preventative maintenance

Most outages announce themselves weeks ahead.

A full disk, a service that restarts every night, a machine three releases behind, a mapped drive that quietly stopped mapping. BE Hub looks for these every night and every heartbeat, and only speaks when something needs a decision.

every night

The nightly pass.

Once a day, at a time you choose per company, BE AI reviews every included machine that is online.

Reads a day of history first

Heartbeats, events, check results, the last pass. Cheap, instant, and enough to know what changed.

Then looks at the machine

Diagnostics, services, recent errors, checks, network facts. Read-only.

Speaks only when it must

Healthy machine: nothing to say, nothing charged beyond the read. Something drifting: a short note, and a proposal if a command would fix it.

every heartbeat

What is watched without asking.

Disks with a date

Growth per day from the last week and a day the disk will be full. On the dashboard, on every fleet row, and in the alert when it comes.

Memory and swap creep

Trends per machine over 7 and 30 days, so a leak shows as a slope, not a crash.

Services and reboots

A service that stops, a reboot that was not planned, a machine that comes back on an old release.

Custom checks every minute

Mapped drives, printers, VPN peers, file age, free space, listening ports. Drift is caught before a person notices.

Behind on releases

Machines not on the latest signed agent, and a roll-out to fix it by tag or company.

Degraded storage

A ZFS pool that goes from ONLINE to DEGRADED is a critical alert before the second disk fails.

on request or on a schedule

Recipes a technician would run if there were time.

Patch health

Pending reboot, failed updates, last successful update, uptime. Ask for it on a machine or a whole company.

Service health

Stopped auto-start services that should be running, crash events, restarts proposed only where a service clearly belongs up.

Health summary

"Give me a health summary of this machine": patching, disk, memory, services, and anything from the event log worth a look.

Verify the effect

After any approved change, one click asks BE AI to check the machine again and confirm what changed.

questions

About the nightly pass.

Does the nightly pass cost a BE AI run per machine per night?

Only for machines that need one. A machine that was healthy at its last pass and has had no events or alerts since is skipped and re-checked every three days. On a quiet fleet most nights cost a handful of runs.

Can I pick the time and the machines?

Per company: on or off, the time and time zone, the days. Per machine: in or out of the pass, in or out of BE AI entirely.

What if it proposes something I do not want done?

Nothing runs without an approval. Reject it, or reply in the thread and say what you would rather do. The nightly pass never acts on its own.

Is this patch management?

No. It tells you patch state, pending reboots and failed updates, and can propose a restart you approve. Deploying patches stays with the tool you use for that.

Let it run for a week on ten machines.

By the second morning the dashboard lists what will break first. Nothing is changed without you.

Start a pilot Talk to Binary Elements