InnerWeb
Skip to main content

Device log: 89-entry window and diagnostic gaps

The employee watch records its event log locally and uploads it to the server in batches, so after a long connection gap the beginning of the record drops out of the queue. For the IT and OT team this sets a concrete boundary: what can be reconstructed after the fact, and what must be secured in another way.

89
entries per batch and in the device queue
Source: karta
78 s
log upload rhythm with a 26 s heartbeat
Source: karta
approx. 2 h
reconstructable connection gap
Source: karta

What the log records and where it goes

The watch records an event log: sensor decisions, the course of an alarm and heartbeat transmissions. Entries are uploaded to the server in batches, which makes it possible to reconstruct the course of an alarm after the fact, even when nobody was watching the panel at the moment of the event.

The heartbeat is a periodic message from the device: beacons it hears, signal strength, battery status and state. The platform knows the employee's position from it without GPS. The log is a separate, richer record of what the device did and why, so it is the source that answers why an alarm did or did not trigger.

The window arithmetic: 26 seconds, 78 seconds, 89 entries

The log is sent every third heartbeat cycle. With a 26 s interval this gives an upload roughly every 78 s, and one batch holds up to 89 entries. The queue on the device also holds 89 positions. While the connection works, the queue is emptied continuously and the server has an almost complete picture.

When the connection disappears, the device still records events, but the queue is limited. After about two hours without a connection the oldest entries are pushed out by newer ones and the beginning of the gap disappears. This is not a defect, but a deliberate trade-off between watch memory and completeness of the record.

For the diagnostician the conclusion is simple: the log on the server is reliable for roughly the last two hours of an outage and for everything that happened after the connection was restored. Longer outages leave a gap at the beginning, which has to be closed with another source.

Parameters to keep in mind

Keep them in one place in the maintenance manual so that a technician does not have to calculate them each time.

  • heartbeat interval: 26 s
  • log upload: every third cycle, approx. 78 s
  • batch: up to 89 entries
  • device queue: 89 positions
  • reconstructable gap limit: approx. 2 hours

Watch time thresholds in seconds

Watch time thresholds in secondsHorizontal bar chart: heartbeat 26 s, log upload 78 s, stillness vibration 120 s, switch to pulse 233 s.050100150200250sHeartbeatHeartbeat: 26 s26Log uploadLog upload: 78 s78Stillness vibrationStillness vibration: 120 s120Switch to pulseSwitch to pulse: 233 s233
Source: Device log function sheet and the observed stillness alarm configuration

The stillness alarm works without the server

A gap in the log does not mean a gap in protection. The stillness alarm on the device works independently of the server. In the configuration observed, the watch vibrates after 120 s without movement, and after 233 s without confirmation from the server it switches to a short pulse every 34 s. It does not vibrate while charging or at rest.

From the IT and OT perspective this distinction matters. An employee in a zone without coverage still receives a signal on the wrist, even if the server has seen nothing for an hour. The server, however, recovers the description of the event only after the connection returns and only within the limits of the queue window.

These values describe the observed configuration, not fixed product values. Before rollout, record in the installation documentation the thresholds actually set at the given site.

How to organise diagnostics so that gaps do not hurt

The simplest rule: collect the report about a watch that does not respond right after the connection returns, not the next day. Every hour of delay on an active device means more entries pushing out the older ones.

The second rule concerns the order of sources. First the log from the server, then the last known position with a timestamp marked as outdated, then the battery state from the last heartbeat. Only when these three sources do not explain the matter does the device go to a physical inspection.

The third rule is a coverage map. If connection gaps at a given place on the site last longer than two hours, this is a signal to add an access point or to enter the place on the list of zones with limited diagnostics.

  • collect device logs right after the connection returns
  • mark in the documentation the zones where gaps may exceed two hours
  • compare the log with the last positions and the battery state
  • record the stillness thresholds set at the site
  • exhaust the server sources before a physical inspection

A typical set of five alarms as a reference point

Wearable devices usually handle five events: entering a restricted zone, leaving a zone, loss of communication with the tag, low battery and an emergency situation reported by the user. The last known position is shown as outdated with a timestamp.

When planning maintenance, assign each of these five events an owner on the site side and a place in the log. Then the discussion about a gap in the log ends with something concrete: which event fell within the window and which fell out of it.

Video

Frequently asked questions

Does an outage longer than two hours erase the whole log?

No. The beginning of the outage is lost, because the 89-position queue is pushed out by newer entries. Events from the end of the outage and after the connection returns remain.

Does the stillness alarm work when the server cannot see the watch?

Yes. The stillness alarm runs on the device independently of the server, so the employee receives the signal even without coverage.

How is the position known if the log has gaps?

From the heartbeat: beacons heard, signal strength, battery and status. The last known position is marked as outdated together with its timestamp.

Let's plan diagnostics for your site

We will show the device log on a live example and agree with your IT and OT team on thresholds, zones without coverage and the owner of each alarm.

Let's talk