fix(unifi): the Overview's history buckets on Postgres; devices name their state (#5207)
CI / lint (push) Successful in 2s
CI / integration (push) Successful in 1m24s
CI / unit (push) Successful in 1m35s
CI / publish (push) Successful in 1m29s

#5207 "UniFi Overview tab: func.strftime is SQLite-only": the WAN and
client sparklines bucketed with strftime('%s'), which Postgres lacks, so
/plugins/unifi/rows failed. Traefik's date_bin bucket moves to
core/time_range.time_bucket and both plugins use it. Devices list offline
and failing first, and a state other than online is named (Upgrading,
Adopting, Heartbeat missed...) instead of drawn as Down.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
2026-10-08 17:28:19 -04:00
co-authored by Claude Opus 5.5
parent 46a301652b
commit ea31deabf9
6 changed files with 145 additions and 49 deletions
+12
View File
@@ -4,6 +4,8 @@ from __future__ import annotations
from datetime import datetime, timedelta, timezone
from typing import TypeVar
from sqlalchemy import func
RANGES: dict[str, timedelta] = {
"1h": timedelta(hours=1),
"6h": timedelta(hours=6),
@@ -33,6 +35,16 @@ def bucket_seconds(since: datetime, target_points: int = 80) -> int:
return max(60, int(range_secs / target_points))
def time_bucket(column, seconds: int):
"""The start of the `seconds`-wide window a timestamp falls in, aligned to
the epoch, as a Postgres timestamp labelled "bucket". Group and order by
it. (An integer epoch divided by an int rendered as numeric division in
Postgres, so no two scrapes shared a bucket, and overflowed int4 after
2038 (#5256); strftime('%s') is SQLite-only (#5207).)"""
return func.date_bin(func.make_interval(0, 0, 0, 0, 0, 0, int(seconds)), column,
func.to_timestamp(0)).label("bucket")
T = TypeVar("T")