- support@haventech.guru
- Mon - Fri: 6.30 am - 5.00 pm
We are creative, ambitious and ready for challenges! Hire Us
Alert Management
AI Labs / NetCentral / Alert Management
AI Labs · NetCentral · Alert ManagementAck, Snooze, Resolve, Standby:
what holds and what comes back.
When an engineer deals with an alert, the console should show exactly what they did, and nothing should quietly come back as a new alarm. NetCentral keeps two things apart: what the monitoring says, and what a person decided.
Talk to us: 312-432-4480Email usThe idea
Two things are kept apart
Most alert tools blur them together, which is how a link that is down on purpose ends up paging someone every night.
The condition
What the monitoring says right now
Reported by the sources NetCentral reads: is the link, port or device actually down?
- Only the sources can change it
- A fault still being reported comes back, whatever anyone clicked
The management state
What a person decided about it
Acknowledged, snoozed, on Standby or resolved: a human judgment, recorded with a name.
- Ack, Snooze and Standby change only this
- Resolve claims the condition is over, and is checked first
At a glance
Where each action leaves an alert
Open incidenta source reports a fault
Ack / Snoozeon the board, owned, still scoredSnooze ends after 1, 4 or 24 h and it is open again
Standbyrests: off the board, still polledReported when it wakes; reviewed after 90 days
Resolveclosed, with its signalsStill reported? A new incident opens on the next poll
The four actions
What each one does, and what it holds
AckHolds while reported
“I have this.” The incident is marked as yours, and every poll keeps the same incident, so it stays acknowledged while the fault lasts.
Still on the dashboard and in the score, with a check mark.Snooze1, 4 or 24 hours
Like Ack, for a fixed time. When it runs out, the incident is open again by itself.
Quiet for the time chosen, then back.StandbyDown on purpose
For something that is down by design or no longer used. Its incidents rest: still polled, still recorded, reported if it wakes. A reason is required and every Standby is reviewed after 90 days.
Off the board and out of the score, at once.ResolveIt is fixed
Closes the incident and its signals. If a source still sees the fault, the next poll opens a brand-new incident, so Resolve is checked before it acts.
Leaves the board; nothing is deleted.Pressing Resolve
Checked before it acts
Resolving a fault that is still being reported only makes it come back as a new alert, so NetCentral looks first and offers the action that will actually hold.
You press Resolve. Has a source reported this fault in the last 10 minutes?
Yes: still being reported
It would be back within a minute
The window says so and leads with Put on Standby (down on purpose) or Ack instead (you are working it). Resolve is offered last, as “Resolve anyway”.
No: only leftovers
Old log lines waiting to age out
They are listed with their ages; Resolve closes them and the incident leaves the board.
No open signals
Nothing is reporting it
Resolved at once.
Standby, in one click
Down on purpose should look calm, not like a meltdown
A VPN tunnel that is no longer used but still configured, a spare link, a box being repaired: put it on Standby and the dashboard, the score and the ticker clear at once, while NetCentral keeps listening in case it wakes.
Standby…
From any incident window, the incident page, or the device page.
Refusals first
Anything that cannot rest is said before you type a reason.
Reason and mode
Automatic or Once; other links down on the same device come pre-ticked.
Clears at once
The incidents rest in the same click and the whole screen repaints.
Audited, reviewed
Every Standby has a name, a reason, and a review after 90 days.
Safety rails
What Standby will never do
Never half of a failover pair
The pair keeps its both-lines-down alarm, so a real double failure is never silenced.
Several devices: all or nothing
An incident spanning several devices rests only when every device in it rests.
Every action says what really happened
A refused action says so, and only an action that was done writes the audit row.
Operators only
Viewers can see every state; only operators and admins can change one, and each change carries their name.
Found and fixed
Why an alert used to come back
The director spotted alerts returning after they had been dealt with. The AI crew traced each cause, fixed it, and drew the logic as an infographic inside NetCentral, so the explanation stays tied to the code that does it.
| Found in testing | What an operator saw | Now |
|---|---|---|
| Resolve on a link that was still down | It closed, then reopened as a new incident seconds later, again and again | The window warns first and offers Standby or Ack |
| Acknowledge on a device page closed the signal | The alert vanished, then returned as a new incident | It acknowledges the incident, which holds |
| Standby cleared one link at a time | The board stayed red until the next poll | Rests and repaints at once; links down on the same device are offered too |
| Standby refused after typing a reason | A message, and nothing saved | Any refusal is said first, with what can rest |
| One dead port watched twice | Two critical alerts for one fault | One incident, one name |
How this was built
Directed by a person, built by an AI crew
- The director described the problem in his own words and approved each fix
- Claude Opus 5.5 (Anthropic) traced the causes, wrote the code and the tests
- Each fix ships with an automated check that fails if it ever regresses
- The in-app infographic is re-checked against the code on every build
Let’s talk
Want alerts your team can trust?
Call or email and a Haventech engineer will walk you through what AI can do for your business. Mon – Fri, 6:30 am – 5:00 pm.