Transformations
Not everything that arrives at GridNMS should be treated exactly as-is. A chatty informational trap might deserve a lower severity, a known cosmetic warning shouldn’t page anyone, and a firehose of debug-level logging doesn’t need to be stored forever just because it exists. Transformations is where you shape that — two independent rule engines, reached from one page, that act before their data settles into its final place:
- Event Rules reshape an incoming event — tag it, change its severity, or suppress it — before it lands on Events & Alerts.
- Log Rules cut ingest volume — drop or divert matching log records — before they’re ever written to storage, so noisy sources don’t cost you search performance or retained storage.
Where to find it
Section titled “Where to find it”Open Configure → Logs and Events and select the Transformations tab. Two sub-tabs switch between the two engines: Event Rules and Log Rules. Each list is evaluated in order, and the two differ in what happens after a match: Log Rules stop at the first matching rule, while Event Rules keep going (see below).
Event Rules
Section titled “Event Rules”Every inbound event — from a device check, an SNMP trap, or a log detection — passes through Event Rules before it’s written to the feed. Rules run in the order they were created, oldest first, and every rule that matches is applied, so a later rule sees what an earlier one did: a tag it set, a severity it changed. A Suppress rule ends it — the event is dropped and the rules after it never run.
Rules apply to events as they are created. Events already in the feed are never rewritten, and a rule you add, edit or delete takes effect from the next event.
Match conditions
Section titled “Match conditions”- Message Filter — a regular expression tested against the event’s full message text.
- Tag Filter — a regular expression tested against the event’s current tag (set by the source, or by an earlier rule in the list).
- Match Condition — Either fires the rule if the message or the tag matches; Both requires both to match, for narrower targeting. Leaving a filter blank means it always matches.
Actions
Section titled “Actions”| Action | What it does |
|---|---|
| Set Tag | Attach a tag string to the event, for filtering and for later rules to key on. |
| Set Severity | Override the event’s severity — for example, promote a warning to critical, or downgrade a chatty trap to Info. |
| Suppress | Drop the event entirely. It’s never stored or shown anywhere — it doesn’t open or join a case and it sends no notification. |
Events are permanent records and are never closed, so there is no rule action for closing one. To group related events and work them to a conclusion, use a case.
Click New Rule to open the editor, or click an existing rule to edit it. The status badge toggles a rule Active/Inactive without deleting it — handy for disabling something during a maintenance window.
Suppress only decides whether the event is recorded. GridNMS still knows the thing it described happened: if you suppress a “device down” event, the device is still treated as down, and the matching “back up” event is still raised when it recovers (unless a rule suppresses that one too). Suppress both halves if you want neither in the feed.
Common examples
Section titled “Common examples”The editor offers a set of ready-made examples you can start from and adjust: suppressing routine “configuration saved” syslog spam, tagging link-down events for easier filtering, and escalating UPS low-battery traps to critical.
Log Rules
Section titled “Log Rules”Log Rules act earlier in the pipeline than Event Rules — on the raw ingested log record, before it’s written to storage at all. Each rule matches on a sourcetype (assigned by a parser) and an optional message filter regex over the raw line, then takes an action.
Actions
Section titled “Actions”| Action | What it does |
|---|---|
| Drop | Discard the matching record before it’s ever written to storage. |
| Route internal | Keep the record, but divert it to the hidden internal index instead of your searchable logs. |
Sampling and dry run
Section titled “Sampling and dry run”- Sample 1-in-N — instead of acting on every match, keep a deterministic slice and apply the action to the rest. For example, N = 100 keeps 1 record in every 100 (a load balancer’s health-check hits, say) and drops the other 99. N = 1 (the default) acts on every match.
- Dry run — a dry-run rule is evaluated and counted, but never actually drops or routes anything. New rules default to dry run, so you can see a rule’s blast radius from its counters before arming it for real.
Live counters
Section titled “Live counters”The list shows a running Matched / Acted count per rule (with a kept count alongside it when sampling is in play), refreshed automatically — so what a rule is discarding is never invisible, even though the discarded records themselves never reach storage.
Common examples
Section titled “Common examples”Starter examples cover dropping DEBUG-level syslog chatter, sampling
load-balancer health-check spam down to 1-in-100, and routing sudo/audit lines
to the internal index instead of the tenant log search.
Where to go next
Section titled “Where to go next”- Assign the sourcetypes Log Rules match against in Parsers.
- Pull structured fields out of what makes it through in Field Extraction.
- Write rules that watch what’s left in Detections.
- See how retention and the cold tier fit in on Log Storage.
- Triage the events Event Rules shape on Events & Alerts.
docs built 2026-11-12 · 945a2dec2