Opsview Alert Integration

​

Configure in Opsview

​

Step 1: Create the notification script

#!/bin/sh
# Opsview custom notification method: forward host/service notifications to a
# Flashduty Standard Alert Event integration. The push URL is the first argument.
URL="$1"
[ -n "$URL" ] || exit 1

case "$NAGIOS_NOTIFICATIONTYPE" in
  PROBLEM|RECOVERY) ;;
  *) exit 0 ;;   # ignore acknowledgements, flapping and downtime notifications
esac

if [ -n "$NAGIOS_SERVICEDESC" ]; then
  KEY="opsview:$NAGIOS_HOSTNAME:$NAGIOS_SERVICEDESC"
  STATE="$NAGIOS_SERVICESTATE"
  OUTPUT="$NAGIOS_SERVICEOUTPUT"
  CHECK="$NAGIOS_SERVICEDESC"
  TITLE="$NAGIOS_HOSTNAME - $NAGIOS_SERVICEDESC - $STATE"
else
  KEY="opsview:$NAGIOS_HOSTNAME"
  STATE="$NAGIOS_HOSTSTATE"
  OUTPUT="$NAGIOS_HOSTOUTPUT"
  CHECK="host"
  TITLE="$NAGIOS_HOSTNAME - $STATE"
fi

if [ "$NAGIOS_NOTIFICATIONTYPE" = "RECOVERY" ]; then
  STATUS=Ok
else
  case "$STATE" in
    CRITICAL|DOWN|UNREACHABLE) STATUS=Critical ;;
    WARNING)                   STATUS=Warning ;;
    OK|UP)                     STATUS=Ok ;;
    *)                         STATUS=Info ;;
  esac
fi

esc() { printf '%s' "$1" | tr '\n\r\t' '   ' | sed 's/\\/\\\\/g; s/"/\\"/g'; }

JSON=$(printf '{"title_rule":"%s","event_status":"%s","alert_key":"%s","description":"%s","labels":{"resource":"%s","check":"%s","host_address":"%s","host_alias":"%s","opsview_keywords":"%s"}}' \
  "$(esc "$TITLE")" "$STATUS" "$(esc "$KEY")" "$(esc "$OUTPUT")" "$(esc "$NAGIOS_HOSTNAME")" "$(esc "$CHECK")" \
  "$(esc "$NAGIOS_HOSTADDRESS")" "$(esc "$NAGIOS_HOSTALIAS")" "$(esc "$OPSVIEW_KEYWORDS")")

curl -fsS -m 10 -X POST -H 'Content-Type: application/json' -d "$JSON" "$URL" >/dev/null

​

Step 2: Install and test the script

sudo -u opsview /opt/opsview/orchestrator/bin/orchestratorimportscripts notifications /path/to/notify_by_flashduty

# Host problem
sudo -u opsview /opt/opsview/coreutils/utils/test_notifications hostproblem /path/to/notify_by_flashduty '<push URL>'
# Service problem
sudo -u opsview /opt/opsview/coreutils/utils/test_notifications serviceproblem /path/to/notify_by_flashduty '<push URL>'

​

Step 3: Add the notification method

1. Go to Configuration → Apply Changes to finish installing the script
2. Go to Configuration → Notification Methods and click Add New
3. Enter Flashduty as Name, tick Enable, and choose whether Run on is the Orchestrator or the Collector
4. Enter notify_by_flashduty <push URL> as Command (scripts are read from /opt/opsview/monitoringscripts/notifications by default, and arguments follow the script name)
5. After saving, go to Configuration → Apply Changes again to activate the new notification method

​

Step 4: Add the method to a notification profile

1. Go to Configuration → Users and Roles → Users and edit the user (or open My Profile from your username at the top right), open the Notification Profiles tab and click Add New
2. Select Flashduty under Alert me by and the active time period under During
3. Under Host and Services, select the host groups and service checks to notify about and tick the states to notify on; include both the problem states and the recovery states, otherwise Flashduty receives no recovery notification
4. Click Update, then Submit Changes, and finally go to Configuration → Apply Changes to make it live

​

Recovery and deduplication

• A host or service always uses the same Alert Key, so repeated problem notifications (for example from Opsview’s re-notification interval) only update the same alert, and the RECOVERY notification recovers it
• When a service goes from WARNING to CRITICAL, the severity is updated on the same alert
• Acknowledgement (ACKNOWLEDGEMENT), flapping (FLAPPING*) and downtime (DOWNTIME*) notifications are ignored by the script and produce no event in Flashduty

​

Troubleshooting

• No alert arrives: confirm the Notification Profile selects Flashduty and Apply Changes was run; check in Opsview’s Notifications View whether the notification was sent; test the script on its own with test_notifications
• The test works but real alerts are not pushed: in a real notification the script reads the environment variables Opsview passes; add { date; echo "Called with $@"; env; echo; } >> /tmp/env.txt to the top of the script (the debugging method given in the Opsview documentation) to see what it actually receives
• The alert does not recover: confirm the Notification Profile ticks the recovery states; Nagios only sends a recovery notification after a problem notification was sent
• Flashduty returns a parameter error: confirm the push URL in Command is complete, including integration_key