Updated BOINC alarms to be a bit more sensible.
Austin S. Hemmelgarn committed
Jun 5, 2018 at 15:36 UTC
60cfd06c4e913366132c3580c79b77f7d4490c67
1 file changed
+6
-6
conf.d/health.d/boinc.conf
+6
-6
@@ -6,11 +6,11 @@ template: boinc_compute_errors
6
os: *
7
hosts: *
8
families: *
9
- lookup: sum -10m unaligned of comperror
9
+ lookup: average -10m unaligned of comperror
10
units: tasks
11
every: 1m
12
- warn: $this > 1
13
- crit: $this > 4
12
+ warn: $this > 0
13
+ crit: $this > 1
14
delay: up 1m down 5m multiplier 1.5 max 1h
15
info: the total number of compute errors over the past 10 minutes
16
to: sysadmin
@@ -21,11 +21,11 @@ template: boinc_upload_errors
21
os: *
22
hosts: *
23
families: *
24
- lookup: max -10m unaligned of upload_failed
24
+ lookup: average -10m unaligned of upload_failed
25
units: tasks
26
every: 1m
27
- warn: $this > 2
28
- crit: $this > 4
27
+ warn: $this > 0
28
+ crit: $this > 1
29
delay: up 1m down 5m multiplier 1.5 max 1h
30
info: the average number of failed uploads over the past 10 minutes
31
to: sysadmin