[local-preview] Run gitpod-telemetry during exit

While #11077 adds a way
to send telemetry once a `local-preview` instance is instantiated
and is running, The user data is not very useful here as the user
hasn't performed any actions, etc. This data is only useful right now
to know that a new install has been performed.

This PR updates the `local-preview` to also do the same whenever
a exit signal is created, By doing this we also get the same
data once a instance is being closed and hence giving use anonymous
info on the usage.

Signed-off-by: Tarun Pothulapati <[email protected]>

install/preview/entrypoint.sh

@@ -4,11 +4,6 @@
 
 set -e
 
-if [ "$1" != "logging" ]; then
-  $0 logging 2>&1 | /prettylog
-  exit
-fi
-
 # check for minimum requirements
 REQUIRED_MEM_KB=$((6 * 1024 * 1024))
 total_mem_kb=$(awk '/MemTotal:/ {print $2}' /proc/meminfo)
@@ -136,8 +131,7 @@ yq eval-all -i 'del(.spec.template.spec.initContainers[0])' /var/lib/rancher/k3s
 for f in /var/lib/rancher/k3s/server/manifests/gitpod/*.yaml; do (cat "$f"; echo) >> /var/lib/rancher/k3s/server/manifests/gitpod.yaml; done
 rm -rf /var/lib/rancher/k3s/server/manifests/gitpod
 
-# waits for gitpod pods to be ready, and manually runs the `gitpod-telemetry` cronjob
-run_telemetry(){
+run_init_telemetry(){
   # wait for the k3s cluster to be ready and Gitpod workloads are added
   sleep 100
   # indefinitely wait for Gitpod pods to be ready
@@ -146,7 +140,18 @@ run_telemetry(){
   kubectl create job gitpod-telemetry-init --from=cronjob/gitpod-telemetry
 }
 
-run_telemetry 2>&1 &
+run_exit_telemetry(){
+  # wait for Gitpod pods to be ready
+  kubectl wait --timeout=-1s --for=condition=ready pod -l app=gitpod,component!=migrations
+  # manually tun the cronjob
+  kubectl create job gitpod-telemetry-exit --from=cronjob/gitpod-telemetry
+  # wait for the job to finish
+  sleep 30
+}
+
+trap 'run_exit_telemetry 2>&1' EXIT INT HUP
+
+run_init_telemetry 2>&1 &
 
 /bin/k3s server --disable traefik \
   --node-label gitpod.io/workload_meta=true \