The alert
{severity} - {customer} - {site} ISP link is unstable
The cloud management server has noticed that “Hello” messages from the gateway are dropping out for short periods (under 15 minutes at a time), but the link as a whole isn’t fully down. End users typically experience this as a flaky or slow Internet connection. If the pattern persists, contact Uplevel support for a closer look.
What triggers it
The ISP Unstable alert fires when the gateway’s health-check probes fail repeatedly but recover within a five-minute window each time. The failures aren’t long enough to count as a full outage, but they’re frequent enough that something is clearly wrong with the upstream service.
A one-off occurrence is usually nothing — minor disruptions on the ISP’s side are common. A consistent pattern over hours or days is the signal that the link itself is degraded.
How to triage
- Check the ISP. Contact the ISP’s support line or check their public service-status page. They may already have an incident open.
- Measure stability over time. If the pattern continues, monitor the link with a continuous probe. PingPlotter is the tool we typically reach for; Uplevel can provide a short-term PingPlotter Agent for triage. For long-running monitoring, take it up directly with PingPlotter as a customer.
- Escalate to the ISP with data. If the alerts continue and PingPlotter shows packet loss or latency spikes pointing upstream of the gateway, open a ticket with the ISP and include the timestamps of the alerts and the PingPlotter results. With concrete data the ISP can run diagnostics against the customer’s circuit.
Periodic ISP wobble is part of normal operation. The goal isn’t to drive the alert count to zero, but to catch the cases where the wobble is sustained or worsening before it turns into a real outage.