MCX Monitor In development
Every site: green, minor alarm or major alarm.
MCX Monitor watches the network under the radios: base stations, site controllers, routers, links and power. It is being built now. This page says what is finished, what is next and how it is put together.
A small collector at each site, one core in the cloud
The collector does the local work and pushes what it finds up a connection it opens itself. There is no VPN and nothing inbound to open at a customer.
- No VPN
- The collector opens one encrypted connection out on port 443. It works behind NAT, cellular and strict firewalls.
- A site that goes quiet
- If the backhaul drops, the collector keeps polling and queues what it finds. The core shows that site grey for unreachable, not red.
- One collector per network
- A large radio network with isolated sites gets one at each site, since an alarm cannot cross a dead link.
Built today: the collector
The collector is finished and tested. It installs as a Windows service on a server the site already has.
- SNMP polling
- Versions 2c and 3, on a schedule, with a profile for each device model.
- SNMP traps
- Alarms the equipment pushes arrive in under a second. Versions 1, 2c and 3.
- Syslog
- A listener for the log lines equipment sends.
- A queue on disk
- Nothing is lost while the link to the core is down.
- Encrypted link
- One outbound connection to the core. Each collector has its own token, which can be revoked at once.
- Installer
- Registers the service and adds the firewall rules it needs. Uninstalling removes both.
Next: the core
The core receives and stores what collectors send today. The alarm engine, map and logins below are designed and approved, and are being built against real equipment.
- Radio infrastructure
- P25 base stations and site controllers first, through their vendor MIB files. DMR equipment follows.
- Routers
- Customer routers through the manufacturer's cloud management API, with direct SNMP as the fallback.
- Everything else
- Switches, UPS units and microwave links through standard and vendor MIBs, and ping and port checks for anything without SNMP.
- Templates and discovery
- A saved set of monitors for each model, applied in one click, and a subnet scan that suggests the right one.
Alarms that mean something
The hard part of monitoring is not raising alarms. It is raising one alarm for one problem.
- Two levels
- Every monitor has a minor and a major threshold, shown yellow and red.
- Site rollup
- A site shows the worst state of anything inside it.
- Dependencies
- If the site router is down, the equipment behind it shows unreachable and does not page anyone.
- Flap damping
- Several failed polls in a row before an alarm, so a flaky link does not cause a storm.
- Acknowledge and maintain
- Acknowledge an alarm with a note, and schedule maintenance so planned work stays green.
- Notifications
- Email, text, push and webhook, routed by severity, customer and time of day, escalating if nobody acknowledges.
- Call denies
- An alarm when a radio site denies more than a set number of calls within a set number of minutes.
- Tunnel latency
- An alarm when the active link of a bonded tunnel is slow. An idle backup link that is slow is logged, not alarmed.
- Two sources disagree
- When the router's cloud says online and the collector cannot reach it, that is its own alarm.
Views for the technician, the wall and the customer
- Site map
- Sites as pins on a real map in green, yellow, red or grey. Click one to see its equipment.
- Site diagram
- The equipment at a site drawn with live status.
- Wall board
- Full screen, no sign-in timeout, the map beside the active alarms.
- Customer logins
- A customer sees only their own sites, view only. Your technicians see everything.
- History and graphs
- Signal levels, temperatures and latency kept and graphed over any range.
- Reports
- A searchable alarm log and monthly uptime for each site and customer.
Want it at your sites first?
We are choosing the first networks now. Tell us what equipment you run.
Sentinel MCX