You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Commit 852be00
Browse filesBrowse the repository at this point in the historyBrowse files
console: Take the freshness "Now" column from the latest reading
Readings are binned to 60 points per range, and each point is the worst
reading in its span. "Now" read the last point, so at a 24 hour range it
answered "the worst of the last 24 minutes" to the question "is this bad
right now". An object that spiked twenty minutes ago and recovered still
read high, and under the "right now" predicate it was counted as
exceeding the threshold and coloured on the graph.
A second query returns each object's newest reading, batched into the
same request. Its two minute filter keeps it cheap whatever range is
being graphed: measured on a 300 object, 24 hour fleet it adds 100ms to
a 499ms request, against 2,594ms for computing every statistic from
unbinned readings.
The graph still plots bin maxima, since that is all 60 points can show,
so a point can sit above its row's "Now" value. The footer says so, and
drops the mz_internal relation names, which are not a stable interface.
Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
0 commit comments