common linux performance monitoring tools

 common linux performance monitoring tools

besides the 'top' command which is the top useful command in linux performance monitoring, see my blog http://rafaxu.blogspot.com.au/2013/12/linux-top-command-in-brief.html, there are a few useful tools for basic performance checking:

System performance checking commands:

#vmstata: give you the basic idea how the system performance.

·         r: The number of processes waiting for run time.

·         b: The number of processes in uninterruptible sleep.

·         swpd: the amount of virtual memory used.

·         free: the amount of idle memory.

·         buff: the amount of memory used as buffers.

·         cache: the amount of memory used as cache.

·         si: Amount of memory swapped in from disk (/s).

·         so: Amount of memory swapped to disk (/s).

·         bi: Blocks received from a block device (blocks/s).

·         bo: Blocks sent to a block device (blocks/s).

·         us: Time spent running non-kernel code. (user time, including nice time)

·         sy: Time spent running kernel code. (system time)

·         id: Time spent idle. Prior to Linux 2.5.41, this includes IO-wait time.

·         wa: Time spent waiting for IO. Prior to Linux 2.5.41, included in idle.

·         st: Time stolen from a virtual machine. Prior to Linux 2.6.11, unknown.

#losf : list of files. Used to list the processes opening files.

It is very useful to check which process occupies the file

Here is the example.

We open a file in python shell

Then if using lsof command the check which process using the file, you will get the python shell PID and other related information.

#tcpdump, a sniffer like tool to capture the packages

Some useful options and scenarios

Check all the packages for one physical interface: tcpdump –i [interface]

Check all the packages from one IP: tcpdump host [ip]

Check all the packages for one port: tcpdump port [port]

Save the packages to the file: tcpdump –w [file]

Read from the file: tcpdump –r [file]

Other options could be found in ‘man tcpdump’ or ‘tcpdump -help’

#netstat: show the network status of the OS

Some useful options and scenarios

Show the routing table:  netstat –r

Show the listening ports: netstat –l –t (tcp sockets listening)

Show all information : netstat –an (do not resolve to the hostname), you can use grep to find the searching information in the results.

#iostat: show the disk IO status of the OS

To collect and shows the disk related information.

tps: Indicate  the  number  of  transfers per second that were issued to the device.

Blk_read/s: blocks read rate per second

Blk_read: the block read numbers

Blk_write/s: blocks write rate per second

Blk_write: the block write numbers

#ps: shows the running processes

Usually used as ‘ps –ef’ to show all active processes

‘ps -auwwx’ to show the long parameters