Fix CrashLoopBackOff, ramdisk-logs no inotifywait

When updating the operators we end up recreating the conductor and
inspector deployments. But the old service images does not yet have
the required command: inotifywait - because of the missing command
the pod enter CrashLoopBackOff state.

To resolve the issue, the OpenstackVersion must be patch to update
the container images.

To workaround this issue, add a check for the inotifywait command,
and in case it does not exist:
 a) Print a Warning
 b) Starta a tail for follow /dev/null to have the container do
    nothing forever. (e.g. until it is restarted with new service
    container images after patching the OpenstackVersion.)

Jira: OSPRH-17520

Author: hjensas

templates/common/bin/runlogwatch.sh

@@ -3,6 +3,13 @@
 # Ramdisk logs path
 LOG_DIR=${LOG_DIR:-/var/lib/ironic/ramdisk-logs}
 
+# NOTE(hjensas): OSPRH-17520
+if ! command -v inotifywait; then
+    echo "WARNING: ramdisk-logs will not be captured, because the inotifywait command is missing. Please patch the OpenstackVersion to update container images."
+    # Start a tail on /dev/null to do nothing forever
+    tail -f /dev/null
+fi
+
 inotifywait -m "${LOG_DIR}" -e close_write |
     while read -r path _action file; do
         echo "************ Contents of ${path}${file} ramdisk log file bundle **************"