bonamin/bellsystems-cp
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix(mqtt): dedupe boot_report against the latest row, not any boot_count

Browse Source 
98dd16b skipped a boot_report if ANY earlier row had the same boot_count.
That is wrong: the firmware's lifetime boot counter gets reset (reflash /
telemetry reset), and the data shows counts 1-4 recurring in July and again
in September. With that rule a real later boot reusing a number would be
dropped forever.

A retained redelivery is always a copy of the device's most recent boot, so
compare only against the latest row (boot_count + reset_reason). A genuine
new boot always differs from it - the counter moves forward or was reset.

Note: the one-off cleanup run on 2026-09-30 used the same wrong
(serial, boot_count) key and deleted some genuine boot rows along with the
redelivery duplicates; see the session notes / heartbeat-based reboot list.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
bonamin
2026-09-30 15:55:26 +03:00
co-authored by Claude Opus 5.5
parent 98dd16b597
commit 18e57c6e5f
1 changed files with 18 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files
backend/database/pg_mqtt.py
+18 -10
View File
@@ -445,22 +445,30 @@ async def insert_boot_event(device_serial: str, boot_count: int | None,
crash_pc: int | None = None,
crash_exc_cause: int | None = None,
crash_exc_vaddr: int | None = None) -> int | None:
"""Insert a boot event, unless one with the same boot_count already exists
for this device (returns None then). boot_report is published retained on
system/info, so the broker redelivers the last one every time the backend
(re)subscribes — without this check each backend restart logged the
device's last boot (crash detail included) again as a brand-new event."""
"""Insert a boot event, unless it is a repeat of the device's most recent
one (returns None then). boot_report is published retained on system/info,
so the broker redelivers the last one every time the backend (re)subscribes
— without this check each backend restart logged the device's last boot
(crash detail included) again as a brand-new event.
Only the LATEST row is compared, never "any row with this boot_count": the
firmware's lifetime counter gets reset (reflash / telemetry reset), so the
same boot_count legitimately appears again for a later, different boot. A
genuine new boot always differs from the latest row — the counter either
moved forward or was reset to a lower number."""
async with AsyncSessionLocal() as session:
if boot_count is not None:
dup = await session.execute(
latest = await session.execute(
text("""
SELECT 1 FROM device_boot_events
WHERE device_serial = :serial AND boot_count = :boot_count
SELECT boot_count, reset_reason FROM device_boot_events
WHERE device_serial = :serial
ORDER BY occurred_at DESC
LIMIT 1
"""),
{"serial": device_serial, "boot_count": boot_count},
{"serial": device_serial},
)
if dup.first() is not None:
row = latest.first()
if row is not None and row.boot_count == boot_count and row.reset_reason == reset_reason:
return None
result = await session.execute(
text("""