1. Overview
It's quite common to find ourselves in a situation where we need to know the resource usage of each process and thread in our system. For example, we might want to know which process is slowing down our system.
In this tutorial, we'll look at how to can get this kind of insight using the top command.
2. A Default Interface
We can use top by simply typing top in the command line, after which we'll get an interactive interface:
top
top - 04:05:27 up 3 days, 12:02, 1 user, load average: 0.55, 1.06, 1.27
Tasks: 362 total, 2 running, 290 sleeping, 0 stopped, 0 zombie
%Cpu(s): 35.8 us, 10.7 sy, 0.0 ni, 52.4 id, 0.3 wa, 0.0 hi, 0.7 si, 0.0 st
KiB Mem : 8060436 total, 150704 free, 4438276 used, 3471456 buff/cache
KiB Swap: 2097148 total, 1656152 free, 440996 used. 2557604 avail Mem
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
32081 abhishe+ 20 0 879676 198164 106096 S 152.6 2.5 0:10.16 chrome
582 abhishe+ 20 0 51448 4088 3372 R 15.8 0.1 0:00.04 top
503 root -51 0 0 0 0 S 5.3 0.0 11:05.61 irq/130-iwlwifi
875 message+ 20 0 53120 5900 3204 S 5.3 0.1 10:10.14 dbus-daemon
6855 abhishe+ 20 0 1564544 170444 22924 S 5.3 2.1 75:21.88 deluge-gtk
1 root 20 0 225840 7200 4720 S 0.0 0.1 4:51.28 systemd
2 root 20 0 0 0 0 S 0.0 0.0 0:00.20 kthreadd
4 root 0 -20 0 0 0 I 0.0 0.0 0:00.00 kworker/0:0H
6 root 0 -20 0 0 0 I 0.0 0.0 0:00.00 mm_percpu_wq
This interactive screen is divided into four sections:
- Summary
- Fields/Columns Header
- Input/Message Line
- Tasks
2.1. Summary
The first line consists of five things:
- name of the window
- current time
- length of time since last boot
- total number of users, and
- system load averaged over the last 1, 5 and 15 minutes
top - 04:05:27 up 3 days, 12:02, 1 user, load average: 0.55, 1.06, 1.27
We can see that the second line gives the count of various processes and threads, divided into four categories: running, sleeping, stopped, and zombie.
And the next line tells us about the CPU state percentage, that is, time taken by user and kernel processes:
Tasks: 362 total, 2 running, 290 sleeping, 0 stopped, 0 zombie %Cpu(s): 35.8 us, 10.7 sy, 0.0 ni, 52.4 id, 0.3 wa, 0.0 hi, 0.7 si, 0.0 st
The meaning of various symbols in the above example is the time taken by the CPU in running various processes:
- us – user processes (that are defined without any user-defined priority – un-niced user processes)
- sy – kernel processes
- ni – niced user processes
- id – kernel idle handler
- wa – I/O completion
- hi – hardware interrupts
- si – software interrupts
- st – time stolen from this VM by the hypervisor
We can notice that line 4 describes the state of physical memory; while line 5 describes the virtual memory:
KiB Mem : 8060436 total, 150704 free, 4438276 used, 3471456 buff/cache KiB Swap: 2097148 total, 1656152 free, 440996 used. 2557604 avail Mem
3. The top Headers
As we can see in the example given above there are various fields describing the status of various processes and threads.
Let's learn about the meaning of these headers one-by-one:
- PID (Process ID): The unique id of the task that is defined by task_struct – it's used by the kernel to identify any process
- USER (User Name): The effective username of the task's owner
- PR (Priority): The scheduling priority of the task. The rt values under this field mean that the task is running under real-time scheduling prioritization
- NI (Nice Value): Also depicts the priority of the task. The difference between PR and NI is that PR is the real priority of a process as seen by the kernel, while NI is just a priority hint for the kernel. A negative nice value means higher priority, whereas a positive nice value means lower priority
- TIME+ (CPU Time): Depicts the total CPU time the task has used since it started, having the granularity of hundredths of a second
- COMMAND (Command Name): Displays the command line used to start a task or the name of the associated program
3.1. Memory Headers
The headers that are used to summarize various parameters related to memory are described below:
- VIRT (Virtual Memory Size in KiB): Depicts the total amount of virtual memory used by the task. Virtual memory includes all code, data, and shared libraries. It also includes pages that have been swapped out and pages that have been mapped but not used
- RES (Resident Memory Size in KiB): Stands for a subset of the virtual memory space (VIRT) representing the non-swapped physical memory a task is currently using
- SHR (Shared Memory Size in KiB): Stands for a subset of resident memory (RES) that may be used by other processes
- %CPU (CPU Usage): Stands for the task's share of the elapsed CPU time since the last screen update, expressed as a percentage of total CPU time. A value greater than 100% can be reported for a multi-threaded process when top is not running in Threads Mode
- %MEM (Memory Usage -RES): The task's current share of available physical memory
4. Interactive Commands
We can interact with the top interface using various commands:
- The simplest one being seeing the help menu by pressing the h button.
- We can use the d or s button to change the refresh rate of top. The default refresh rate is 3.0 seconds.
- To quit from the top interface, we can press the q button.
We can kill a task by pressing the k button, after that the “Input Line” will be active and we'll need to enter the PID of the task.
We can also change the renice value of a task by pressing the r button. After that, we'll enter the PID and then the renice value of that task. Ordinary users can only increase the nice value and are prevented from lowering it.
We can change the unit used for showing memory in the Summary Area from KiB by pressing E:
MiB Mem : 7871.520 total, 995.176 free, 4501.594 used, 2374.750 buff/cache MiB Swap: 2047.996 total, 1607.332 free, 440.664 used. 2275.230 avail Mem
To change the memory unit used in the Task Area, we can press e:
22011 abhishe+ 20 0 4049.7m 266.1m 138.3m S 13.2 3.4 18:08.67 gnome-shell 920 cyberea+ 20 0 2545.5m 110.4m 8.6m S 7.9 1.4 92:37.54 cybereason-sens 1554 abhishe+ 20 0 489.2m 69.9m 53.0m S 6.6 0.9 97:43.26 Xorg 6855 abhishe+ 20 0 1536.8m 174.6m 21.6m S 6.6 2.2 85:00.29 deluge-gtk 23393 abhishe+ 20 0 1689.2m 197.4m 63.4m S 6.0 2.5 3:09.83 _Postman
Both of these will lead to the cycling of memory units starting from KiB and going all the way up to EiB (exbibytes).
4.1. Global Modes
There are different modes that are useful in various cases, one of those being Threads Mode.
By default, top displays a summation of all threads in each process. We can change this by pressing the H button. After this top will display individual threads of each process:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 6855 abhishe+ 20 0 1573660 178760 22124 S 2.6 2.2 45:11.77 deluge-gtk 6899 abhishe+ 20 0 1573660 178760 22124 S 2.3 2.2 37:41.68 deluge-gtk
As we can notice in the previous example, the application named deluge was mentioned only once as the underlying threads were not shown, while in this example we can see two different threads that are used by this application.
The other mode is Solaris Mode, which can be toggled off by pressing the I button. When operating under this mode, a task's CPU usage will be divided by the total number of CPUs.
4.2. Interaction With the Task Window
We can change the fields that are displayed and their order by pressing the f button. The field menu will open up and then we can select the fields to be shown, their order, sort by fields, etc.
One of the most useful views that is presented by top is Forest View Mode. In this mode, the tasks will be ordered like a tree and all child tasks will be aligned under their respective parents:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
1 root 20 0 225840 7196 4716 S 0.0 0.1 5:46.79 systemd
283 root 19 -1 148972 37300 36300 S 0.0 0.5 0:23.01 `- systemd-journal
336 root 20 0 47060 4000 2528 S 0.0 0.0 0:01.10 `- systemd-udevd
862 systemd+ 20 0 146112 1276 1208 S 0.0 0.0 0:00.35 `- systemd-timesyn
864 systemd+ 20 0 71072 4556 3916 S 0.0 0.1 0:12.47 `- systemd-resolve
867 root 20 0 70728 3732 3448 S 0.0 0.0 0:03.05 `- systemd-logind
871 root 20 0 38428 2748 2652 S 0.0 0.0 0:00.27 `- cron
We can use the x key to highlight the sorted field. We can use the > and < to change the sorted field to the right or left respectively. Some fields have direct key bindings for their sorting, M for %MEM, N for PID, P for %CPU, T for TIME+.
5. Command-Line Options
We can use top in batch mode by passing the -b flag. When in batch mode top doesn't receive any input and will run till the program is killed. This is quite useful for passing the output of top command to some other program or file.
To fix the number of iterations, we can use -n flag:
top -b -n10
To change the refresh rate, we can use the -d flag. We can use fractional seconds with this flag:
top -d2.5
To see all the output fields supported by top, we can use the -O flag:
top -O PID PPID UID USER ... more output omitted
We can use these field names to define the sorting order by passing it after the -o flag. So, if we want to sort the output of top by virtual memory, we can use:
top -o VIRT PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 23584 abhishe+ 20 0 14.593g 554600 58412 S 2.3 6.9 10:25.55 _Postman 22011 abhishe+ 20 0 4142400 277884 141424 S 0.7 3.4 22:00.86 gnome-shell 1183 gdm 20 0 3664328 114104 72160 S 0.0 1.4 6:33.79 gnome-shell 2008 abhishe+ 20 0 2782760 22520 15096 S 0.0 0.3 0:35.15 copyq
Next, we can use various filters for monitoring tasks on the basis of PIDs, users, etc. To filter task on the basis of PIDs, we can pass up to 20 PIDs with the -p flag:
top -p23584,22011 PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 22011 abhishe+ 20 0 4144624 276900 141368 S 6.2 3.4 22:16.92 gnome-shell 23584 abhishe+ 20 0 14.593g 554600 58412 S 0.0 6.9 10:29.91 _Postman
Finally, to filter on the basis of users, we can use the -u flag:
top -u root PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 503 root -51 0 0 0 0 S 6.2 0.0 12:55.09 irq/130-iwlwifi 1 root 20 0 225840 7196 4716 S 0.0 0.1 5:43.72 systemd
6. Conclusion
In this tutorial, we saw how top is useful in knowing the memory usage of various processes and threads and monitor them.
We saw its interactive screen and explored the meaning and use of various fields.
We also saw various handy command-line options.