Glances

This manual describes Glances version 2.2.

Copyright © 2012-2014 Nicolas Hennion <nicolas@nicolargo.com>

December 2014

Table of Contents

Introduction

Glances is a cross-platform curses-based system monitoring tool which aims to present a maximum of information in a minimum of space, ideally to fit in a classical 80x24 terminal or higher to have additional information. It can adapt dynamically the displayed information depending on the terminal size.

Glances can also work in client/server mode. Remote monitoring could be done via terminal or web interface.

Glances is written in Python and uses the psutil library to get information from your system.

Console (80x24)

images/screenshot.png

Full view (>80x24)

images/screenshot-wide.png

Web interface (Firefox)

images/screenshot-web.png

Usage

Standalone Mode

Simply run:

$ glances

Client/Server Mode

If you want to remotely monitor a machine, called server, from another one, called client, just run on the server:

server$ glances -s

and on the client:

client$ glances -c @server

where @server is the IP address or hostname of the server.

Glances can centralize available Glances servers using the --browser option. The server list can be staticaly defined in the Glances configuration file (section [serverlist]). Glances can also detect and display all Glances servers available on you network (auto discover mode is based on the the Zeroconf protocol only available on GNU/Linux and Mac OS X):

client$ glances --browser

It is possible to disable the auto discover mode --disable-autodiscover.

In server mode, you can set the bind address -B ADDRESS and listening TCP port -p PORT.

In client mode, you can set the TCP port of the server -p PORT.

You can set a password to access to the server --password.

Default binding address is 0.0.0.0 (Glances will listen on all the available network interfaces) and TCP port is 61209.

In client/server mode, limits are set by the server side.

Glances is IPv6 compatible. Just use the -B :: option to bind to all IPv6 addresses.

As an experimental feature, if Glances server is not detected by the client, the latter will try to grab stats using the SNMP protocol:

client$ glances -c @snmpserver

Note: Stats grabbed by SNMP request are limited (operating system dependent).

Web Server Mode

If you want to remotely monitor a machine, called server, from any device with a web browser, just run the server with the -w option:

server$ glances -w

and on the client enter the following URL in your favorite web browser:

http://@server:61208

where @server is the IP address or hostname of the server.

The Glances web interface follows responsive web design principles.

Screenshot from Chrome on Android

images/screenshot-web2.png

Command Reference

Command-Line Options

-h, --help show this help message and exit
-V, --version show program's version number and exit
-d, --debug Enable debug mode
-C CONF_FILE, --config CONF_FILE
 path to the configuration file
--enable-history
 enable the history mode
--disable-bold disable bold mode in the terminal
--disable-diskio
 disable disk I/O module
--disable-fs disable filesystem module
--disable-network
 disable network module
--disable-sensors
 disable sensors module
--disable-left-sidebar
 disable left sidebar
--disable-process
 disable process module
--disable-log disable log module
--enable-process-extended
 enable extended stats on top process
--enable-history
 enable the history mode
--path-history PATH_HISTORY
 Set the export path for graph history
--output-csv OUTPUT_CSV
 export stats to a CSV file
-c CLIENT, --client CLIENT
 connect to a Glances server by IPv4/IPv6 address or hostname
-s, --server run Glances in server mode
--browser run the Glances client browser (list of Glances server)
--disable-autodiscover
 disable autodiscover feature
-p PORT, --port PORT
 define the client/server TCP port [default: 61209]
-B BIND_ADDRESS, --bind BIND_ADDRESS
 bind server to the given IPv4/IPv6 address or hostname
--password-badidea PASSWORD_ARG
 define password from the command line
--password define a client/server password from the prompt or file
--disable-autodiscover
 Hide Glances server from the auto discover feature
--snmp-community SNMP_COMMUNITY
 SNMP community
--snmp-port SNMP_PORT
 SNMP port
--snmp-version SNMP_VERSION
 SNMP version (1, 2c or 3)
--snmp-user SNMP_USER
 SNMP username (only for SNMPv3)
--snmp-auth SNMP_AUTH
 SNMP authentication key (only for SNMPv3)
--snmp-force force SNMP mode
-t TIME, --time TIME
 set refresh time in seconds [default: 3 sec]
-w, --webserver
 run Glances in web server mode
-f PROCESS_FILTER, --process-filter PROCESS_FILTER
 set the process filter patern (regular expression)
--process-short-name
 force short name for processes name
--hide-kernel-threads
 hide kernel threads in process list
--tree display processes as a tree
-b, --byte display network rate in byte per second
-1, --percpu start Glances in per CPU mode
--fs-free-space
 display FS free space instead of used
--theme-white optimize display for white background

Interactive Commands

The following commands (key pressed) are supported while in Glances:

ENTER

Set the process filter Filter is a regular expression pattern:

  • gnome: all processes starting with the gnome string
  • .*gnome.*: all processes containing the gnome string
a

Sort process list automatically

  • If CPU iowait >60%, sort processes by I/O read and write
  • If CPU >70%, sort processes by CPU usage
  • If MEM >70%, sort processes by memory usage
b
Switch between bit/s or Byte/s for network I/O
c
Sort processes by CPU usage
d
Show/hide disk I/O stats
e
Enable/disable top extended stats
f
Show/hide file system stats
F
Switch between FS used and free space
g
Generate hraphs for current history
h
Show/hide the help screen
i
Sort processes by I/O rate
l
Show/hide log messages
m
Sort processes by MEM usage
n
Show/hide network stats
p
Sort processes by name
q or ESC
Quit the current Glances session
r
Reset history
s
Show/hide sensors stats
t
Sort process by CPU times (TIME+)
T
View network I/O as combination
u
View cumulative network I/O
w
Delete finished warning log messages
x
Delete finished warning and critical log messages
z
Show/hide processes stats
1
Switch between global CPU and per-CPU stats
2
Enable/disable left sidebar
/
Switch between short name / command line (processes name)

In the Glances client browser (accessible through the --browser command line argument):

ENTER
Run Glances client to the selected server
UP
Up in the servers list
DOWN
Down in the servers list
q or ESC
Quit Glances

Configuration

No configuration file is mandatory to use Glances.

Furthermore a configuration file is needed to set up limits, disks or network interfaces to hide and/or monitored processes list or to define alias.

By default, the configuration file is under:

Linux:/etc/glances/glances.conf
*BSD and OS X:/usr/local/etc/glances/glances.conf
Windows:%APPDATA%\glances\glances.conf

On Windows XP, the %APPDATA% path is:

C:\Documents and Settings\<User>\Application Data

Since Windows Vista and newer versions:

C:\Users\<User>\AppData\Roaming
or
%userprofile%\AppData\Roaming

You can override the default configuration, located in one of the above directories on your system, except for Windows.

Just copy the glances.conf file to your $XDG_CONFIG_HOME directory, e.g., on Linux:

mkdir -p $XDG_CONFIG_HOME/glances
cp /usr/share/doc/glances/glances.conf $XDG_CONFIG_HOME/glances/

On OS X, you should copy the configuration file to ~/Library/Application Support/glances/.

Logs and debug mode

Glances logs all its internal messages to a log file. By default, only INFO & WARNING & ERROR &CRITICAL levels are logged, but DEBUG messages can ben logged using the -d option on the command line.

By default, the log file is under:

Linux, *BSD and OS X:
 /tmp/glances.log
Windows:%APPDATA%\Local\temp\glances.log

Anatomy Of The Application

Legend

GREEN stat counter is "OK"
BLUE stat counter is "CAREFUL"
MAGENTA stat counter is "WARNING"
RED stat counter is "CRITICAL"

Note: only stats with colored background will be logged in the alert view.

CPU

Short view:

images/cpu.png

If enough horizontal space is available, extended CPU information are displayed.

Extended view:

images/cpu-wide.png

To switch to per-CPU stats, just hit the 1 key:

images/per-cpu.png

The CPU stats are shown as a percentage and for the configured refresh time. The total CPU usage is displayed on the first line.

If user|system CPU is <50%, then status is set to "OK"
If user|system CPU is >50%, then status is set to "CAREFUL"
If user|system CPU is >70%, then status is set to "WARNING"
If user|system CPU is >90%, then status is set to "CRITICAL"

Note: limit values can be overwritten in the configuration file under the [cpu] and/or [percpu] sections.

Load

images/load.png

On the No Sheep blog, Zachary Tirrell defines the load average [1]:

"In short it is the average sum of the number of processes waiting in the run-queue plus the number currently executing over 1, 5, and 15 minutes time periods."

Glances gets the number of CPU core to adapt the alerts. Alerts on load average are only set on 15 minutes time period. The first line also displays the number of CPU core.

If load average is <0.7*core, then status is set to "OK"
If load average is >0.7*core, then status is set to "CAREFUL"
If load average is >1*core, then status is set to "WARNING"
If load average is >5*core, then status is set to "CRITICAL"

Note: limit values can be overwritten in the configuration file under the [load] section.

Memory

Glances uses two columns: one for the RAM and one for the SWAP.

images/mem.png

If enough space is available, Glances displays extended information for the RAM:

images/mem-wide.png

Alerts are only set for used memory and used swap.

If used memory|swap is <50%, then status is set to "OK"
If used memory|swap is >50%, then status is set to "CAREFUL"
If used memory|swap is >70%, then status is set to "WARNING"
If used memory|swap is >90%, then status is set to "CRITICAL"

Note: limit values can be overwritten in the configuration file under the [memory] and/or [memswap] sections.

Network

images/network.png

Glances displays the network interface bit rate. The unit is adapted dynamically (bits per second, kbits per second, Mbits per second, etc).

Alerts are only set if the maximum speed per network interface is available (see sample in the configuration file).

Note: it is possibile to define a list of network interfaces to hide and per-interface limit values in the [network] section of the configuration file and aliases for interface name.

Disk I/O

images/diskio.png

Glances displays the disk I/O throughput. The unit is adapted dynamically.

There is no alert on this information.

Note: it is possible to define a list of disks to hide under the [diskio] section in the configuration file and aliases for disk name.

File System

images/fs.png

Glances displays the used and total file system disk space. The unit is adapted dynamically.

Alerts are set for used disk space.

If used disk is <50%, then status is set to "OK"
If used disk is >50%, then status is set to "CAREFUL"
If used disk is >70%, then status is set to "WARNING"
If used disk is >90%, then status is set to "CRITICAL"

Note: limit values can be overwritten in the configuration file under the [filesystem] section.

Sensors

Glances can displays the sensors information using lm-sensors, hddtemp and batinfo [2].

All of the above libraries are available only on Linux.

As of lm-sensors, a filter is being applied in order to display temperature only.

images/sensors.png

There is no alert on this information.

Note: limit values and sensors alias names can be defined in the configuration file under the [sensors] section.

Processes List

Compact view:

images/processlist.png

Full view:

images/processlist-wide.png

Three views are available for processes:

  • Processes summary
  • Optional monitored processes list (see below)
  • Processes list

The processes summary line display:

  • Tasks number (total number of processes)
  • Threads number
  • Running tasks number
  • Sleeping tasks number
  • Other tasks number (not running or sleeping)
  • Sort key

By default, or if you hit the a key, the processes list is automatically sorted by:

  • CPU if there is no alert (default behavior)
  • CPU if a CPU or LOAD alert is detected
  • MEM if a memory alert is detected
  • Disk I/O if a CPU iowait alert is detected

The number of processes in the list is adapted to the screen size.

CPU%
% of CPU used by the process
MEM%
% of MEM used by the process
VIRT
Total program size - Virtual Memory Size (VMS)
RES
Resident Set Size (RSS)
PID
Process ID
USER
User ID
NI
Nice level of the process (niceness other than 0 is highlighted)
S
Process status (running process is highlighted)
TIME+
Cumulative CPU time used
IOR/s
Per process I/O read rate (in Byte/s)
IOW/s
Per process I/O write rate (in Byte/s)
COMMAND
Process command line User cans switch to the process name by pressing on the / key

Process status legend:

R
Running
S
Sleeping (may be interrupted)
D
Disk sleep (may not be interrupted)
T
Traced / Stopped
Z
Zombie

In standalone mode, additionals informations are provided for the top process:

images/processlist-top.png
  • CPU affinity (number of cores used by the process)
  • Extended memory information (swap, shared, text, lib, data and dirty on Linux)
  • Open threads, files and network sessions (TCP and UDP)
  • IO nice level

The extended stats feature could be enabled using the --enable-process-extended option (command line) or the e key (curses interface).

Note: limit values can be overwritten in the configuration file under the [process] section.

Monitored Processes List

The monitored processes list allows user, through the configuration file, to group processes and quickly show if the number of running processes is not good.

images/monitored.png

Each item is defined by:

  • description: description of the processes (max 16 chars).
  • regex: regular expression of the processes to monitor.
  • command (optional): full path to shell command/script for extended stat. Should return a single line string. Use with caution.
  • countmin (optional): minimal number of processes. A warning will be displayed if number of processes < count.
  • countmax (optional): maximum number of processes. A warning will be displayed if number of processes > count.

Up to 10 items can be defined.

For example, if you want to monitor the Nginx processes on a Web server, the following definition should do the job:

[monitor]
list_1_description=Nginx server
list_1_regex=.*nginx.*
list_1_command=nginx -v
list_1_countmin=1
list_1_countmax=4

If you also want to monitor the PHP-FPM daemon processes, you should add another item:

[monitor]
list_1_description=Nginx server
list_1_regex=.*nginx.*
list_1_command=nginx -v
list_1_countmin=1
list_1_countmax=4
list_1_description=PHP-FPM
list_1_regex=.*php-fpm.*
list_1_countmin=1
list_1_countmax=20

In client/server mode, the list is defined on the server side. A new method, called getAllMonitored, is available in the APIs and get the JSON representation of the monitored processes list.

Alerts are set as following:

If number of processes is 0, then status is set to "CRITICAL"
If number of processes is min < current < max, then status is set to "OK"
Else status is set to "WARNING"

Logs

images/logs.png

A log messages list is displayed in the bottom of the screen if (and only if):

  • at least one WARNING or CRITICAL alert was occurred
  • space is available in the bottom of the console/terminal

Each alert message displays the following information:

  1. start datetime
  2. duration if alert is terminated or ongoing if the alert is still in progress
  3. alert name
  4. {min,avg,max} values or number of running processes for monitored processes list alerts

Other Outputs

It is possible to export statistics to CSV file.

$ glances --output-csv /tmp/glances.csv

CSV files have two lines per stats:

APIs Documentations

Glances includes a XML-RPC server and a RESTFULL-JSON API which and can be used by another client software.

APIs documentations are available at:

Support

To post a question about Glances use case, please post it to the offical Q&A forum.

To report a bug or a feature request use the bug tracking system at https://github.com/nicolargo/glances/issues.

Feel free to contribute !

[1]http://nosheep.net/story/defining-unix-load-average/
[2]https://github.com/nicolargo/batinfo