Kapisi/roles/Sharingan/files/monit/checks/system

34 lines
1.9 KiB
Plaintext
Raw Normal View History

###############################################################################
## Services
###############################################################################
#
## Check general system resources such as load average, cpu and memory
## usage. Each test specifies a resource, conditions and the action to be
## performed should a test fail.
check system $HOST
if loadavg (1min) per core > 2 for 25 times within 30 cycles then exec "/etc/monit.d/scripts/critical Load average is high"
if memory usage > 90% for 8 times within 10 cycles then exec "/etc/monit.d/scripts/critical Memory is overused."
if swap usage > 25% for 8 times within 10 cycles then exec "/etc/monit.d/scripts/critical Swap is overused."
group system
#
## Check filesystem permissions, uid, gid, space usage, inode usage and disk I/O.
## Other services, such as databases, may depend on this resource and an automatically
## graceful stop may be cascaded to them before the filesystem will become full and data
## lost.
check filesystem rootfs with path /
if space usage > 80% for 5 times within 15 cycles then exec "/etc/monit.d/scripts/critical Root filesystem is too full."
if read rate > 500 operations/s for 5 cycles then exec "/etc/monit.d/scripts/critical Too many reads on root filesystem."
if write rate > 500 operations/s for 5 cycles then exec "/etc/monit.d/scripts/critical Too many writes on root filesystem."
2022-05-04 10:30:23 -05:00
# if service time > 50 milliseconds for 3 times within 5 cycles then exec "/etc/monit.d/scripts/critical Root filesystem performing slowly -- check iostat."
group system
#
## Check a network link status (up/down), link capacity changes, saturation
## and bandwidth usage.
#
# check network public with interface eth0
# if failed link then alert
# if changed link then alert
# if saturation > 90% then alert
# if download > 10 MB/s then alert
# if total uploaded > 1 GB in last hour then alert