Severity & Status Codes
GridNMS shows severity in three places, and they do not use the same scale. Check which one you’re reading before comparing two numbers:
| Where you see it | Scale | Direction |
|---|---|---|
| Events, cases, and detections | Critical · Major · Minor · Warning · Info | 1 is the most severe |
| A log line in Log Search | Unknown · Informational · Low · Medium · High · Critical · Fatal | 6 is the most severe |
| Thresholds, and the Logs chart legend | Critical · Error · Warning · Info · Debug | 1 is the most severe |
Event severity
Section titled “Event severity”Used for events, cases, and detections:
| Level | Name | Color | Use |
|---|---|---|---|
| 1 | Critical | Red | Total outage, link down |
| 2 | Major | Orange | Serious but non-total impact |
| 3 | Minor | Amber | Threshold breach, degraded state |
| 4 | Warning | Yellow | Worth noting, not yet a problem |
| 5 | Info | Blue | Routine informational (a device came back up, a link came up) |
When a detection has a severity threshold, you’re choosing the maximum level to be
notified about — a threshold of 3 notifies on Critical, Major, and Minor.
Log severity
Section titled “Log severity”Each line in Log Search carries its own severity, which comes from the log data model rather than the event scale above. It runs the opposite way — the number climbs as the line gets more serious:
| Level | Name |
|---|---|
| 0 | Unknown |
| 1 | Informational |
| 2 | Low |
| 3 | Medium |
| 4 | High |
| 5 | Critical |
| 6 | Fatal |
So a log line marked 5 is Critical, while an event marked 5 is Info. The badge
beside each log line always spells the name out, so read the name rather than the
number when you are moving between the two pages.
Threshold severity
Section titled “Threshold severity”When you set a threshold — on an interface, a container, or a metric — you pick how serious a breach should be from a third, shorter list:
Critical · Error · Warning · Info · Debug
The same five names label the bars in the chart above your log search results.
Events are immutable
Section titled “Events are immutable”An event is a point-in-time record of something that happened — it’s written once and never changes, so there’s no open/acknowledged/closed status on an event itself. The ongoing investigation lifecycle (open → investigating → resolved → closed) lives on Cases: a detection can group related events into a case, and that case is what you triage. An event that recovers (for example a device coming back up) is its own new event, not an edit to the original one — the two are linked by a shared tag so you can see the down/up pair together. See Events & Alerts.
Device reachability
Section titled “Device reachability”Alive value |
Meaning |
|---|---|
| 1 | Up — a collector reported the device reachable |
| 0 | Down — a collector reported it unreachable |
| 2 | Unknown — added, but no collector has reported on it yet |
| (none) | Unknown |
A device is down only when a collector said so (0). A device that has been
added but never reported on carries 2, which is the usual value on a new
instance and on any device no collector owns — it is not a problem state. If you
read this field in your own integration, treat anything that is not 1 or 0 as
unknown rather than as a yes/no: a plain truth test counts 2 as up and will
report devices that have never been checked as healthy.
Alive: 1 also goes stale. If no collector has reported on the device in the
last five minutes the interface shows it as Stale rather than Up, because a
silent collector is not evidence the device is still reachable.
Common event tags
Section titled “Common event tags”| Tag | Raised by | Meaning |
|---|---|---|
PING_DOWN |
Device Down detection | Device became unreachable (Critical) |
PING_UP |
Device Down detection | Device returned (Info); pairs with PING_DOWN |
See Events & Alerts for how detections turn raw signal — reachability, SNMP traps, log lines — into tagged events, and how to author your own.
docs built 2026-11-12 · 945a2dec2