Skip to content

Commit 1638148

Browse files
jdonelsonclaude
andcommitted
console: correct the freshness kept-row note
The kept-row count does not depend on the recording interval: mz_wallclock_global_lag_history keeps one reading per object per minute. Co-Authored-By: Claude Opus 5.5 <[email protected]>
1 parent ea6fde4 commit 1638148

1 file changed

Lines changed: 4 additions & 4 deletions

File tree

‎console/src/api/materialize/freshness/lagHistory.ts‎

Lines changed: 4 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -152,10 +152,10 @@ export function rankFromWorst(p: number, n: number) {
152152
* The most readings one object can have in a range: one a minute, plus one
153153
* because both ends of the window can hold a reading.
154154
*
155-
* NOTE: assumes `wallclock_lag_recording_interval` is at its 60 s default. An
156-
* environment that records more often has more readings than this, so the
157-
* percentile picks fall past the rows the queries keep and come back NULL,
158-
* which reads as unreadable.
155+
* NOTE: holds whatever `wallclock_lag_recording_interval` is.
156+
* `mz_wallclock_global_lag_history` truncates each reading to its minute and
157+
* keeps one per object per minute, so recording more often adds no rows here,
158+
* and recording less often only leaves fewer.
159159
*/
160160
function maxReadingsPerObject(lookbackMs: number) {
161161
return lookbackMs / LATEST_READING_INTERVAL_MS + 1;

0 commit comments

Comments
 (0)