From 6218755c3defc5fbde81b4e59c4be472d3f9c69a Mon Sep 17 00:00:00 2001
From: Bin Qian <bin.qian@windriver.com>
Date: Wed, 14 Aug 2019 10:51:12 -0400
Subject: [PATCH] Soften NTP alarm language for syncing with peer

In IPv6 setup, NTP refid is hash result of reference's IPv6 address.
In such case, do not try to intepret the refid to tell if the peer
has a reliable source.

When the NTP service uses peer controller as reference, the alarm
is a reminder to the admin user instead of reporting an issue. This
is a minor alarm.

Closes-Bug: 1834071
Change-Id: Ia2770ba7ed77640e58e8c35254a504b57487ff8f
Signed-off-by: Bin Qian <bin.qian@windriver.com>
---
 monitoring/collectd-extensions/src/ntpq.py | 32 ++++++++++++++++------
 1 file changed, 23 insertions(+), 9 deletions(-)

diff --git a/monitoring/collectd-extensions/src/ntpq.py b/monitoring/collectd-extensions/src/ntpq.py
index cb400791c..d3ee4538c 100755
--- a/monitoring/collectd-extensions/src/ntpq.py
+++ b/monitoring/collectd-extensions/src/ntpq.py
@@ -88,6 +88,7 @@ class NtpqObject:
     reachable_servers = []          # list of reachable servers
     selected_server = 'None'        # the ip address of the selected server
     selected_server_save = 'None'   # the last selected server ; note change
+    peer_selected = False           # true when peer is selected
 
     # variables used to raise alarms to FM
     suppression = True
@@ -161,10 +162,16 @@ def _raise_alarm(ip=None):
         if obj.alarm_raised is True:
             return False
 
-        reason = "NTP configuration does not contain any valid "
-        reason += "or reachable NTP servers."
+        if obj.peer_selected:
+            reason = "NTP cannot reach external time source; " \
+                     "syncing with peer controller only"
+            fm_severity = fm_constants.FM_ALARM_SEVERITY_MINOR
+        else:
+            reason = "NTP configuration does not contain any valid "
+            reason += "or reachable NTP servers."
+            fm_severity = fm_constants.FM_ALARM_SEVERITY_MAJOR
+
         eid = obj.base_eid
-        fm_severity = fm_constants.FM_ALARM_SEVERITY_MAJOR
 
     else:
         reason = "NTP address "
@@ -725,7 +732,14 @@ def read_func():
             cols = obj.ntpq[i].split(' ')
             ip = cols[0][1:]
             if ip:
-                if _is_controller(ip) is False:
+                ip_family = _is_ip_address(ip)
+                obj.peer_selected = _is_controller(ip)
+                if ip != obj.selected_server and obj.alarm_raised is True:
+                    # a new ntp server is selected, old alarm may not be
+                    # valid
+                    _clear_base_alarm()
+                    obj.alarm_raised = False
+                if obj.peer_selected is False:
                     if obj.selected_server:
                         # done update the selected server if more selections
                         # are found. go with the first one found.
@@ -747,15 +761,12 @@ def read_func():
 
                     if refid not in ('', '127.0.0.1') and \
                             not _is_controller(refid) and \
-                            socket.AF_INET == _is_ip_address(ip):
+                            socket.AF_INET == ip_family:
                         # ipv4, peer controller refer to a time source is not
                         # itself or a controller (this node)
                         obj.selected_server = ip
                         collectd.debug("peer controller has a reliable "
                                        "source")
-                    else:
-                        collectd.debug("peer controller does not have a "
-                                       "reliable source")
 
         # anything else is unreachable
         else:
@@ -796,7 +807,10 @@ def read_func():
             _clear_base_alarm()
 
     elif obj.alarm_raised is False:
-        collectd.error("%s no selected server" % PLUGIN)
+        if obj.peer_selected:
+            collectd.info("%s peer is selected" % PLUGIN)
+        else:
+            collectd.error("%s no selected server" % PLUGIN)
         if _raise_alarm() is False:
             obj.selected_server_save = 'None'