In this article, we want to introduce you to 'vmstat' command, which is one of the top monitoring tools that Linux/UNIX users should know of. The vmstat command is used to retrieve the statistics of CPU, virtual memory, paging, processes, threads, disks, traps, I/O blocks, and so on.
This useful command will help you a lot in managing, diagnosing, and troubleshooting system resources.
Installation
To use vmstat, your Linux/UNIX systems need to have "sysstat" package installed. In case your operating system does not have it installed yet, run the following command:
For Debian/Ubuntu
$ sudo apt-get install sysstat
On CentOS/RedHat
$ sudo yum install sysstat
or
$ sudo dnf install sysstat
Command Syntax
To use vmstat, refer to the command syntax below:
$ vmstat [options] [delay [count]]
- [options]: There are some options you can use with vmstat command, depending on the use cases. For instance, run command 'man vmstat' to get detailed information about vmstat and its versions.
- [delay]: Displays the delay between updates in seconds. If no delay is specified, only one report is printed with the average values since boot.
- [count]: Outputs number of updates. If no count is specified when delay is defined, the default value is infinite.
Using the vmstat command with examples
Let's take a look at how we can use vmstat command in some example cases.
1. Display memory, process and CPU statistics
By running command vmstat only without any options, output will give us information about processes, memory, swap, IO, system, and CPU.
These outputs are the average for each statistic since the computer was last rebooted. They are not statistics at the time when executing the command.
$ vmstat procs -----------memory----------- ---swap--- ----io---- --system-- ------cpu------ r b swpd free buff cache si so bi bo in cs us sy id wa st 0 0 0 2165936 66984 1087708 0 0 57 13 78 104 0 0 99 1 0
Output explanation
- Procs (Process) column
- r: Number of executable processes. These are processes that have been running or waiting for a new cycle.
- b: The number of processes blocked, waiting for I/O to complete. These processes are in a continuous sleep state. In fact, these processes are not "sleeping", but they are just blocking the system call. When the program calls a function or a service located in the OS kernel, it cannot be interrupted until the operation is completed.
- Memory column
- swpd: the amount of swap memory used
- free: the amount of idle memory
- buff: the amount of memory used as buffers
- cache: the amount of memory used as cache
- Swap column
- si: Amount of memory swapped in from disk
- so: Amount of memory swapped to disk
- IO column
- bi: Blocks are received from a block device. Number of data blocks used to swap virtual memory back to RAM (KiB/s)
- bo: Blocks sent to a block device. Number of data blocks used to swap virtual memory out of RAM and into swap space (KiB/s)
- System column
- in: The number of interrupts per second, including the clock
- cs: The number of context switches per second
- CPU column
- us: Time spent running non-kernel code (user time, including nice time)
- sy: Time spent running kernel code
- id: Time spent idle
- wa: Time spent waiting for IO
- st: Time stolen from a virtual machine. This is the time a virtual machine has to wait for the hypervisor to finish servicing other virtual machines before it can come back and process this virtual machine.
2. Display active and inactive memory
You can get the active or inactive memory information by using '-a' or '--active' option.
$ vmstat -a procs -----------memory---------- ---swap-- -----io---- --system-- ------cpu----- r b swpd free inact active si so bi bo in cs us sy id wa st 0 0 0 2165936 1105960 410608 0 0 68 15 80 107 0 0 99 1 0
3. To monitor statistics in real-time
The syntax of vmstat command has delay value to output statistics in specific intervals of time (second). You can combine delay with option '-t' or '--timestamp' to monitor system resources in every 2 seconds.
$ vmstat -t 2 procs -----------memory---------- ---swap-- -----io---- --system-- ------cpu----- -----timestamp----- r b swpd free buff cache si so bi bo in cs us sy id wa st +07 2 0 0 1956760 72860 1304724 0 0 40 13 75 96 0 0 99 1 0 2023-08-28 14:23:57 0 0 0 1956760 72860 1304724 0 0 0 0 152 172 0 0 100 0 0 2023-08-28 14:23:59 0 0 0 1956760 72860 1304724 0 0 0 0 163 187 0 0 100 0 0 2023-08-28 14:24:01 0 0 0 1956760 72860 1304724 0 0 0 0 165 203 0 0 100 0 0 2023-08-28 14:24:03 0 0 0 1956760 72860 1304724 0 0 0 0 156 178 0 0 100 0 0 2023-08-28 14:24:05
To limit the number output for reporting, assume you only need 7 records in every 5 seconds, run command with 'count' value after 'delay' value:
$ vmstat -t 5 7 procs -----------memory---------- ---swap-- -----io---- --system-- ------cpu----- -----timestamp----- r b swpd free buff cache si so bi bo in cs us sy id wa st +07 0 0 0 1986580 72884 1304724 0 0 39 12 75 96 0 0 99 1 0 2023-08-28 14:29:17 0 0 0 1986580 72884 1304724 0 0 0 0 143 175 0 0 100 0 0 2023-08-28 14:29:22 0 0 0 1986580 72884 1304724 0 0 0 0 129 163 0 0 100 0 0 2023-08-28 14:29:27 0 0 0 1986580 72884 1304724 0 0 0 0 131 170 0 0 100 0 0 2023-08-28 14:29:32 2 0 0 1986580 72884 1304724 0 0 0 0 132 168 0 0 100 0 0 2023-08-28 14:29:37 1 0 0 1986580 72884 1304724 0 0 0 0 138 171 0 0 100 0 0 2023-08-28 14:29:42 0 0 0 1986580 72884 1304724 0 0 0 0 140 167 0 0 100 0 0 2023-08-28 14:29:47
4. Change output units between kilobyte or megabyte
You can use "-S
- k: 1000 bytes
- K: 1024 bytes
- m: 1000000 bytes
- M: 1048576 bytes
$ vmstat procs -----------memory---------- ---swap-- -----io---- -system-- ------cpu----- r b swpd free buff cache si so bi bo in cs us sy id wa st 1 0 0 1976780 73060 1304752 0 0 37 12 75 95 0 0 99 1 0
$ vmstat -S m procs -----------memory---------- ---swap-- -----io---- -system-- ------cpu----- r b swpd free buff cache si so bi bo in cs us sy id wa st 2 0 0 2024 74 1336 0 0 37 12 75 95 0 0 99 1 0
$ vmstat -S M procs -----------memory---------- ---swap-- -----io---- -system-- ------cpu----- r b swpd free buff cache si so bi bo in cs us sy id wa st 1 0 0 1930 71 1274 0 0 37 12 75 95 0 0 99 1 0
$ vmstat -S k procs -----------memory---------- ---swap-- -----io---- -system-- ------cpu----- r b swpd free buff cache si so bi bo in cs us sy id wa st 2 0 0 2023964 74813 1336066 0 0 37 12 75 95 0 0 99 1 0
$ vmstat -S K procs -----------memory---------- ---swap-- -----io---- -system-- ------cpu----- r b swpd free buff cache si so bi bo in cs us sy id wa st 1 0 0 1976528 73060 1304752 0 0 37 12 75 95 0 0 99 1 0
5. Retrieving disks read/write and i/o statistics
'vmstat' command provides option "-d" (d is lowercase) or "–disk" to report disk statistics. Run command as following:
$ vmstat -d
disk- ------------reads------------ ------------writes----------- -----IO------
total merged sectors ms total merged sectors ms cur sec
loop0 14 0 34 178 0 0 0 0 0 0
loop1 52 0 2156 3633 0 0 0 0 0 2
loop2 59 0 2172 2872 0 0 0 0 0 2
loop3 57 0 2188 2582 0 0 0 0 0 2
loop4 81 0 2302 1078 0 0 0 0 0 1
loop5 83 0 880 676 0 0 0 0 0 0
loop6 48 0 718 2637 0 0 0 0 0 2
loop7 43 0 690 794 0 0 0 0 0 0
sda 26306 9352 2073482 613971 8499 10980 683138 12540 0 121
sr0 27 0 2096 3494 0 0 0 0 0 3
dm-0 34116 0 2043738 1656604 19321 0 683128 18900 0 121
dm-1 194 0 8616 328 0 0 0 0 0 0
loop8 50 0 720 6787 0 0 0 0 0 6
loop9 54 0 2138 2307 0 0 0 0 0 1
loop10 360 0 8612 4899 0 0 0 0 0 3
6. Retrieving specific partition read/write statistics
To view statistics regarding a particular partition, use the "-p" or "--partition" option including partition identifier as parameter. With the example below, we try to get the information of "sda3" partition:
$ vmstat -p sda3
sda3 reads read sectors writes requested writes
26001 2055050 9114 691584
The information returned the total number of 'reads' and 'writes' issued to that partition, as well as the number of 'sectors' included.
7. Display summarization of disk statistics
To quickly display summary statistics for disk activity, use the "-D" (D is uppercase) or "--disk-sum" option.
$ vmstat -D
31 disks
3 partitions
63988 total reads
9352 merged reads
4240974 read sectors
2351606 milli reading
29764 writes
11408 merged writes
1387402 written sectors
32699 milli writing
0 inprogress IO
303 milli spent IO
8. Display slab information
A slab represents a contiguous piece of memory, usually made of several virtually contiguous pages. The slab is the actual container of data associated with objects of a specific kind of the containing cache. To retrieve slab information, use the "-m" or "--slabs" option as shown in the following example.
Note: this option is required to be run as sudo privileges.
$ vmstat -m vmstat: your kernel does not support slabinfo or your permissions are insufficient $ sudo vmstat -m Cache Num Total Size Pages isofs_inode_cache 23 23 688 23 ovl_inode 66 66 728 22 nf_conntrack 97 160 256 16 kvm_async_pf 0 0 136 30 kvm_vcpu 0 0 11520 1 kvm_mmu_page_header 0 0 184 22 x86_emulator 0 0 2688 12 ext4_groupinfo_4k 374 374 184 22 …..
- cache: Cache name
- num: Number of currently active objects in the cache
- total: Total number of available objects in the cache
- size: Size of each object in the cache
- pages: Number of memory pages with at least one active object currently associated with the cache
9. Displays a table of various event counters and memory statistics
To display a page of event counters and memory statistics, use the "-s" (s is lowercase) or "--stats" option.
$ vmstat -s
4005240 K total memory
1133636 K used memory
665568 K active memory
1558884 K inactive memory
1347348 K free memory
70840 K buffer memory
1453416 K swap cache
3805180 K total swap
0 K used swap
3805180 K free swap
3029 non-nice user cpu ticks
642 nice user cpu ticks
4449 system cpu ticks
206506 idle cpu ticks
20692 IO-wait cpu ticks
0 IRQ cpu ticks
453 softirq cpu ticks
0 stolen cpu ticks
1258429 pages paged in
436673 pages paged out
0 pages swapped in
0 pages swapped out
300233 interrupts
564701 CPU context switches
1693303116 boot time
5055 forks
Although the reported statistics are largely the same as those generated from the default vmstat command, some of the data is broken down in more detail. For example, the default output combines both 'nice user' and 'non-nice user' CPU times into the 'us' column, while the "--stats" option lists these statistics separately.
$ vmstat procs -----------memory---------- ---swap-- -----io---- --system-- ------cpu----- r b swpd free buff cache si so bi bo in cs us sy id wa st 2 0 0 1347348 71024 1453428 0 0 278 97 89 162 1 1 93 5 0
10. Display fork information
A fork is a Unix or Linux system call to create a new process from an existing running process. To display the number of forks since system is started, run command vmstat with option "-f" or "--forks".
$ vmstat -f
27108 forks
11. Combine vmstat with other commands to modify the output
In some situations, you want to get some specific statistics of output only from vmstat command. By combining vmstat with other commands like 'grep', 'sed' or 'awk', you can display specific output. For example:
Display all detail statistics of only CPU and Memory
$ vmstat -s | grep -i 'cpu\|memory' | sed 's/ *//' 4005240 K total memory 1171764 K used memory 1228184 K active memory 1427588 K inactive memory 848032 K free memory 103044 K buffer memory 12399 non-nice user cpu ticks 4805 nice user cpu ticks 14775 system cpu ticks 1260667 idle cpu ticks 33472 IO-wait cpu ticks 0 IRQ cpu ticks 1064 softirq cpu ticks 0 stolen cpu ticks 2210691 CPU context switches
Display some specific statistics like process (r), memory (free), io (bi, bo), cpu (us, sy, id, wa) and change its header name
$ vmstat | awk ' END {print "No_process_waiting_for_CPU: " $1, "\nMemory_Idle: " $4, "\n\nSwapped_In_memory: " $9, "\nSwapped_Out_Memory: " $10, "\n\nCPU_time_runnning_user_code: " $15, "\nCPU_time_runnning_system_code: " $16, "\n\nInterupts_per_sec: " $13, "\nContext_switches_per_sec: " $14} '| column -t
No_process_waiting_for_CPU: 2
Memory_Idle: 848032
Swapped_In_memory: 161
Swapped_Out_Memory: 258
CPU_time_runnning_user_code: 95
CPU_time_runnning_system_code: 2
Interupts_per_sec: 1
Context_switches_per_sec: 1
Batch Mode
In batch mode, vmstat will print only the data rows in real-time without any headers. This is useful when you need to continuously monitor the values of various system parameters. Here are some examples:
vmstat -n 1 | (while read; do echo "$(date +%Y%m%d.%H%M%S) $REPLY"; done) | tail -n +3
$ vmstat -n 1 | (while read; do echo "$(date +%Y%m%d.%H%M%S) $REPLY"; done) | tail -n +3 20230902.113628 2 0 9376224 4176940 270532 4831568 2 10 23 89 353 41 32 8 60 0 0 20230902.113629 2 0 9376224 4168888 270532 4842960 0 0 0 276 2533 9369 7 4 88 0 0 20230902.113630 1 0 9376224 4175212 270532 4834124 4 0 4 0 2571 8564 10 5 85 0 0 20230902.113631 2 0 9376224 4201604 270532 4833476 0 0 0 152 2483 8038 9 3 88 0 0 20230902.113632 1 0 9376224 4200876 270532 4830404 0 0 0 0 2413 8063 8 3 89 0 0 ^C $
Another command example to print data in batch mode:
while sleep 1 ; do echo `date '+%Y%m%d.%H%M%S'` `vmstat | head -3 | tail -1` ; done
$ while sleep 1 ; do echo `date '+%Y%m%d.%H%M%S'` `vmstat | head -3 | tail -1` ; done 20230902.113715 1 0 9375968 4171416 271128 4843600 2 10 23 89 353 41 32 8 60 0 0 20230902.113716 1 0 9375968 4195572 271128 4830948 2 10 23 89 353 41 32 8 60 0 0 20230902.113717 1 0 9375968 4194060 271128 4828620 2 10 23 89 353 41 32 8 60 0 0 20230902.113718 1 0 9375968 4212476 271136 4828676 2 10 23 89 353 41 32 8 60 0 0 20230902.113719 2 0 9375968 4203164 271136 4824796 2 10 23 89 353 41 32 8 60 0 0 ^C $
Batch mode printing is useful for logging the output for future examination or in the event of some kind of system problem. Then can be logged to a file, or even a database.
Print CPU Usage
Here is a special usecase of vmstat that prints the average cpu usage:
echo $[100-$(vmstat 1 2|tail -1|awk '{print $15}')]
16
This command can be useful to detect the load on the cpu at any given time.
Conclusion
vmstat command can be a useful tool for performance problems related to system resources. We hope you enjoyed and will have more knowledge in the management of Linux systems after going through the examples explained in this article.
