I’ve compiled 25 performance monitoring and debugging tools that will be helpful when you are working on Linux environment. This list is not comprehensive or authoritative by any means.
However this list has enough tools for you to play around and pick the one that is suitable your specific debugging and monitoring scenario.
Using sar utility you can do two things: 1) Monitor system real time performance (CPU, Memory, I/O, etc) 2) Collect performance data in the background on an on-going basis and do analysis on the historical data to identify bottlenecks.
Sar is part of the sysstat package. The following are some of the things you can do using sar utility.
Collective CPU usage
Individual CPU statistics
Memory used and available
Swap space used and available
Overall I/O activities of the system
Individual device I/O activities
Context switch statistics
Run queue and load average data
Network statistics
Report sar data from a specific time
and lot more..
The following sar command will display the system CPU statistics 3 times (with 1 second interval).
The following “sar -b” command reports I/O statistics. &#” indicates that the sar -b will be executed for every 1 second for a total of 3 times.
$ sar -b 1 3
Linux 2.6.18-194.el5PAE (dev-db)
03/26/2011
01:56:28 PM
01:56:29 PM
01:56:30 PM
01:56:31 PM
More SAR examples:
2. Tcpdump
tcpdump is a network packet analyzer. Using tcpdump you can capture the packets and analyze it for any performance bottlenecks.
The following tcpdump command example displays captured packets in ASCII.
$ tcpdump -A -i eth0
tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on eth0, link-type EN10MB (Ethernet), capture size 96 bytes
14:34:50.913995 IP valh4.lell.net.ssh & yy.domain.innetbcp.net.11006: P :(116) ack
E.....@.@..]..i...9...*.V...]...P....h....E...&{..U=...g.
......G..7\+KA....A...L.
14:34:51.423640 IP valh4.lell.net.ssh & yy.domain.innetbcp.net.11006: P 116:232(116) ack 1 win 63652
E.....@.@..\..i...9...*.V..*]...P....h....7......X..!....Im.S.g.u:*..O&....^#Ba...
E..(R.@.|.....9...i.*...]...V..*P..OWp........
Using tcpdump you can capture packets based on several custom conditions. For example, capture packets that flow through a particular port, capture tcp communication between two specific hosts, capture packets that belongs to a specific protocol type, etc.
More tcpdump examples:
Nagios is an open source monitoring solution that can monitor pretty much anything in your IT infrastructure. For example, when a server goes down it can send a notification to your sysadmin team, when a database goes down it can page your DBA team, when the a web server goes down it can notify the appropriate team.
You can also set warning and critical threshold level for various services to help you proactively address the issue. For example, it can notify sysadmin team when a disk partition becomes 80% full, which will give enough time for the sysadmin team to work on adding more space before the issue becomes critical.
Nagios also has a very good user interface from where you can monitor the health of your overall IT infrastructure.
The following are some of the things you can monitor using Nagios:
Any hardware (servers, switches, routers, etc)
Linux servers and Windows servers
Databases (Oracle, MySQL, PostgreSQL, etc)
Various services running on your OS (sendmail, nis, nfs, ldap, etc)
Web servers
Your custom application
More Nagios examples: , , and .
iostat reports CPU, disk I/O, and NFS statistics. The following are some of iostat command examples.
Iostat without any argument displays information about the CPU usage, and I/O statistics about all the partitions on the system as shown below.
Linux 2.6.32-100.28.5.el6.x86_64 (dev-db)
07/09/2011
%nice %system %iowait
Blk_read/s
Blk_wrtn/s
By default iostat displays I/O data for all the disks available in the system. To view statistics for a specific device (For example, /dev/sda), use the option -p as shown below.
$ iostat -p sda
Linux 2.6.32-100.28.5.el6.x86_64 (dev-db)
07/09/2011
%nice %system %iowait
Blk_read/s
Blk_wrtn/s
mpstat reports processors statistics. The following are some of mpstat command examples.
Option -A, displays all the information that can be displayed by the mpstat command as shown below. This is really equivalent to “mpstat -I ALL -u -P ALL” command.
$ mpstat -A
Linux 2.6.32-100.28.5.el6.x86_64 (dev-db)
07/09/2011
10:26:34 PM
%sys %iowait
10:26:34 PM
10:26:34 PM
10:26:34 PM
10:26:34 PM
10:26:34 PM
10:26:34 PM
10:26:34 PM
10:26:34 PM
10:26:34 PM
10:26:34 PM
10:26:34 PM
10:26:34 PM
10:26:34 PM
10:26:34 PM
10:26:34 PM
10:26:34 PM
mpstat Option -P ALL, displays all the individual CPUs (or Cores) along with its statistics as shown below.
$ mpstat -P ALL
Linux 2.6.32-100.28.5.el6.x86_64 (dev-db)
07/09/2011
10:28:04 PM
%sys %iowait
10:28:04 PM
10:28:04 PM
10:28:04 PM
10:28:04 PM
10:28:04 PM
vmstat reports virtual memory statistics. The following are some of vmstat command examples.
vmstat by default will display the memory usage (including swap) as shown below.
procs -----------memory---------- ---swap-- -----io---- --system-- -----cpu------
cs us sy id wa st
To execute vmstat every 2 seconds for 10 times, do the following. After executing 10 times, it will stop automatically.
$ vmstat 2 10
procs -----------memory---------- ---swap-- -----io---- --system-- -----cpu-----
cs us sy id wa st
0 736 6789320
0 736 6789320
iostat and vmstat are part of the sar utility. You should install sysstat package to get iostat and vmstat working.
More examples:
7. PS Command
Process is a running instance of a program. Linux is a multitasking operating system, which means that more than one process can be active at once. Use ps command to find out what processes are running on your system.
ps command also give you lot of additional information about the running process which will help you identify any performance bottlenecks on your system.
The following are few ps command examples.
Use -u option to display the process that belongs to a specific username. When you have multiple username, separate them using a comma. The example below displays all the process that are owned by user wwwrun, or postfix.
$ ps -f -u wwwrun,postfix
C STIME TTY
00:00:00 qmgr -l -t fifo -u
00:00:00 /usr/sbin/httpd2-prefork -f /etc/apache2/httpd.conf
00:00:00 /usr/sbin/httpd2-prefork -f /etc/apache2/httpd.conf
00:00:00 /usr/sbin/httpd2-prefork -f /etc/apache2/httpd.conf
00:00:00 /usr/sbin/httpd2-prefork -f /etc/apache2/httpd.conf
00:00:00 /usr/sbin/httpd2-prefork -f /etc/apache2/httpd.conf
00:00:00 /usr/sbin/httpd2-prefork -f /etc/apache2/httpd.conf
00:00:00 /usr/sbin/httpd2-prefork -f /etc/apache2/httpd.conf
00:00:00 pickup -l -t fifo -u
The example below display the process Id and commands in a hierarchy. –forest is an argument to ps command which displays ASCII art of process tree. From this tree, we can identify which is the parent process and the child processes it forked in a recursive manner.
$ ps -e -o pid,args --forest
\_ sshd: root@pts/7
\_ sshd: root@pts/11
\_ vi ./117/journal
\_ sshd: root@pts/1
\_ sshd: root@pts/5
More ps examples:
Free command displays information about the physical (RAM) and swap memory of your system.
In the example below, the total physical memory on this system is 1GB. The values displayed below are in KB.
Mem: 1034624
-/+ buffers/cache:
The following example will display the total memory on your system including RAM and Swap.
In the following command:
option m displays the values in MB
option t displays the “Total” line, which is sum of physical and swap memory values
option o is to hide the buffers/cache line from the above example.
# free -mto
Top command displays all the running process in the system ordered by certain columns. This displays the information real-time.
You can kill a process without exiting from top. Once you’ve located a process that needs to be killed, press “k” which will ask for the process id, and signal to send.
If you have the privilege to kill that particular PID, it will get killed successfully.
PID to kill: 1309
Kill PID 1309 with signal [15]:
SHR S %CPU %MEM
45:31.32 gagent
22:38.97 gagent
0:00.39 nautilus
Use top -u to display a specific user processes only in the top command output.
$ top -u geek
While unix top command is running, press u which will ask for username as shown below.
Which user (blank for all): geek
SHR S %CPU %MEM
45:31.32 gagent
22:38.97 gagent
More top examples:
pmap command displays the memory map of a given process. You need to pass the pid as an argument to the pmap command.
The following example displays the memory map of the current bash shell. In this example, 5732 is the PID of the bash shell.
$ pmap 5732
104K r-x--
/lib/ld-2.5.so
1272K r-x--
/lib/libc-2.5.so
/lib/libdl-2.5.so
/lib/libtermcap.so.2.0.8
/lib/libnsl-2.5.so
/lib/libnss_nis-2.5.so
/lib/libnss_files-2.5.so
2048K r----
/usr/lib/locale/locale-archive
pmap -x gives some additional information about the memory maps.
pmap -x 5732
Locked Mode
libc-2.5.so
libdl-2.5.so
libtermcap.so.2.0.8
libnsl-2.5.so
libnss_nis-2.5.so
libnss_files-2.5.so
locale-archive
-------- ------- ------- ------- -------
To display the device information of the process maps use ‘pamp -d pid’.
11. Netstat
Netstat command displays various network related information such as network connections, routing tables, interface statistics, masquerade connections, multicast memberships etc.,
The following are some netstat command examples.
List all ports (both listening and non listening) using netstat -a as shown below.
# netstat -a | more
Active Internet connections (servers and established)
Proto Recv-Q Send-Q Local Address
Foreign Address
0 localhost:30037
0 *:bootpc
Active UNIX domain sockets (servers and established)
Proto RefCnt Flags
/tmp/.X11-unix/X0
/var/run/acpid.socket
Use the following netstat command to find out on which port a program is running.
# netstat -ap | grep ssh
(Not all processes could be identified, non-owned process info
will not be shown, you would have to be root to see it all.)
0 dev-db:ssh
101.174.100.22:39213
CLOSE_WAIT
0 dev-db:ssh
101.174.100.22:57643
CLOSE_WAIT
Use the following netstat command to find out which process is using a particular port.
# netstat -an | grep ':80'
More netstat examples:
12. IPTraf
IPTraf is a IP Network Monitoring Software. The following are some of the main features of IPTraf:
It is a console based (text-based) utility.
This displays IP traffic crossing over your network. This displays TCP flag, packet and byte counts, ICMP, OSPF packet types, etc.
Displays extended interface statistics (including IP, TCP, UDP, ICMP, packet size and count, checksum errors, etc.)
LAN module discovers hosts automatically and displays their activities
Protocol display filters to view selective protocol traffic
Advanced Logging features
Apart from ethernet interface it also supports FDDI, ISDN, SLIP, PPP, and loopback
You can also run the utility in full screen mode. This also has a text-based menu.
More info: . .
13. Strace
Strace is used for debugging and troubleshooting the execution of an executable on Linux environment. It displays the system calls used by the process, and the signals received by the process.
Strace monitors the system calls and signals of a specific program. It is helpful when you do not have the source code and would like to debug the execution of a program. strace provides you the execution sequence of a binary from start to end.
Trace a Specific System Calls in an Executable Using Option -e
Be default, strace displays all system calls for the given executable. The following example shows the output of strace for the Linux ls command.
$ strace ls
execve("/bin/ls", ["ls"], [/* 21 vars */]) = 0
= 0x8c31000
access("/etc/ld.so.nohwcap", F_OK)
= -1 ENOENT (No such file or directory)
mmap2(NULL, 8192, PROT_READ, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xb78c7000
access("/etc/ld.so.preload", R_OK)
= -1 ENOENT (No such file or directory)
open("/etc/ld.so.cache", O_RDONLY)
fstat64(3, {st_mode=S_IFREG|0644, st_size=65354, ...}) = 0
To display only a specific system call, use the strace -e option as shown below.
$ strace -e open ls
open("/etc/ld.so.cache", O_RDONLY)
open("/lib/libselinux.so.1", O_RDONLY)
open("/lib/librt.so.1", O_RDONLY)
open("/lib/libacl.so.1", O_RDONLY)
open("/lib/libc.so.6", O_RDONLY)
open("/lib/libdl.so.2", O_RDONLY)
open("/lib/libpthread.so.0", O_RDONLY)
open("/lib/libattr.so.1", O_RDONLY)
open("/proc/filesystems", O_RDONLY|O_LARGEFILE) = 3
open("/usr/lib/locale/locale-archive", O_RDONLY|O_LARGEFILE) = 3
open(".", O_RDONLY|O_NONBLOCK|O_LARGEFILE|O_DIRECTORY|O_CLOEXEC) = 3
More strace examples:
Lsof stands for ls open files, which will list all the open files in the system. The open files include network connection, devices and directories. The output of the lsof command will have the following columns:
COMMAND process name.
PID process ID
USER Username
FD file descriptor
TYPE node type of the file
DEVICE device number
SIZE file size
NODE node number
NAME full path of the file name.
To view all open files of the system, execute the lsof command without any parameter as shown below.
# lsof | more
983101 /sbin/init
166798 /lib/ld-2.3.4.so
166799 /lib/tls/libc-2.3.4.so
163964 /lib/libsepol.so.1
166811 /lib/libselinux.so.1
972 /dev/initctl
To view open files by a specific user, use lsof -u option to display all the files opened by a specific user.
# lsof -u ramesh
7190 ramesh
475196 /bin/vi
7163 ramesh
TCP dev-db:ssh-&abc-12-12-12-12.
To list users of a particular file, use lsof as shown below. In this example, it displays all users who are currently using vi.
# lsof /bin/vi
TYPE DEVICE
8,1 196 /bin/vi
ramesh txt
8,1 196 /bin/vi
Ntop is just like top, but for network traffic. ntop is a network traffic monitor that displays the network usage.
You can also access ntop from browser to get the traffic information and network status.
The following are some the key features of ntop:
Display network traffic broken down by protocols
Sort the network traffic output based on several criteria
Display network traffic statistics
Ability to store the network traffic statistics using RRD
Identify the identify of the users, and host os
Ability to analyze and display IT traffic
Ability to work as NetFlow/sFlow collector for routers and switches
Displays network traffic statistics similar to RMON
Works on Linux, MacOS and Windows
More info:
16. GkrellM
GKrellM stands for GNU Krell Monitors, or GTK Krell Meters. It is GTK+ toolkit based monitoring program, that monitors various sytem resources. The UI is stakable. i.e you can add as many monitoring objects you want one on top of another. Just like any other desktop UI based monitoring tools, it can monitor CPU, memory, file system, network usage, etc. But using plugins you can monitoring external applications.
More info:
17. w and uptime
While monitoring system performance, w command will hlep to know who is logged on to the system.
09:35:06 up 21 days, 23:28,
load average: 0.00, 0.00, 0.00
21days 1:05
1:05 /usr/bin/Xorg :0 -nr -verbose
192.168.1.10
15.55s 0.26s sshd: localuser [priv]
192.168.1.11
19.05s 0.20s sshd: localuser [priv]
192.168.1.12
21.15s 0.16s sshd: localuser [priv]
For each and every user who is logged on, it displays the following info:
Remote host ip-address
Login time of the user
How long the user has been idle
JCPU and PCUP
The command of the current process the user is executing
Line 1 of the w command output is similar to the uptime command output. It displays the following:
Current time
How long the system has been up and running
Total number of users who are currently logged on the system
Load average for the last 1, 5 and 15 minutes
If you want only the uptime information, use the uptime command.
09:35:02 up 106 days, 28 min,
load average: 0.08, 0.11, 0.05
Please note that both w and uptime command gets the information from the /var/run/utmp data file.
/proc is a virtual file system. For example, if you do ls -l /proc/stat, you’ll notice that it has a size of 0 bytes, but if you do “cat /proc/stat”, you’ll see some content inside the file.
Do a ls -l /proc, and you’ll see lot of directories with just numbers. These numbers represents the process ids, the files inside this numbered directory corresponds to the process with that particular PID.
The following are the important files located under each numbered directory (for each process):
cmdline – command line of the command.
environ – environment variables.
fd – Contains the file descriptors which is linked to the appropriate files.
limits – Contains the information about the specific limits to the process.
mounts – mount related information
The following are the important links under each numbered directory (for each process):
cwd – Link to current working directory of the process.
exe – Link to executable of the process.
root – Link to the root directory of the process.
More /proc examples:
19. KDE System Guard
This is also called as KSysGuard. On Linux desktops that run KDE, you can use this tool to monitor system resources. Apart from monitoring the local system, this can also monitor remote systems.
If you are running KDE desktop, go to Applications -& System -& System Monitor, which will launch the KSysGuard. You can also type ksysguard from the command line to launch it.
This tool displays the following two tabs:
Process Table – Displays all active processes. You can sort, kill, or change priority of the processes from here
System Load – Displays graphs for CPU, Memory, and Network usages. These graphs can be customized by right cliking on any of these graphs.
To connect to a remote host and monitor it, click on File menu -& Monitor Remote Machine -& specify the ip-address of the host, the connection method (for example, ssh). This will ask you for the username/password on the remote machine. Once connected, this will display the system usage of the remote machine in the Process Table and System Load tabs.
20. GNOME System Monitor
On Linux desktops that run GNOME, you can use the this tool to monitor processes, system resources, and file systems from a graphical interface. Apart from monitoring, you can also use this UI tool to kill a process, change the priority of a process.
If you are running GNOME desktop, go to System -& Administration -& System Monitor, which will launch the GNOME System Monitor. You can also type gnome-system-monitor from the command line to launch it.
This tool has the following four tabs:
System – Displays the system information including Linux distribution version, system resources, and hardware information.
Processes – Displays all active processes that can be sorted based on various fields
Resources – Displays CPU, memory and network usages
File Systems – Displays information about currently mounted file systems
More info:
Conky is a system monitor or X. Conky displays information in the UI using what it calls objects. By default there are more than 250 objects that are bundled with conky, which displays various monitoring information (CPU, memory, network, disk, etc.). It supports IMAP, POP3, several audio players.
You can monitor and display any external application by craeting your own objects using scripting. The monitoring information can be displays in various format: Text, graphs, progress bars, etc. This utility is extremly configurable.
More info:
Cacti is a PHP based UI frontend for the RRDTool. Cacti stores the data required to generate the graph in a MySQL database.
The following are some high-level features of Cacti:
Ability to perform the data gathering and store it in MySQL database (or round robin archives)
Several advanced graphing featurs are available (grouping of GPRINT graph items, auto-padding for graphs, manipulate graph data using CDEF math function, all RRDTool graph items are supported)
The data source can gather local or remote data for the graph
Ability to fully customize Round robin archive (RRA) settings
User can define custom scripts to gather data
SNMP support (php-snmp, ucd-snmp, or net-snmp) for data gathering
Built-in poller helps to execute custom scripts, get SNMP data, update RRD files, etc.
Highly flexible graph template features
User friendly and customizable graph display options
Create different users with various permission sets to access the cacti frontend
Granular permission levels can be set for the individual user
and lot more..
More info:
23. Vnstat
vnstat is a command line utility that displays and logs network traffic of the interfaces on your systems. This depends on the network statistics provided by the kernel. So, vnstat doesn’t add any additional load to your system for monitoring and logging the network traffic.
vnstat without any argument will give you a quick summary with the following info:
The last time when the vnStat datbase located under /var/lib/vnstat/ was updated
From when it started collecting the statistics for a specific interface
The network statistic data (bytes transmitted, bytes received) for the last two months, and last two days.
Database updated: Sat Oct 15 11:54:00 2011
eth0 since 10/01/11
------------------------+-------------+-------------+---------------
12.90 MiB |
6.90 MiB |
19.81 MiB |
0.14 kbit/s
12.89 MiB |
6.94 MiB |
19.82 MiB |
0.15 kbit/s
------------------------+-------------+-------------+---------------
------------------------+-------------+-------------+---------------
4.30 MiB |
2.42 MiB |
6.72 MiB |
0.64 kbit/s
2.03 MiB |
1.07 MiB |
3.10 MiB |
0.59 kbit/s
------------------------+-------------+-------------+---------------
Use “vnstat -t” or “vnstat –top10” to display all time top 10 traffic days.
$ vnstat --top10
-----------------------------+-------------+-------------+---------------
4.30 MiB |
2.42 MiB |
6.72 MiB |
0.64 kbit/s
4.07 MiB |
2.17 MiB |
6.24 MiB |
0.59 kbit/s
2.48 MiB |
1.28 MiB |
3.76 MiB |
0.36 kbit/s
-----------------------------+-------------+-------------+---------------
More vnstat Examples:
htop is a ncurses-based process viewer. This is similar to top, but is more flexible and user friendly. You can interact with the htop using mouse. You can scroll vertically to view the full process list, and scroll horizontally to view the full command line of the process.
htop output consists of three sections 1) header 2) body and 3) footer.
Header displays the following three bars, and few vital system information. You can change any of these from the htop setup menu.
CPU Usage: Displays the %used in text at the end of the bar. The bar itself will show different colors. Low-priority in blue, normal in green, kernel in red.
Memory Usage
Swap Usage
Body displays the list of processes sorted by %CPU usage. Use arrow keys, page up, page down key to scoll the processes.
Footer displays htop menu commands.
More info:
25. Socket Statistics – SS
ss stands for socket statistics. This displays information that are similar to netstat command.
To display all listening sockets, do ss -l as shown below.
Recv-Q Send-Q
Local Address:Port
Peer Address:Port
:::webcache
The following displays only the established connection.
$ ss -o state established
Recv-Q Send-Q
Local Address:Port
Peer Address:Port
192.168.1.10:ssh
192.168.2.11:55969
timer:(on,414ms,0)
The following displays socket summary statistics. This displays the total number of sockets broken down by the type.
Total: 688 (kernel 721)
16 (estab 1, closed 0, orphaned 0, synrecv 0, timewait 0/0), ports 11
Transport Total
What tool do you use to monitor performance on your Linux environment? Did I miss any of your favorite performance monitoring tool? .
If you enjoyed this article, you might also like..
Next post:
Previous post:
& - Practical Examples to Build a Strong Foundation in Linux
- Take Control of Your Bash Command Line and Shell Scripting
- Enhance Your UNIX / Linux Life with Sed and Awk
- Practical Examples for Becoming Fast and Productive in Vim Editor
- Monitor Everything, Be Proactive, and Sleep Well
POPULAR POSTS
CATEGORIES
Copyright &
Ramesh Natarajan. All rights reserved