Process management under Linux

In operating system systems, processes are a very important concept.

I. Knowledge of processes in Linux

1. What is a process?

In layman's terms, the process is a running program. The process descriptor (PID) is the only indicator of the process, and the process is defined and managed by Task_struck and Task_list in the Linux kernel.

2, the classification of the process

1) According to the different modes of Linux operation is divided into:

Kernel mentality: This type of process runs in kernel mode and executes some kernel instructions (Ring 0).

User state: This type of process works in user mode and executes user instructions (Ring 3).

If the user-state process is to perform some kernel-mentality instructions, a system call is generated, and the system call requests the kernel to complete the relevant request and return the result to the user-state process.

2) According to the status of the process can be divided into:

Running state: Running running process

Can interrupt sleep state: The process is asleep, but can be interrupted

Non-disruptive sleep state: the process is asleep but cannot be interrupted

Stop state: stoped will not be dispatched by the kernel

Zombie State: Zombie is caused by the fact that after the process is over, its parent process does not wait for it.

3) According to the intensity of the operation

CPU Intensive: Processes that consume more CPU time while they are running.

I/O intensive: Processes that consume more I/O time during runtime.

Typically, I/o-intensive priorities are higher than CPU-intensive.

4) According to the processing method of the process

Batch process:

Interactive process:

Real-time processes:

3. Priority of the process

The priority of the process is represented by 0-139 numbers, and the number priority is from small to large, in order: 0-99, 139-100.

Priority is divided into 2 categories:

Real-time priority: 0-99, which is maintained by the kernel

Static priority: 100-139, you can use nice to adjust, the value of the Nice value range is [ -20,19], respectively, corresponding to 100 to 139. The nice default value is 0.

Dynamic priority: Dynamically maintained by the kernel and dynamically adjusted.

II. management tools for the process

1. Pstree command to view the number of processes. common usage is as follows:

2. PS command to view the related status of the process. support for SYSV and BSD two style options.

Common selection:

A terminal-related process

x non-terminal processes

U show the user running the process

Common combination options: PS aux

2. # The meanings of each field described below
4. # User processes run as users
6. # PID Process Descriptor is unique
8. # Percentage of CPU when the%CPU process is running
10. # The percentage of memory that the%MEM process runs
12. # VSZ Virtual Memory size
14. # RSS resident Memory set, all memory sets that cannot be swapped out
16. # STAT indicates memory status
18. # The usual statuses are:
20. # S: Can sleep in the middle of the state
22. # R: Run state
24. # D: Non-disruptive sleep state
26. # T: Stop state
28. # Z: Zombie State
30. # S:session leader leader of the so-called process
32. # +: Indicates a foreground process
34. # L: Multithreaded process
36. # N: Low-priority process
38. # <: high-priority process
40. # TTY is used to indicate that the terminal is displayed as "? "is a terminal-independent process
42. # START Process Starting Time
44. # Time Process Runtime
46. # command to execute a process if the command is surrounded by "[]", the kernel thread

-E Show All processes

-F Display Completion format information

Commonly used together: PS-EF

But in some cases, our commands sometimes show incomplete

When you want to display the completion of PS-EFWW

-F: Show additional information

-H: Show hierarchy of processes

Common Combination Method: Ps-efh

We may use the most options in the future:

-O We can customize the display field

1. # commonly used are:
2. # pid command psr pri ni%cpu%mem rsz vsz etc

3, Pgrep,pidof

Pgrep Common selection:

-U View the process number of the specified user

-G View the process number of the specified user group

-L Show process name and process number

Pidof: Displays only the PID of the started process

1. [[email protected] ~]# pidof httpd # Process not started is not displayed
2. [Email protected] ~]# pidof man
3. 5283

4. Top command

Real-time Monitoring system resources

1. # after executing the top command, enter interactive mode

2. # Some of the interactive commands in top:
4. # L: Controls whether the first line is displayed, load balancing information
6. # T: Controls whether to show process information by and CPU information
8. # m: Controls whether memory is displayed, exchange information
10. # I or 1 (number): whether to display each CPU information separately
12. # M: Sorted by%mem, from big to small
14. # K:kill Kill the process
16. # s: Modify default refresh time by default is 3 seconds
18. # The following explains the header information:
20. Top-21:35:17 up 10:03, 4 users, Load average:0.00, 0.00, 0.00
22. # system time Start time logon users load balancing: 1min 5min 15min
24. # What is the system load? This refers to the average number of processes waiting in the process queue
26. # The information shown is equivalent to the uptime command
28. tasks:165 Total, 1 running, 164 sleeping, 0 stopped, 0 zombie
30. # Number of processes running number of sleep states number of dead zombies processes
32. Cpu (s): 0.0%us,0.0%sy, 0.0%ni, 100.0%id, 0.0%wa, 0.0%hi, 0.0%si, 0.0%st
34. # 0.0%us:user Space: Percentage of CPU used by programs running user space
36. # 0.0%sy:system Space: Percentage of CPU used by programs running kernel space
38. # 0.0%ni:nice value call time accounted for CPU hundred percent
40. # 100.0%id: Percentage of system CPU idle
42. # 0.0%wa: Percentage of CPU used to wait for I/O
44. # 0.0%hi: Percent of CPU for hard interrupts
46. # 0.0%si: Percent of CPU for soft interrupts
48. # 0.0%st: The percentage of the CPU that the system was "stolen", which generally refers to the CPU used to run the virtual machine
50. mem:1012548k Total, 396328k used, 616220k free, 99444k buffers
52. # Total memory size use memory size the size of the remaining memory size cache
54. swap:2097144k Total, 0k used, 2097144k free, 144156k cached
56. # Swap The total size of the partition using the remaining buffer size
58. # The information shown here is equivalent to the free command

Common options:

-D #: Specify the Refresh time interval

-B: Show top refresh in batches

-N #: Displayed batches

Example: top-d 4-b 2-n 3

4, Htop

The Htop command is an upgraded version of the top command, which is better than the top command, both functionally and in the interface display.

U: Interactively Select the process that displays the specified user

L: Displays the list of files opened by the process where the cursor is located

S: Displays the system call that the process executing the cursor

A: Bind the process to the specified CPU

#: Quickly position the cursor to the PID # process

is the Htop interface. This tool is more powerful, use more functions, here first to do a brief introduction. In the future I will write Htop blog specifically.

5, Vmstat

Wmstat Viewing virtual usage

2. # Common usage:
4. # Vmstat Display information will be refreshed by default 1 seconds, always displayed
6. # vmstart-n 2 Display information will be refreshed 2 seconds, always displayed
8. # vmstat-n 1 4 Display information is refreshed 1 seconds, refreshed 4 times
10. [Email protected] ~]# Vmstat-n 1 1
12. procs-----------Memory-------------Swap-------io------System-------CPU-----
14. R b swpd free buff cache si so bi bo in CS us sy ID WA St
16. 0 0 0 614392 100468 144776 0 0 2 1 6 5 0 0 100 0 0
18. # Let's explain what each field means:
20. # procs fields about the process
22. # R refers to the number of processes running the queue, which may be low CPU performance if it is too long
24. # b blocks the length of the queue, which is typically used to wait for I/O to complete. If it is too large, it indicates low I/O performance
26. # memory field about RAM use
28. # Swap memory size used by swap
30. # free memory size Its value = Total size-buff-cache-used
32. # Buff buffer size, purpose is to speed up I/O write operations (usually disk)
34. # cache buffer size, mapping is to speed up the I/O read operation (usually disk)
36. # Swap field Description Swap memory
38. # si swapin refers to the rate of data entering the Swap memory unit: KB/S
40. # so swapout refers to the rate of data-out swap memory in units: kb/s
42. # If these 2 values are larger, there will be a jitter phenomenon. Recommended additional memory
44. # Description of Io field I/O
46. # Bi:block in the rate of memory read from the block device kb/s
48. # Bo:block out save to block device rate kb/s
50. # This is what we usually say about disk read and write performance, which can be improved by raid.
52. # system fields about the systems
54. # in:interruppt the rate at which interrupts occur
56. # cs: Rate of context switching (process scheduling)
58. # CPU Field Description CPU Usage
60. # Us:user Space: Percentage of CPU used by programs running user space
62. # Sy:system Space: Percentage of CPU used by programs running kernel space
64. # ID: Percentage of system CPU idle
66. # wa: Percentage of CPU waiting for I/O, due to too many differences in CPU and I/O speed
68. # st: Percentage of the CPU that the system was "stolen", typically referring to the percentage of time spent on the CPU used by the virtual machine to run

6, Nice,renice

Adjusts the priority of the process.

Nice sets the priority when the process starts.

1. # Common parameters:
2. #-N Nice for example: Nice-n 3 httpd
3. # in general, nice values are negative settings that are normally set by an administrator. Normal users can only set nice to positive values.
4. # If you do not specify the-n parameter, the default nice value is 10

Renice Reset the priority of the started process.

1. # Common options are:
2. #-N Nice reset the value of Nice
3. #-P PID setting process PID

7, Kill,killall

For people with Linux C programming experience, we know that one of the ways of IPC communication is through the semaphore (signal), so for the kill and Killall commands, they have a lot to do with the semaphore, or kill, The Killall command allows us to manually pass signals to the process to control the process by signaling the semaphore.

The common semaphore is as follows:

2. [Email protected] ~]# kill-l
4. 1) SIGHUP 2) SIGINT 3) Sigquit 4) Sigill 5) SIGTRAP
6. 6) SIGABRT 7) Sigbus 8) SIGFPE 9) SIGKILL) SIGUSR1
8. One) (SIGSEGV) (SIGUSR2) sigpipe) sigalrm) SIGTERM
10. Sigstkflt) (SIGCHLD) Sigcont SIGSTOP) SIGTSTP
12. (Sigttin) Sigttou () Sigurg) sigxcpu) Sigxfsz
14. (SIGVTALRM) sigprof) sigwinch SIGIO) SIGPWR
16. Sigsys) (sigrtmin) sigrtmin+1) sigrtmin+2 Notoginseng) sigrtmin+3
18. sigrtmin+4) sigrtmin+5 (sigrtmin+6) sigrtmin+7) sigrtmin+8
20. sigrtmin+9) (sigrtmin+10) sigrtmin+11 () sigrtmin+12) sigrtmin+13
22. (sigrtmin+14) sigrtmin+15 () SIGRTMAX-14) SIGRTMAX-13) SIGRTMAX-12
24. SIGRTMAX-11) SIGRTMAX-10 SIGRTMAX-9) SIGRTMAX-8 () SIGRTMAX-7
26. (SIGRTMAX-6) (SIGRTMAX-5) SIGRTMAX-4) SIGRTMAX-3) SIGRTMAX-2
28. SIGRTMAX-1) Sigrtmax
30. # The signals we use are:
32. # 1 SIGHUP When the process is not closed, reread the configuration file. Ngnix is doing quite well in this area and can even be upgraded online.
34. # 2 SIGINT interrupt signal equals CTRL + C
36. # 9 SIGKILL violence kills
38. # Sifterm graceful off by default this is the case
40. # Kill usage is as follows:
42. # Kill [-signal] PID
44. # for signal there are three kinds of representations: for example: 1) -9-15-1-2 2)-sigkill-sighup-sigterm 3)-hup-kill-term-int etc.
46. #
48. # Killall is killing a class of processes
50. # For example: Killall httpd equivalent to kill ' pidof httpd '

8, JOBS,BG,FG

What is a job? A job is a process in which many processes work together to accomplish a specific task. The job has 2 kinds of foreground job and background job.

Use & or CTRL + Z to get a process into the background.

1. # ping 192.168.0.1 &
2. # This runs into the background of the job, quit the terminal when the job will be terminated.
3. # Use the Nohup command to avoid this problem
4. # nohup Ping 192.168.0.1 &
5. # You can use the Jobs command to view jobs in the background
6. # Each job has a job number to identify the job
7. # Job Control commands
8. # BG [[%]job_num]: Let the jobs sent backstage continue to run in the background
9. # FG [[%]job_num]: Returning background jobs to foreground
10. # Kill%job_num: Terminates the specified job

9. Dstat command to view various status information

Usage: dstat [-AFV] [Options ...] [Delay [Count]]

Common parameters	Specific meaning
-C	Display CPU statistics, if there are multiple CPU summary statistics
-D	Show disk statistics, if there are multiple disks, summary statistics
-D DEVICE	Display information for a specific disk
-G	Display page information (specifically what the page is and what he does, as specified in a later Memory tuning blog)
-I.	Show Broken Statistics
-M	Display statistics for memory
-L	Display the load information of the system
-N	To display the related properties of a network interface
-S	Display System Properties
-N Inter_face_name	Display properties for a specific interface
-S	Display properties of swap memory
-P	Show Process Queue
--ipc	Displays IPC Message Queuing, semaphores, and shared memory usage
-A	Equivalent to-cdngy display CPU, disk, NIC, page, System Properties
-F	Display all information in full format,
--TCP,--UDP	Display TCP,UDP status information

See the Man Manual for more parameters.

Example:

10. View Memory Mappings

Pmap PID View memory mapping of the corresponding process, commonly used is: Pman ' pidof process_name '. Of course, this information can also view the/proc/pid/pmap file view.

11. Glances command

A powerful system monitoring tool: real-time monitoring of information such as Cpu,meomory,load,swap,network,mount,disk.

For other features, use H to get help. Because this tool is more powerful, so here is a simple introduction, later in the Htop tool to do a special introduction.

Process management under Linux