# bms-1 (94.23.26.113) — Netdata health override for netdata.dbengine_global_errors
# Deployed to: /etc/netdata/health.d/dbengine.conf   (user health.d overrides the stock
#   /usr/lib/netdata/conf.d/health.d/dbengine.conf at the FILE level — so this file must
#   contain ALL THREE alarms, not just the tuned one, or FS/IO paging would be lost.)
# Issue: radieu/p24-infra#4242
#
# WHY: the stock 10min_dbengine_global_flushing_errors alarm counts transient in-flight page
# flushes that miss their window during Netdata's own dbengine datafile rotation (~every 17 min
# on bms-1). Those spikes (~410–823 errors/10min observed) tripped CRITICAL and paged a human on
# every rotation cycle, even though I/O errors and FS errors stayed 0 and the disk/RAID were
# healthy (self-recovering, no data loss).
#
# FIX (issue option 2): keep CRITICAL paging on the two dimensions that mean REAL trouble
# (I/O errors, FS errors — copied verbatim from stock), and demote the flushing-errors dimension
# to a high-threshold, non-paging (to: silent) informational WARNING. A genuinely failing disk
# surfaces first as I/O errors, which still page CRITICAL.
#
# Netdata v1.19.0 on bms-1 has no `netdatacli`; reload health with:  killall -USR2 netdata

 alarm: 10min_dbengine_global_fs_errors
    on: netdata.dbengine_global_errors
    os: linux freebsd macos
 hosts: *
lookup: sum -10m unaligned of FS errors
 units: errors
 every: 10s
  crit: $this > 0
 delay: down 15m multiplier 1.5 max 1h
  info: number of File-System errors dbengine came across the last 10 minutes (too many open files, wrong permissions etc)
    to: sysadmin

 alarm: 10min_dbengine_global_io_errors
    on: netdata.dbengine_global_errors
    os: linux freebsd macos
 hosts: *
lookup: sum -10m unaligned of I/O errors
 units: errors
 every: 10s
  crit: $this > 0
 delay: down 1h multiplier 1.5 max 3h
  info: number of IO errors dbengine came across the last 10 minutes (CRC errors, out of space, bad disk etc)
    to: sysadmin

 alarm: 10min_dbengine_global_flushing_errors
    on: netdata.dbengine_global_errors
    os: linux freebsd macos
 hosts: *
lookup: sum -10m unaligned of flushing errors
 units: errors
 every: 10s
  warn: $this > 3000
 delay: down 1h multiplier 1.5 max 3h
  info: dbengine failed to fully flush pages to disk in the last 10 minutes. On bms-1 short bursts during dbengine datafile rotation (~every 17 min, ~410-823 errors observed) are EXPECTED and self-recover with no data loss. Tuned per issue #4242 to WARN-only (never CRITICAL) with to:silent, tripping the dashboard badge only above 3000 (well clear of the rotation baseline). Real disk trouble surfaces first as I/O errors — see 10min_dbengine_global_io_errors, which still pages CRITICAL.
    to: silent