首頁 » Linux » Monitoring » monit

1. monit 入門

由 datahunter 在 四, 18/04/2024 - 13:08 發表

最後更新: 2024-04-17

目錄

 

介紹

Program:

LSB executable

Monit can

  • start a process if it does not run
  • restart a process if it does not respond
  • stop a process if it uses too much resources
  • execute meaningful causal actions in error situations

monitor

  • files, directories and filesystems for changes(timestamps , checksum, permission size changes)
  • TCP/IP network checks

Version:

monit -V

    This is Monit version 5.3.2

 

 

Start Options

 

  • -d n                  Run Monit as a daemon once per n seconds. (poll cycle) [Setting: set daemon n]
  • -s statefile         Write state information to this file. (A service's monitoring state is persistent across Monit restart)
  • -l logfile

 

Command

 

  • monitor <name | all>

mon a service

  • unmonitor<name | all>

Disable monitoring of all services listed in the control file.

  • status

    看到 webpanel 有的資料 (亦即是要有 webpanel enable 才用到 !!)

    monit status

    The Monit daemon 5.3.2 uptime: 0m

System 'system_status'
  status                            Running
  monitoring status                 Monitored
  load average                      [0.51] [0.77] [0.81]
  cpu                               16.6%us 8.6%sy 1.0%wa
  memory usage                      4662296 kB [57.6%]
  swap usage                        577540 kB [4.7%]
  data collected                    Wed, 15 Jan 2014 16:04:15

Process 'ssh'
  status                            Running
  monitoring status                 Monitored
  pid                               20952
  parent pid                        1
  uptime                            6d 23h 54m
  children                          3
  memory kilobytes                  2424
  memory kilobytes total            8824
  memory percent                    0.0%
  memory percent total              0.1%
  cpu percent                       0.0%
  cpu percent total                 0.0%
  data collected                    Wed, 15 Jan 2014 16:04:15

...
  • summary

# 沒有 status 那麼詳細

monit summary

Monit 5.25.1 uptime: 26m
┌─────────────────────────────────┬────────────────────────────┬───────────────┐
│ Service Name                    │ Status                     │ Type          │
├─────────────────────────────────┼────────────────────────────┼───────────────┤
│ myserver                        │ OK                         │ System        │
├─────────────────────────────────┼────────────────────────────┼───────────────┤
│ public                          │ OK                         │ Network       │
└─────────────────────────────────┴────────────────────────────┴───────────────┘
  • reload                    # The daemon will reread its configuration, close and reopen log files.

 

  • quit

 

  • validate                  # Check all services listed in the control file (相當於人手行一次)
                                 # Default behavior when Monit runs in daemon mode
... info     : Awakened by User defined signal 1
... info     : Monit daemon with PID 536802 awakened
  • start / stop / restart

start <all | name>

stop <all | name>

restart <all | name>

i.e.

# Start all services listed in the control file and enable monitoring for them.

monit start all

 

Config File

 

指定用另一個 config file:

monit -c /var/monit/monitrc

Default config file:

# If this file does not exist, Monit will try /etc/monitrc

~/.monitrc

Service 的 config file(service monit start):

/ect/monitrc

...
include /etc/monit.d/*.cfg

Test Config File Syntax

monit -t

output:

Control file syntax OK

Global Setting

# Monit's poll cycle
# Monit detaches from the console, goes to sleep for the given poll interval, 
# wakes up and start monitoring again in an endless cycle
set daemon 20
    start delay 120

# Log Setting
set logfile /var/log/monit.log


# unique id for the Monit instance
set idfile /var/lib/monit/id


# saves monitoring states on each cycle.
set statefile /var/lib/monit/state


# Multiple servers may be specified using a comma separator
# (If the first mail server fails, Monit will use the second mail server in the list)
set mailserver smtpgw1,
               smtpgw2 port 1025,
               localhost                  # fallback relay


# By default, the queue is disabled and if the alert handler fails,
# Monit will simply drop the alert message.
set eventqueue
    basedir /var/lib/monit/events              # set the base directory where events will be stored
    slots 100                                  # optionally limit the queue size


# mail format  
set mail-format { from: [email protected] }

# 可以多人收 ALERT, 一行一個人
set alert [email protected]
set alert [email protected]

# built-in mini-httpd server
set httpd port 2812 and
    use address localhost
    allow localhost
    allow admin:ClearTextPW


include /etc/monit.d/*.conf

Config file permission:

# 一定要這個 permission

chow root: /etc/monit.conf

chmod 600 /etc/monit.conf

with start delay

要延遲多耐(秒)才開始 mon

期間 monit status 無效

Cannot create socket to [localhost]:2812 -- Connection refused

Log

log to file

set log /var/log/monit.log

log to syslog

set log syslog

 

Service Poll Time (Poll cycle)

 

一共有兩種方式:

  • EVERY [number] CYCLES
  • EVERY [cron]

Example:

# cycles
# 每 "set daemon N" 的 2N 行一次
check process nginx with pidfile /var/run/nginx.pid every 2 cycles

# cron
# cron jobs format. 星期一至五, 8 a.m. 點至 7 p.m 內的每分鐘
check program nginx with pidfile /var/run/nginx.pid every "* 8-19 * * 1-5"

P.S.

Strongly recommend to use an asterix in the minute field or at minimum a range,

* Never use a specific minute as Monit may not run on that minute.

因為 monit 的 scheduler 是 serial polling, 所以不能保證執行時間

e.g.

# a range

0-15

 

My Settings

 

Centos 6

/etc/logrotate.d/monit

/var/log/monit.log {
    missingok
    notifempty
    size 100k
    create 0644 root root
    postrotate
        # /bin/systemctl reload monit.service > /dev/null 2>&1 || :
        /sbin/service monit condrestart > /dev/null 2>&1 || :
    endscript
}

行以下 cmd 先 ~

mkdir /var/lib/monit

rm /etc/monit.d/logging

/etc/monit.conf

# My Basic Config

set daemon 10
    start delay 60

set logfile /var/log/monit.log

set idfile /var/lib/monit/id
set statefile /var/lib/monit/state

set mailserver localhost
set alert xxx@xxx

set httpd port 2812 and
    use address localhost
    allow localhost

include /etc/monit.d/*.conf

RHEL 8

set daemon 10
    start delay 60

set idfile /var/lib/monit/id
set statefile /var/lib/monit/state

set mailserver localhost
set alert [email protected]

set httpd unixsocket /var/run/monit.sock
    permission 600
    allow localhost

check system $HOST
    if loadavg (1min) per core > 2 for 5 cycles then alert
    if loadavg (5min) per core > 1.5 for 10 cycles then alert
    if cpu usage > 95% for 10 cycles then alert
    if memory usage > 80% then alert
    if swap usage > 80% then alert

include /etc/monit.d/*

Notes

 * 不用加 "set log syslog" 因為有 /etc/monit.d/logging

 * 建立目錄 /var/lib/monit 去放 idfile 及 statefile, 否則會放在 /root/{.monit.id,.monit.state}

 * "set httpd unixsocket" 必須加 "allow localhost" 才用到

 * include 不用 *.conf, 因為 upgrade 後又會有 /etc/monit.d/logging

 

Web-Interface

 

Default: Disable.

Default TCP port 127.0.0.1:2812

 * If security is a concern, bind the HTTP interface to localhost / Unix Socket

Monit HTTPD Authentication:

set httpd port 2812
    allow localhost
    allow 10.1.1.1
    allow 192.168.1.0/255.255.255.0
    allow 10.0.0.0/8
    allow myuser:mypassword
    allow md5 /etc/httpd/htpasswd john paul ringo george

* allow 們是 "AND" 的關係

Web UI:

UNIX SOCKET

SET HTTPD UNIXSOCKET <path>
    [UID <uid | username>]
    [GID <gid | groupname>]
    [PERMISSION <octal number>]
    ...

UID, GID

optional, defaults to the user who executes Monit

PERMISSION

optional, absolute octal mode

i.e.

set httpd unixsocket /var/run/monit.sock
    permission 600
    allow localhost

signature

hide Monit version

set httpd
  port 2812
  signature disable

 

Monit Action

 

Available actions

IF <TEST> THEN ACTION

ACTION:

  • alert
  • restart          # restarts the service and sends an alert
  • start
  • stop
  • exec             # EXEC can be used to execute an arbitrary program and send an alert.
  • unmonitor

EXEC (重點: repeat, as)

repeat

The program is executed only once if the test fails.

You can enable execute repetition if the error persists for a given number of cycles:

# 當每個 cycles 係 30 秒時(set daemon 30), 以下設定即每 5 分析.

if failed <test> then exec "/usr/local/bin/sms.sh"
     as uid "nobody" and gid "nobody"
     repeat every 10 cycles

as

# You may optionally specify the uid and gid

exec "/root/scripts/fixit.sh" as uid nobody and gid nobody

 * If Monit is run by root, then all programs executed by Monit will be started with superuser privileges

注意, 如果是 shell script, 那 Script 一定要  "#!/bin/bash"

Notes

ALERT 用 "WITH REMINDER ON N CYCLES", EXEC 用 "REPEAT EVERY N CYCLES"

i.e.

CHECK HOST MyVPN ADDRESS 192.168.88.20
  ALERT [email protected] WITH REMINDER ON 2 CYCLES
  IF FAILED PING
    COUNT 5
    THEN EXEC /home/fortivpn/vpn/start-vpn.sh
    AS uid fortivpn AND gid fortivpn
    REPEAT EVERY 2 CYCLES

 

Service Monitoring Mode

 

MODE < ACTIVE | PASSIVE >

  • ACTIVE: raise alerts and restart the service    # DEFAULT
  • PASSIVE: raise alerts only

i.e.

# Monit will not try to (re)start this service if it is not running:

check process sybase with pidfile /var/run/sybase.pid
  mode passive
  start = "/etc/init.d/sybase start"
  stop  = "/etc/init.d/sybase stop"

 

可用的監測(CHECK X)

 

  • CHECK SYSTEM <UniqueName>                                 # 系統資源
  • CHECK FILESYSTEM <UniqueName> PATH <path>      # Disk IO 及 Space Usage
  • CHECK PROCESS <UniqueName> <PIDFILE path | MATCHING regex>
  • CHECK FILE <UniqueName> PATH <path>
  • CHECK FIFO <UniqueName> PATH <path>
  • CHECK DIRECTORY <UniqueName> PATH <path>
  • CHECK HOST <UniqueName> ADDRESS <Address>
  • CHECK PROGRAM <UniqueName> PATH <executable file> [TIMEOUT <number> SECONDS]

 

HOST

CHECK HOST <UniqueName> ADDRESS <Address>

UniqueName 是一個 Label 來. 行 monit status 時會顯示它

e.g.

Settings:

CHECK HOST localhost ADDRESS 127.0.0.1

Output:

Remote Host 'localhost'
  status                       OK
  monitoring status            Monitored
  ...

Address 是用於之後的 "IF" checking. 它可以是 Domain / IP

e.g.

CHECK HOST MyVPN ADDRESS 192.168.88.20
  IF FAILED PING COUNT 5
    THEN EXEC /home/fortivpn/vpn/start-vpn.sh

 

PROGRAM

If the program does not finish executing within <number> seconds,
Monit will terminate it. The default program timeout is 300 seconds

The "status test" allows one to check the program's exit status.

IF STATUS operator value THEN action

 

System resource (CPU, Memory)

 

To monitor general system resources such as CPU usage, total memory usage or load average.

If you use the variable $HOST as the name, it will expand to the hostname.

 

CPU

$HOST 會是 hostname

check system $HOST
  if loadavg (1min) > 4 then alert
  if loadavg (5min) > 2 then alert
  if memory usage > 75% for 6 cycles then alert
  if swap usage > 25% then alert
  if cpu usage > 95% for 10 cycles then alert
  if cpu usage (user) > 70% then alert
  if cpu usage (system) > 30% then alert
  if cpu usage (wait) > 20% then alert

Memory usage

What does monit consider to be memory usage?

# On latest Monit (ie: 5.25.x) the memory usage value accounts for ZFS ARC cache

Code

si->memory.usage.bytes = systeminfo.memory.size - zfsarcsize - (uint64_t)(mem_free + buffers + cached + slabreclaimable) * 1024;

# 對比

monit status | grep 'memory usage'

memory usage                 2.5 GB [32.4%]

grep -w -e MemTotal -e Buffers -e Cached -e MemFree -e Slab  /proc/meminfo

MemTotal:        8060728 kB
MemFree:          375184 kB
Buffers:          652612 kB
Cached:          3048636 kB
Slab:            1472140 kB

Usage = MemTotal - MemFree - Buffers - Cached  - Slab = 2.45 G

Network

 * Unit: Byte

# 10 min up/down > 5 MByte

# 1 hr up size > 4 GByte

check network public with interface eth0
    if failed link then alert
    if changed link then alert
    if download > 5 MB/s for 20 cycles then alert
    if upload > 5 MB/s for 20 cycles then alert
    if total uploaded > 4 GB in last hour then alert

 

Disk I/O (filesystem)

 

Monit will normally need to run as the root user to access this metrics.

# Unit:  "B","KB","MB","GB"

i.e.

check filesystem datafs with path /dev/sda1
    # Usage
    if space usage > 90% then alert
    if inode usage > 90% then alert
    # IO
    if read rate > 10 MB/s for 5 cycles then alert
    if read rate > 500 operations/s for 5 cycles then alert
    if write rate > 10 MB/s for 5 cycles then alert
    if write rate > 500 operations/s for 5 cycles then alert
    if service time > 10 milliseconds for 3 times within 5 cycles then alert

* Per-process I/O activity statistics by platform: Byte

Service time per operation

Service Time is the time taken to complete a read or a write operation.

If it grows, it means that the disk is not able to handle the operations fast enough.

# Unit is "ms" (millisecond) or "s" (second)

if service time > 10 milliseconds
    for 3 times within 5 cycles
then alert

 

Monitoring a directory

 

check directory bin with path /bin
    if failed permission 755 then unmonitor
    if failed uid 0 then unmonitor
    if failed gid 0 then unmonitor

 

Monitoring file

 

TEST:

  • IF FAILED [MD5|SHA1] CHECKSUM [EXPECT checksum] THEN action
  • IF CHANGED [MD5|SHA1] CHECKSUM THEN action
  • IF TIMESTAMP [[operator] value [unit]] THEN action
  • IF CHANGED TIMESTAMP THEN action
  • IF [DOES] NOT EXIST THEN action
  • IF SIZE [[operator] value [unit]] THEN action
  • IF CHANGED SIZE THEN action

 

Changed checksum

if failed
   checksum expect 8f7f419955cefa0b33a2ba316cba3659
then alert

OR

if changed checksum then exec "/usr/bin/apachectl graceful"

Changed timestamp (unit: "SECOND", "MINUTE", "HOUR" or "DAY")

# If the file is older then N minutes, then things are broken

IF TIMESTAMP > 1 MINUTE THEN alert

EXIST

IF [DOES] NOT EXIST THEN action

 

PERMISSION TESTING

  • IF FAILED PERM(ISSION) octalnumber THEN action
  • IF FAILED [E]UID user THEN action
  • IF FAILED GID group THEN action

Example:

 check file shadow with path /etc/shadow
       if failed permission 0640 then alert

 check file shadow with path /etc/shadow
       if failed uid root then alert

 check file shadow with path /etc/shadow
       if failed gid shadow then alert

 

重複次數(WITHIN, FOR)

 

WITHIN

[[<X>] [TIMES WITHIN] <Y> CYCLES]

IF CHANGED <TEST> [[<X>] [TIMES WITHIN] <Y> CYCLES] THEN ACTION

e.g.

# An alert is delivered each time the condition becomes true.

# tenth cycle if a service remains in a state

alert foo@bar with reminder on 10 cycles

# tcp port 80 在 5 cycles 內有 3 次無反應時出 alert

if failed port 80 for 3 times within 5 cycles then alert

# 連續一段時間:

if cpu is greater than 50% for 5 cycles then restart

# 限制 restart 的次數 (IF N RESTART WITHIN M CYCLES THEN <action>)

# restarted 2 times within 3 cycles

if 2 restarts within 3 cycles then unmonitor

FOR

Syntax: ... FOR N CYCLES ...

Requires X consecutive events before switching the state

# cycle is failure (1-0-1-0-1-0-...), then "for 2 cycles" condition will never match

IF FAILED
   PORT 80
   FOR 2 CYCLES
THEN alert

 

Monit 的 Limits

 

monit 自身的 resource limit

Default values:

set limits {
    programOutput:     512 B,      # check program's output truncate limit
    sendExpectBuffer:  256 B,      # limit for send/expect protocol test
    fileContentBuffer: 512 B,      # limit for file content test
    httpContentBuffer: 1 MB,       # limit for HTTP content test
    networkTimeout:    5 seconds   # timeout for network I/O
    programTimeout:    300 seconds # timeout for check program
    stopTimeout:       30 seconds  # timeout for service stop
    startTimeout:      30 seconds  # timeout for service start
    restartTimeout:    30 seconds  # timeout for service restart
}

 

Cheat Sheet

 

monit -t                   # syntax checking

monit reload             # config file 更新後必須 reload, 否則 "monitor <name>" 反映不到新 settings

monit summary         # 純看 Status(OK|Initializing)

monit status             # 有 process 的 info (uptime, threads, children, cpu, memory, io)

 

Doc

 

# 很完整的 doc

 

 

»
  • 瀏覽次數： 248

Creative Commons license icon Creative Commons license icon