[reboot-cause]: Move reboot cause processing to its own service, 'process-reboot-cause' (#3102)

files/build_templates/sonic_debian_extension.j2

@@ -220,6 +220,11 @@ sudo cp $IMAGE_CONFIGS/caclmgrd/caclmgrd.service  $FILESYSTEM_ROOT/etc/systemd/s
 sudo LANG=C chroot $FILESYSTEM_ROOT systemctl enable caclmgrd.service
 sudo cp $IMAGE_CONFIGS/caclmgrd/caclmgrd $FILESYSTEM_ROOT/usr/bin/
 
+# Copy process-reboot-cause service files
+sudo cp $IMAGE_CONFIGS/process-reboot-cause/process-reboot-cause.service  $FILESYSTEM_ROOT/etc/systemd/system/
+sudo LANG=C chroot $FILESYSTEM_ROOT systemctl enable process-reboot-cause.service
+sudo cp $IMAGE_CONFIGS/process-reboot-cause/process-reboot-cause $FILESYSTEM_ROOT/usr/bin/
+
 ## Install package without starting service
 ## ref: https://wiki.debian.org/chroot
 sudo tee -a $FILESYSTEM_ROOT/usr/sbin/policy-rc.d > /dev/null <<EOF

files/image_config/platform/rc.local

@@ -106,35 +106,6 @@ value_extract() {
     done
 }
 
-# Set up previous and next reboot cause files accordingly
-process_reboot_cause() {
-    REBOOT_CAUSE_DIR="/host/reboot-cause"
-    REBOOT_CAUSE_FILE="${REBOOT_CAUSE_DIR}/reboot-cause.txt"
-    PREVIOUS_REBOOT_CAUSE_FILE="${REBOOT_CAUSE_DIR}/previous-reboot-cause.txt"
-
-    mkdir -p $REBOOT_CAUSE_DIR
-
-    # If this is the first boot after an image install, store that as the
-    # previous reboot cause.
-    if [ -f $FIRST_BOOT_FILE ]; then
-        echo "SONiC image installation" > $PREVIOUS_REBOOT_CAUSE_FILE
-    fi
-
-    # If there is an existing REBOOT_CAUSE_FILE, copy that file to
-    # PREVIOUS_REBOOT_CAUSE_FILE.
-    if [ -f $REBOOT_CAUSE_FILE ]; then
-        mv -f $REBOOT_CAUSE_FILE $PREVIOUS_REBOOT_CAUSE_FILE
-    else
-        echo "Unknown reboot cause" > $PREVIOUS_REBOOT_CAUSE_FILE
-    fi
-
-    # Log the previous reboot cause to the syslog
-    logger "Previous reboot cause: $(cat $PREVIOUS_REBOOT_CAUSE_FILE)"
-
-    # Set the default cause for the next reboot
-    echo "Unexpected reboot" > $REBOOT_CAUSE_FILE
-}
-
 program_console_speed()
 {
     speed=$(cat /proc/cmdline | grep -Eo 'console=ttyS[0-9]+,[0-9]+' | cut -d "," -f2)
@@ -150,9 +121,6 @@ program_console_speed()
 
 #### Begin Main Body ####
 
-# Set up previous and next reboot cause files
-process_reboot_cause
-
 # If the machine.conf is absent, it indicates that the unit booted
 # into SONiC from another NOS. Extract the machine.conf from ONIE.
 if [ ! -e /host/machine.conf ]; then

files/image_config/process-reboot-cause/process-reboot-cause

@@ -0,0 +1,89 @@
+#!/usr/bin/env python
+#
+# process-reboot-cause
+#
+# Program designed to run once, soon after system boot which will
+# determine the cause of the previous reboot and store it to the disk,
+#
+
+try:
+    import os
+    import pwd
+    import subprocess
+    import sys
+    import syslog
+    import sonic_platform
+    from sonic_daemon_base.daemon_base import Logger
+except ImportError as err:
+    raise ImportError("%s - required module not found" % str(err))
+
+VERSION = "1.0"
+
+SYSLOG_IDENTIFIER = "process-reboot-cause"
+
+# Global logger instance
+logger = Logger(SYSLOG_IDENTIFIER)
+
+REBOOT_CAUSE_DIR = "/host/reboot-cause/"
+REBOOT_CAUSE_FILE = REBOOT_CAUSE_DIR + "reboot-cause.txt"
+PREVIOUS_REBOOT_CAUSE_FILE = REBOOT_CAUSE_DIR + "previous-reboot-cause.txt"
+
+UNKNOWN_REBOOT_CAUSE = "Unknown"
+
+
+def main():
+    logger.log_info("Starting up...")
+
+    if not os.geteuid() == 0:
+        logger.log_error("User {} does not have permission to execute".format(pwd.getpwuid(os.getuid()).pw_name))
+        sys.exit("This utility must be run as root")
+
+    # Create REBOOT_CAUSE_DIR if it doesn't exist
+    if not os.path.exists(REBOOT_CAUSE_DIR):
+        os.makedirs(REBOOT_CAUSE_DIR)
+
+    # Remove stale PREVIOUS_REBOOT_CAUSE_FILE if it exists
+    if os.path.exists(PREVIOUS_REBOOT_CAUSE_FILE):
+        os.remove(PREVIOUS_REBOOT_CAUSE_FILE)
+
+    # Check if the previous reboot was caused by hardware
+    platform = sonic_platform.platform.Platform()
+
+    chassis = platform.get_chassis()
+
+    hardware_reboot_cause, optional_details = chassis.get_reboot_cause()
+
+    # Set a default previous reboot cause
+    previous_reboot_cause = UNKNOWN_REBOOT_CAUSE
+
+    if hardware_reboot_cause == chassis.REBOOT_CAUSE_NON_HARDWARE:
+        # The reboot was not caused by hardware. If there is a REBOOT_CAUSE_FILE, it will
+        # contain any software-related reboot info. We will use it as the previous cause.
+        if os.path.isfile(REBOOT_CAUSE_FILE):
+            cause_file = open(REBOOT_CAUSE_FILE, "r")
+            previous_reboot_cause = cause_file.readline().rstrip('\n')
+            cause_file.close()
+    elif hardware_reboot_cause == chassis.REBOOT_CAUSE_HARDWARE_OTHER:
+        previous_reboot_cause = "{} ({})".format(hardware_reboot_cause, optional_details)
+    else:
+        previous_reboot_cause = hardware_reboot_cause
+
+    # Write the previous reboot cause to PREVIOUS_REBOOT_CAUSE_FILE
+    prev_cause_file = open(PREVIOUS_REBOOT_CAUSE_FILE, "w")
+    prev_cause_file.write(previous_reboot_cause)
+    prev_cause_file.close()
+
+    # Also log the previous reboot cause to the syslog
+    logger.log_info("Previous reboot cause: {}".format(previous_reboot_cause))
+
+    # Remove the old REBOOT_CAUSE_FILE
+    os.remove(REBOOT_CAUSE_FILE)
+
+    # Write a new default reboot cause file for the next reboot
+    cause_file = open(REBOOT_CAUSE_FILE, "w")
+    cause_file.write(UNKNOWN_REBOOT_CAUSE)
+    cause_file.close()
+
+
+if __name__ == "__main__":
+    main()

files/image_config/process-reboot-cause/process-reboot-cause.service

@@ -0,0 +1,10 @@
+[Unit]
+Description=Reboot cause determination service
+After=rc-local.service
+
+[Service]
+Type=simple
+ExecStart=/usr/bin/process-reboot-cause
+
+[Install]
+WantedBy=multi-user.target