then
             return 1
         fi
+        abort_job "$app_name"
+        job_id=$(midclt call chart.release.scale "$app_name" '{"replica_count": '"$replica_count"'}') || return 1
+        wait_for_redeploy_methods "$app_name"
+        midclt call core.job_abort "$job_id" > /dev/null 2>&1
+    elif [[ $output == *"${app_name},stopAll-off"* ]]; then
+        job_id=$(midclt call chart.release.scale "$app_name" '{"replica_count": '"$replica_count"'}') || return 1
+        wait_for_redeploy_methods "$app_name"
+        midclt call core.job_abort "$job_id" > /dev/null 2>&1
     else
-        if ! cli -c 'app chart_release scale release_name='\""$app_name"\"\ 'scale_options={"replica_count": '"$replica_count}" > /dev/null; then
+        if ! cli -c 'app chart_release scale release_name='\""$app_name"\"\ 'scale_options={"replica_count": '"$replica_count}" > /dev/null 2>&1; then
             return 1
         fi
     fi
     return 0
-}
+}

utils/stop_app.sh

@@ -70,13 +70,13 @@ stop_app() {
     fi
 
     # Check if the output contains the desired namespace and "cnpg" or "operator"
-    if echo "$output" | grep -q "${app_name},stopAll-*"; then
-        timeout "${timeout}s" cli -c "app chart_release update chart_release=\"$app_name\" values={\"global\": {\"stopAll\": true}}" > /dev/null
-        handle_timeout $?
-    elif echo "$output" | grep -q "${app_name},operator"; then
+    if echo "$output" | grep -q "${app_name},operator"; then
         return 3
+    elif echo "$output" | grep -q "${app_name},stopAll-*"; then
+        timeout "${timeout}s" cli -c "app chart_release update chart_release=\"$app_name\" values={\"global\": {\"stopAll\": true}}" > /dev/null 2>&1
+        handle_timeout $?
     else
-        timeout "${timeout}s" cli -c 'app chart_release scale release_name='\""$app_name"\"\ 'scale_options={"replica_count": 0}' > /dev/null
+        timeout "${timeout}s" cli -c "app chart_release update chart_release=\"$app_name\" values={\"global\": {\"stopAll\": true}}" > /dev/null 2>&1
         handle_timeout $?
     fi
 }