How Observability Measurements Work¶

The volume measurement is the container's row count.

Turning it on. Volume Tracking is a per-container toggle in the container's Observability Settings, and needs the Editor team permission. Turning it on reads the container once to confirm Qualytics can measure it: if that reading fails, the setting is not saved and the error explains why. Turning it off stops new measurements and also turns off Create AI Volumetric checks. Existing checks are kept.

The hourly measurement. Every hour at minute 30 UTC, Qualytics measures the containers that still need a row count for the current UTC day. Each container gets at most one count per UTC day from this cycle: the first successful reading sticks, and later ticks skip the container. On a large datastore it can take a few ticks before every container is measured, and a container missed by one tick is picked up by the next.

Other sources of counts. The hourly measurement is not the only source. These operations record a row count too, and on a day that already has one, the hourly measurement skips the container:

• Every Scan that evaluates a Volumetric check records the count it reads.
• A successful Profile, and a Full Scan that read every record, record the total row count they processed.

A day can therefore hold several counts. The Volumetric check uses their average. See The History the Check Needs.

When a measurement fails. After three consecutive failed measurements, the container is marked Inaccessible and the hourly cycle skips it. Aborted or timed-out readings do not restore it. Once access is fixed, run a Sync on the datastore: a successful Sync that finds the container marks it Available again, and the hourly cycle measures it from then on. A Sync does not discover a Computed Table, Computed File, or Computed Join, so for those, open the container, click Edit, then Validate and Save its definition.

The chart. The volume chart is a line, with one point per period of the Group By setting. Each point is the most recent count of its period.

The Min and Max lines belong to the selected check. For the change comparisons, they move with the data and start from the count one Measurement Period earlier. For Absolute Change, each point's limits are that count plus the check's Min and Max. For Percentage Change, they are that count raised or lowered by the Min and Max percentages: an earlier count of 1,000 with a Max of 20% puts the upper line at 1,200.

The point for the current period pulses in amber while its count is still being recorded, and the segment leading to it is dashed.

The tooltip. Hovering a point shows:

• Last row count, with an arrow and the difference from the previous point.
• Min threshold and Max threshold, when the check has them.
• An orange badge with the anomaly count, when the period has anomalies.

While the period is still being recorded, the count shows — and a note reads Measurements are still being recorded for the selected report date.

The freshness measurement is the container's last modification time, and from it the staleness: how long before the measurement the container last changed. Where that time comes from depends on the container and the platform. See Where the Last Modification Time Comes From.

Turning it on. Freshness Tracking is a per-container toggle in the container's Observability Settings, and needs the Editor team permission. Turning it on reads the container once to confirm Qualytics can measure it: if that reading fails, the setting is not saved and the error explains why. Turning it off stops new measurements and also turns off Create AI Freshness checks. Existing checks are kept.

The hourly measurement. Every hour at minute 30 UTC, Qualytics measures the containers that still need a freshness reading for the current UTC day, at most once per container per UTC day. The cycle tracks volume and freshness separately, so a container that already has a row count for the day can still be measured for freshness.

Other sources of readings. Every Scan that evaluates a Freshness check records the last modification time it reads. A Profile never records one.

When a measurement fails. The same rules as for volume apply: three consecutive failures mark the container Inaccessible, the hourly cycle skips it, and a successful Sync marks it Available again. For a computed container, edit, validate, and save its definition instead.

The chart. The freshness chart shows bars, with one bar per period of the Group By setting. Each bar is the staleness of the period's most recent measurement.

A yellow line marks the selected check's Maximum Age, when it is set.

The tooltip. Hovering a bar shows two groups:

• Max age, in compact form such as 2h 30m, and Staleness, shown in red when it is past the Maximum Age.
• Observed, when the measurement was taken, and Last modified, the container's last modification time, both in your browser's time zone.

A missing value shows as —, and an orange badge shows the anomaly count when the period has anomalies. The current period has no marker on the chart. While it is still being recorded, its tooltip shows — and the note Measurements are still being recorded for the selected report date.

The metric measurement is the value of one field on one row, recorded by a Metric check.

No tracking toggle. Metric measurements are not part of the hourly cycle and do not depend on Volume Tracking or Freshness Tracking. They come only from Scans: every Scan that evaluates an Active Metric check records its value, whether the check passes or fails. A Profile never records one. A Scan that finds no row, or more than one, records nothing, because the check is not asserted. See Exactly One Row.

The card. The Metrics tab shows one card per Metric check, with the check's most recent measurements (up to 11).

The chart. The metric chart is a line of the measured values, one point per period of the card's Group By setting, each point being the most recent value of its period. A value outside the limits is highlighted. The limits are drawn as a band:

• For Absolute Value, the band is the check's Min and Max.
• For Absolute Change, each point's band is the previous value plus the check's Min and Max.
• For Percentage Change, each point's band is the previous value raised or lowered by the Min and Max percentages.

In both change comparisons, the first measurement has no previous value, so the chart starts from the second.

The chart needs a minimum number of measurements before it draws, and shows No measures until then:

The tooltip. Hovering a point shows the Measurement and the Min threshold and Max threshold that apply to it. A missing value shows as —. There is no anomaly count badge: a value outside the limits is shown by its color.