Quick links

• Home page
• Blog
• Download
• Forum
• What’s new

About

From project’s home page:

psutil (python system and process utilities) is a cross-platform library for retrieving information on running 
processes and
system utilization (CPU, memory, disks, network) in 
Python. It is useful mainly for 
system monitoring, 
profiling and
limiting process resources and 
management of running processes. It implements many functionalities offered by command line tools such as: 
ps, top, lsof, netstat, ifconfig, who, df, kill, free, nice, ionice, iostat, iotop, uptime, pidof, tty, taskset, pmap. It currently supports 
Linux, Windows, OSX, FreeBSD and 
Sun Solaris, both 
32-bit and 
64-bitarchitectures, with Python versions from 
2.6 to 3.4 (users of Python 2.4 and 2.5 may use 
2.1.3 version). 
PyPy is also known to work.

The psutil documentation you’re reading is distributed as a single HTML page.

CPU

psutil.
cpu_times
(
percpu=False
)
[source]

Return system CPU times as a namedtuple. Every attribute represents the seconds the CPU has spent in the given mode. The attributes availability varies depending on the platform:

• user
• system
• idle
• nice (UNIX)
• iowait (Linux)
• irq (Linux, FreeBSD)
• softirq (Linux)
• steal (Linux 2.6.11+)
• guest (Linux 2.6.24+)
• guest_nice (Linux 3.2.0+)

When percpu is True return a list of nameduples for each logical CPU on the system. First element of the list refers to first CPU, second element to second CPU and so on. The order of the list is consistent across calls. Example output on Linux:

>>>

>>> import psutil
>>> psutil.cpu_times()
scputimes(user=17411.7, nice=77.99, system=3797.02, idle=51266.57, iowait=732.58, irq=0.01, softirq=142.43, steal=0.0, guest=0.0, guest_nice=0.0)

psutil.
cpu_percent
(
interval=None, 
percpu=False
)
[source]

Return a float representing the current system-wide CPU utilization as a percentage. When interval is > 0.0 compares system CPU times elapsed before and after the interval (blocking). When interval is 0.0 or None compares system CPU times elapsed since last call or module import, returning immediately. That means the first time this is called it will return a meaningless 0.0value which you are supposed to ignore. In this case is recommended for accuracy that this function be called with at least 0.1seconds between calls. When percpu is True returns a list of floats representing the utilization as a percentage for each CPU. First element of the list refers to first CPU, second element to second CPU and so on. The order of the list is consistent across calls.

>>>

>>> import psutil
>>> # blocking
>>> psutil.cpu_percent(interval=1)
2.0
>>> # non-blocking (percentage since last call)
>>> psutil.cpu_percent(interval=None)
2.9
>>> # blocking, per-cpu
>>> psutil.cpu_percent(interval=1, percpu=True)
[2.0, 1.0]
>>>

Warning

 

the first time this function is called with interval = 0.0 or None it will return a meaningless 0.0 value which you are supposed to ignore.

psutil.
cpu_times_percent
(
interval=None, 
percpu=False
)
[source]

Same as cpu_percent() but provides utilization percentages for each specific CPU time as is returned bypsutil.cpu_times(percpu=True). interval and percpu arguments have the same meaning as in cpu_percent().

Warning

 

the first time this function is called with interval = 0.0 or None it will return a meaningless 0.0 value which you are supposed to ignore.

psutil.
cpu_count
(
logical=True
)
[source]

Return the number of logical CPUs in the system (same as os.cpu_count() in Python 3.4). If logical is False return the number of physical cores only (hyper thread CPUs are excluded). Return None if undetermined.

>>>

>>> import psutil
>>> psutil.cpu_count()
4
>>> psutil.cpu_count(logical=False)
2
>>>

Memory

psutil.
virtual_memory
(
)
[source]

Return statistics about system memory usage as a namedtuple including the following fields, expressed in bytes:

• total: total physical memory available.
• available: the actual amount of available memory that can be given instantly to processes that request more memory in bytes; this is calculated by summing different memory values depending on the platform (e.g. free + buffers + cached on Linux) and it is supposed to be used to monitor actual memory usage in a cross platform fashion.
• percent: the percentage usage calculated as (total - available) / total * 100.
• used: memory used, calculated differently depending on the platform and designed for informational purposes only.
• free: memory not being used at all (zeroed) that is readily available; note that this doesn’t reflect the actual memory available (use ‘available’ instead).

Platform-specific fields:

• active: (UNIX): memory currently in use or very recently used, and so it is in RAM.
• inactive: (UNIX): memory that is marked as not used.
• buffers: (Linux, BSD): cache for things like file system metadata.
• cached: (Linux, BSD): cache for various things.
• wired: (BSD, OSX): memory that is marked to always stay in RAM. It is never moved to disk.
• shared: (BSD): memory that may be simultaneously accessed by multiple processes.

The sum of used and available does not necessarily equal total. On Windows available and free are the same. Seeexamples/meminfo.py script providing an example on how to convert bytes in a human readable form.

>>>

>>> import psutil
>>> mem = psutil.virtual_memory()
>>> mem
svmem(total=8374149120L, available=1247768576L, percent=85.1, used=8246628352L, free=127520768L, active=3208777728, inactive=1133408256, buffers=342413312L, cached=777834496)
>>>
>>> THRESHOLD = 100 * 1024 * 1024  # 100MB
>>> if mem.available <= THRESHOLD:
...     print("warning")
...
>>>

psutil.
swap_memory
(
)
[source]

Return system swap memory statistics as a namedtuple including the following fields:

• total: total swap memory in bytes
• used: used swap memory in bytes
• free: free swap memory in bytes
• percent: the percentage usage
• sin: the number of bytes the system has swapped in from disk (cumulative)
• sout: the number of bytes the system has swapped out from disk (cumulative)

sin and sout on Windows are meaningless and are always set to 0. See examples/meminfo.py script providing an example on how to convert bytes in a human readable form.

>>>

>>> import psutil
>>> psutil.swap_memory()
sswap(total=2097147904L, used=886620160L, free=1210527744L, percent=42.3, sin=1050411008, sout=1906720768)

Disks

psutil.
disk_partitions
(
all=False
)
[source]

Return all mounted disk partitions as a list of namedtuples including device, mount point and filesystem type, similarly to “df” command on UNIX. If all parameter is False return physical devices only (e.g. hard disks, cd-rom drives, USB keys) and ignore all others (e.g. memory partitions such as /dev/shm). Namedtuple’s fstype field is a string which varies depending on the platform. On Linux it can be one of the values found in /proc/filesystems (e.g. 'ext3' for an ext3 hard drive o 'iso9660' for the CD-ROM drive). On Windows it is determined via GetDriveType and can be either "removable", "fixed", "remote", "cdrom", "unmounted" or "ramdisk". On OSX and FreeBSD it is retrieved via getfsstat(2). See disk_usage.py script providing an example usage.

>>>

>>> import psutil
>>> psutil.disk_partitions()
[sdiskpart(device='/dev/sda3', mountpoint='/', fstype='ext4', opts='rw,errors=remount-ro'),
 sdiskpart(device='/dev/sda7', mountpoint='/home', fstype='ext4', opts='rw')]

psutil.
disk_usage
(
path
)
[source]

Return disk usage statistics about the given path as a namedtuple including total, used and free space expressed in bytes, plus the percentage usage. OSError is raised if path does not exist. See examples/disk_usage.py script providing an example usage. Starting from Python 3.3 this is also available as shutil.disk_usage(). See disk_usage.py script providing an example usage.

>>>

>>> import psutil
>>> psutil.disk_usage('/')
sdiskusage(total=21378641920, used=4809781248, free=15482871808, percent=22.5)

psutil.
disk_io_counters
(
perdisk=False
)
[source]

Return system-wide disk I/O statistics as a namedtuple including the following fields:

• read_count: number of reads
• write_count: number of writes
• read_bytes: number of bytes read
• write_bytes: number of bytes written
• read_time: time spent reading from disk (in milliseconds)
• write_time: time spent writing to disk (in milliseconds)

If perdisk is True return the same information for every physical disk installed on the system as a dictionary with partition names as the keys and the namedutuple described above as the values. See examples/iotop.py for an example application.

>>>

>>> import psutil
>>> psutil.disk_io_counters()
sdiskio(read_count=8141, write_count=2431, read_bytes=290203, write_bytes=537676, read_time=5868, write_time=94922)
>>>
>>> psutil.disk_io_counters(perdisk=True)
{'sda1': sdiskio(read_count=920, write_count=1, read_bytes=2933248, write_bytes=512, read_time=6016, write_time=4),
 'sda2': sdiskio(read_count=18707, write_count=8830, read_bytes=6060, write_bytes=3443, read_time=24585, write_time=1572),
 'sdb1': sdiskio(read_count=161, write_count=0, read_bytes=786432, write_bytes=0, read_time=44, write_time=0)}

Network

psutil.
net_io_counters
(
pernic=False
)
[source]

Return system-wide network I/O statistics as a namedtuple including the following attributes:

• bytes_sent: number of bytes sent
• bytes_recv: number of bytes received
• packets_sent: number of packets sent
• packets_recv: number of packets received
• errin: total number of errors while receiving
• errout: total number of errors while sending
• dropin: total number of incoming packets which were dropped
• dropout: total number of outgoing packets which were dropped (always 0 on OSX and BSD)

If pernic is True return the same information for every network interface installed on the system as a dictionary with network interface names as the keys and the namedtuple described above as the values. See examples/nettop.py for an example application.

>>>

>>> import psutil
>>> psutil.net_io_counters()
snetio(bytes_sent=14508483, bytes_recv=62749361, packets_sent=84311, packets_recv=94888, errin=0, errout=0, dropin=0, dropout=0)
>>>
>>> psutil.net_io_counters(pernic=True)
{'lo': snetio(bytes_sent=547971, bytes_recv=547971, packets_sent=5075, packets_recv=5075, errin=0, errout=0, dropin=0, dropout=0),
'wlan0': snetio(bytes_sent=13921765, bytes_recv=62162574, packets_sent=79097, packets_recv=89648, errin=0, errout=0, dropin=0, dropout=0)}

psutil.
net_connections
(
kind=’inet’
)
[source]

Return system-wide socket connections as a list of namedutples. Every namedtuple provides 7 attributes:

• fd: the socket file descriptor, if retrievable, else -1. If the connection refers to the current process this may be passed tosocket.fromfd() to obtain a usable socket object.
• family: the address family, either AF_INET, AF_INET6 or AF_UNIX.
• type: the address type, either SOCK_STREAM or SOCK_DGRAM.
• laddr: the local address as a (ip, port) tuple or a path in case of AF_UNIX sockets.
• raddr: the remote address as a (ip, port) tuple or an absolute path in case of UNIX sockets. When the remote endpoint is not connected you’ll get an empty tuple (AF_INET*) or None (AF_UNIX). On Linux AF_UNIX sockets will always have this set toNone.
• status: represents the status of a TCP connection. The return value is one of the psutil.CONN_* constants (a string). For UDP and UNIX sockets this is always going to be psutil.CONN_NONE.
• pid: the PID of the process which opened the socket, if retrievable, else None. On some platforms (e.g. Linux) the availability of this field changes depending on process privileges (root is needed).

The kind parameter is a string which filters for connections that fit the following criteria:

Kind value	Connections using
“inet”	IPv4 and IPv6
“inet4”	IPv4
“inet6”	IPv6
“tcp”	TCP
“tcp4”	TCP over IPv4
“tcp6”	TCP over IPv6
“udp”	UDP
“udp4”	UDP over IPv4
“udp6”	UDP over IPv6
“unix”	UNIX socket (both UDP and TCP protocols)
“all”	the sum of all the possible families and protocols

To get per-process connections use Process.connections(). Also, see netstat.py sample script. Example:

>>>

>>> import psutil
>>> psutil.net_connections()
[pconn(fd=115, family=2, type=1, laddr=('10.0.0.1', 48776), raddr=('93.186.135.91', 80), status='ESTABLISHED', pid=1254),
 pconn(fd=117, family=2, type=1, laddr=('10.0.0.1', 43761), raddr=('72.14.234.100', 80), status='CLOSING', pid=2987),
 pconn(fd=-1, family=2, type=1, laddr=('10.0.0.1', 60759), raddr=('72.14.234.104', 80), status='ESTABLISHED', pid=None),
 pconn(fd=-1, family=2, type=1, laddr=('10.0.0.1', 51314), raddr=('72.14.234.83', 443), status='SYN_SENT', pid=None)
 ...]

Note

 

(OSX) psutil.AccessDenied is always raised unless running as root (lsof does the same).

Note

 

(Solaris) UNIX sockets are not supported.

New in version 2.1.0.

Other system info

psutil.
users
(
)
[source]

Return users currently connected on the system as a list of namedtuples including the following fields:

• user: the name of the user.
• terminal: the tty or pseudo-tty associated with the user, if any, else None.
• host: the host name associated with the entry, if any.
• started: the creation time as a floating point number expressed in seconds since the epoch.

Example:

>>>

>>> import psutil
>>> psutil.users()
[suser(name='giampaolo', terminal='pts/2', host='localhost', started=1340737536.0),
 suser(name='giampaolo', terminal='pts/3', host='localhost', started=1340737792.0)]

psutil.
boot_time
(
)
[source]

Return the system boot time expressed in seconds since the epoch. Example:

>>>

>>> import psutil, datetime
>>> psutil.boot_time()
1389563460.0
>>> datetime.datetime.fromtimestamp(psutil.boot_time()).strftime("%Y-%m-%d %H:%M:%S")
'2014-01-12 22:51:00'

Functions

psutil.
pids
(
)
[source]

Return a list of current running PIDs. To iterate over all processes process_iter() should be preferred.

psutil.
pid_exists
(
pid
)
[source]

Check whether the given PID exists in the current process list. This is faster than doing "pid in psutil.pids()" and should be preferred.

psutil.
process_iter
(
)
[source]

Return an iterator yielding a Process class instance for all running processes on the local machine. Every instance is only created once and then cached into an internal table which is updated every time an element is yielded. Cached Process instances are checked for identity so that you’re safe in case a PID has been reused by another process, in which case the cached instance is updated. This is should be preferred over psutil.pids() for iterating over processes. Sorting order in which processes are returned is based on their PID. Example usage:

import psutil

for proc in psutil.process_iter():
    try:
        pinfo = proc.as_dict(attrs=['pid', 'name'])
    except psutil.NoSuchProcess:
        pass
    else:
        print(pinfo)

psutil.
wait_procs
(
procs, 
timeout=None, 
callback=None
)
[source]

Convenience function which waits for a list of Process instances to terminate. Return a (gone, alive) tuple indicating which processes are gone and which ones are still alive. The gone ones will have a new returncode attribute indicating process exit status (it may be None). callback is a function which gets called every time a process terminates (a Process instance is passed as callback argument). Function will return as soon as all processes terminate or when timeout occurs. Tipical use case is:

• send SIGTERM to a list of processes
• give them some time to terminate
• send SIGKILL to those ones which are still alive

Example:

import psutil

def on_terminate(proc):
    print("process {} terminated".format(proc))

procs = [...]  # a list of Process instances
for p in procs:
    p.terminate()
gone, alive = wait_procs(procs, timeout=3, callback=on_terminate)
for p in alive:
    p.kill()

Exceptions

class 
psutil.
Error
[source]

Base exception class. All other exceptions inherit from this one.

class 
psutil.
NoSuchProcess
(
pid, 
name=None, 
msg=None
)
[source]

Raised by Process class methods when no process with the given pid is found in the current process list or when a process no longer exists. “name” is the name the process had before disappearing and gets set only if Process.name() was previosly called.

class 
psutil.
AccessDenied
(
pid=None, 
name=None, 
msg=None
)
[source]

Raised by Process class methods when permission to perform an action is denied. “name” is the name of the process (may beNone).

class 
psutil.
TimeoutExpired
(
seconds, 
pid=None, 
name=None, 
msg=None
)
[source]

Raised by Process.wait() if timeout expires and process is still alive.

Process class

class 
psutil.
Process
(
pid=None
)
[source]

Represents an OS process with the given pid. If pid is omitted current process pid (os.getpid()) is used. Raise NoSuchProcess if piddoes not exist. When accessing methods of this class always be prepared to catch NoSuchProcess and AccessDenied exceptions.hash() builtin can be used against instances of this class in order to identify a process univocally over time (the hash is determined by mixing process PID and creation time). As such it can also be used with set()s.

Warning

 

the way this class is bound to a process is uniquely via its PID. That means that if the Process instance is old enough and the PID has been reused by another process in the meantime you might end up interacting with another process. The only exceptions for which process identity is pre-emptively checked (via PID + creation time) and guaranteed are for nice()(set), ionice() (set), cpu_affinity() (set), rlimit() (set), children(), parent(), suspend() resume(), send_signal(), terminate(), and kill()methods. To prevent this problem for all other methods you can use is_running() before querying the process or use process_iter()in case you’re iterating over all processes.

pid
[source]

The process PID.

ppid
(
)
[source]

The process parent pid. On Windows the return value is cached after first call.

name
(
)
[source]

The process name. The return value is cached after first call.

exe
(
)
[source]

The process executable as an absolute path. On some systems this may also be an empty string. The return value is cached after first call.

cmdline
(
)
[source]

The command line this process has been called with.

create_time
(
)
[source]

The process creation time as a floating point number expressed in seconds since the epoch, in UTC. The return value is cached after first call.

>>>

>>> import psutil, datetime
>>> p = psutil.Process()
>>> p.create_time()
1307289803.47
>>> datetime.datetime.fromtimestamp(p.create_time()).strftime("%Y-%m-%d %H:%M:%S")
'2011-03-05 18:03:52'

as_dict
(
attrs=None, 
ad_value=None
)
[source]

Utility method returning process information as a hashable dictionary. If attrs is specified it must be a list of strings reflecting available Process class’s attribute names (e.g. ['cpu_times', 'name']) else all public (read only) attributes are assumed. ad_valueis the value which gets assigned to a dict key in case AccessDenied exception is raised when retrieving that particular process information.

>>>

>>> import psutil
>>> p = psutil.Process()
>>> p.as_dict(attrs=['pid', 'name', 'username'])
{'username': 'giampaolo', 'pid': 12366, 'name': 'python'}

parent
(
)
[source]

Utility method which returns the parent process as a Process object pre-emptively checking whether PID has been reused. If no parent PID is known return None.

status
(
)
[source]

The current process status as a string. The returned string is one of the psutil.STATUS_* constants.

cwd
(
)
[source]

The process current working directory as an absolute path.

username
(
)
[source]

The name of the user that owns the process. On UNIX this is calculated by using real process uid.

uids
(
)
[source]

The real, effective and saved user ids of this process as a nameduple. This is the same as os.getresuid() but can be used for every process PID.

Availability: UNIX

gids
(
)
[source]

The real, effective and saved group ids of this process as a nameduple. This is the same as os.getresgid() but can be used for every process PID.

Availability: UNIX

terminal
(
)
[source]

The terminal associated with this process, if any, else None. This is similar to “tty” command but can be used for every process PID.

Availability: UNIX

nice
(
value=None
)
[source]

Get or set process niceness (priority). On UNIX this is a number which usually goes from -20 to 20. The higher the nice value, the lower the priority of the process.

>>>

>>> import psutil
>>> p = psutil.Process()
>>> p.nice(10)  # set
>>> p.nice()  # get
10
>>>

On Windows this is available as well by using GetPriorityClass and SetPriorityClass and value is one of thepsutil.*_PRIORITY_CLASS constants. Example which increases process priority on Windows:

>>>

>>> p.nice(psutil.HIGH_PRIORITY_CLASS)

Starting from Python 3.3 this same functionality is available as os.getpriority() and os.setpriority().

ionice
(
ioclass=None, 
value=None
)
[source]

Get or set process I/O niceness (priority). On Linux ioclass is one of the psutil.IOPRIO_CLASS_* constants. value is a number which goes from 0 to 7. The higher the value, the lower the I/O priority of the process. On Windows only ioclass is used and it can be set to 2 (normal), 1 (low) or 0 (very low). The example below sets IDLE priority class for the current process, meaning it will only get I/O time when no other process needs the disk:

>>>

>>> import psutil
>>> p = psutil.Process()
>>> p.ionice(psutil.IOPRIO_CLASS_IDLE)  # set
>>> p.ionice()  # get
pionice(ioclass=3, value=0)
>>>

On Windows only ioclass is used and it can be set to 2 (normal), 1 (low) or 0 (very low).

Availability: Linux and Windows > Vista

rlimit
(
resource, 
limits=None
)
[source]

Get or set process resource limits (see man prlimit). resource is one of the psutil.RLIMIT_* constants. limits is a (soft, hard)tuple. This is the same as resource.getrlimit() and resource.setrlimit() but can be used for every process PID and only on Linux. Example:

>>>

>>> import psutil
>>> p = psutil.Process()
>>> # process may open no more than 128 file descriptors
>>> p.rlimit(psutil.RLIMIT_NOFILE, (128, 128))
>>> # process may create files no bigger than 1024 bytes
>>> p.rlimit(psutil.RLIMIT_FSIZE, (1024, 1024))
>>> # get
>>> p.rlimit(psutil.RLIMIT_FSIZE)
(1024, 1024)
>>>

Availability: Linux

io_counters
(
)
[source]

Return process I/O statistics as a namedtuple including the number of read and write operations performed by the process and the amount of bytes read and written. For Linux refer to /proc filesysem documentation. On BSD there’s apparently no way to retrieve bytes counters, hence -1 is returned for read_bytes and write_bytes fields. OSX is not supported.

>>>

>>> import psutil
>>> p = psutil.Process()
>>> p.io_counters()
pio(read_count=454556, write_count=3456, read_bytes=110592, write_bytes=0)

Availability: all platforms except OSX

num_ctx_switches
(
)
[source]

The number voluntary and involuntary context switches performed by this process.

num_fds
(
)
[source]

The number of file descriptors used by this process.

Availability: UNIX

num_handles
(
)
[source]

The number of handles used by this process.

Availability: Windows

num_threads
(
)
[source]

The number of threads currently used by this process.

threads
(
)
[source]

Return threads opened by process as a list of namedtuples including thread id and thread CPU times (user/system).

cpu_times
(
)
[source]

Return a tuple whose values are process CPU user and system times which means the amount of time expressed in seconds that a process has spent in user / system mode. This is similar to os.times() but can be used for every process PID.

cpu_percent
(
interval=None
)
[source]

Return a float representing the process CPU utilization as a percentage. When interval is > 0.0 compares process times to system CPU times elapsed before and after the interval (blocking). When interval is 0.0 or None compares process times to system CPU times elapsed since last call, returning immediately. That means the first time this is called it will return a meaningless 0.0 value which you are supposed to ignore. In this case is recommended for accuracy that this function be called a second time with at least 0.1 seconds between calls. Example:

>>>

>>> import psutil
>>> p = psutil.Process()
>>>
>>> # blocking
>>> p.cpu_percent(interval=1)
2.0
>>> # non-blocking (percentage since last call)
>>> p.cpu_percent(interval=None)
2.9
>>>

Note

 

a percentage > 100 is legitimate as it can result from a process with multiple threads running on different CPU cores.

Warning

 

the first time this method is called with interval = 0.0 or None it will return a meaningless 0.0 value which you are supposed to ignore.

cpu_affinity
(
cpus=None
)
[source]

Get or set process current CPU affinity. CPU affinity consists in telling the OS to run a certain process on a limited set of CPUs only. The number of eligible CPUs can be obtained with list(range(psutil.cpu_count())).

>>>

>>> import psutil
>>> psutil.cpu_count()
4
>>> p = psutil.Process()
>>> p.cpu_affinity()  # get
[0, 1, 2, 3]
>>> p.cpu_affinity([0])  # set; from now on, process will run on CPU #0 only
>>> p.cpu_affinity()
[0]
>>>
>>> # reset affinity against all CPUs
>>> all_cpus = list(range(psutil.cpu_count()))
>>> p.cpu_affinity(all_cpus)
>>>

Availability: Linux, Windows, BSD

Changed in version 2.2.0: added support for FreeBSD

memory_info
(
)
[source]

Return a tuple representing RSS (Resident Set Size) and VMS (Virtual Memory Size) in bytes. On UNIX rss and vms are the same values shown by ps. On Windows rss and vms refer to “Mem Usage” and “VM Size” columns of taskmgr.exe. For more detailed memory stats use memory_info_ex().

memory_info_ex
(
)
[source]

Return a namedtuple with variable fields depending on the platform representing extended memory information about the process. All numbers are expressed in bytes.

Linux	OSX	BSD	SunOS	Windows
rss	rss	rss	rss	num_page_faults
vms	vms	vms	vms	peak_wset
shared	pfaults	text		wset
text	pageins	data		peak_paged_pool
lib		stack		paged_pool
data				peak_nonpaged_pool
dirty				nonpaged_pool
				pagefile
				peak_pagefile
				private

Windows metrics are extracted from PROCESS_MEMORY_COUNTERS_EX structure. Example on Linux:

>>>

>>> import psutil
>>> p = psutil.Process()
>>> p.memory_info_ex()
pextmem(rss=15491072, vms=84025344, shared=5206016, text=2555904, lib=0, data=9891840, dirty=0)

memory_percent
(
)
[source]

Compare physical system memory to process resident memory (RSS) and calculate process memory utilization as a percentage.

memory_maps
(
grouped=True
)
[source]

Return process’s mapped memory regions as a list of nameduples whose fields are variable depending on the platform. As such, portable applications should rely on namedtuple’s path and rss fields only. This method is useful to obtain a detailed representation of process memory usage as explained here. If grouped is True the mapped regions with the same path are grouped together and the different memory fields are summed. If grouped is False every mapped region is shown as a single entity and the namedtuple will also include the mapped region’s address space (addr) and permission set (perms). Seeexamples/pmap.py for an example application.

>>>

>>> import psutil
>>> p = psutil.Process()
>>> p.memory_maps()
[pmmap_grouped(path='/lib/x8664-linux-gnu/libutil-2.15.so', rss=16384, anonymous=8192, swap=0),
 pmmap_grouped(path='/lib/x8664-linux-gnu/libc-2.15.so', rss=6384, anonymous=15, swap=0),
 pmmap_grouped(path='/lib/x8664-linux-gnu/libcrypto.so.0.1', rss=34124, anonymous=1245, swap=0),
 pmmap_grouped(path='[heap]', rss=54653, anonymous=8192, swap=0),
 pmmap_grouped(path='[stack]', rss=1542, anonymous=166, swap=0),
 ...]
>>>

children
(
recursive=False
)
[source]

Return the children of this process as a list of Process objects, pre-emptively checking whether PID has been reused. If recursive is True return all the parent descendants. Example assuming A == this process:

A ─┐
   │
   ├─ B (child) ─┐
   │             └─ X (grandchild) ─┐
   │                                └─ Y (great grandchild)
   ├─ C (child)
   └─ D (child)

>>> p.children()
B, C, D
>>> p.children(recursive=True)
B, X, Y, C, D

Note that in the example above if process X disappears process Y won’t be returned either as the reference to process A is lost.

open_files
(
)
[source]

Return regular files opened by process as a list of namedtuples including the absolute file name and the file descriptor number (on Windows this is always -1). Example:

>>>

>>> import psutil
>>> f = open('file.ext', 'w')
>>> p = psutil.Process()
>>> p.open_files()
[popenfile(path='/home/giampaolo/svn/psutil/file.ext', fd=3)]

connections
(
kind=”inet”
)
[source]

Return socket connections opened by process as a list of namedutples. To get system-wide connections usepsutil.net_connections(). Every namedtuple provides 6 attributes:

• fd: the socket file descriptor. This can be passed to socket.fromfd() to obtain a usable socket object. This is only available on UNIX; on Windows -1 is always returned.
• family: the address family, either AF_INET, AF_INET6 or AF_UNIX.
• type: the address type, either SOCK_STREAM or SOCK_DGRAM.
• laddr: the local address as a (ip, port) tuple or a path in case of AF_UNIX sockets.
• raddr: the remote address as a (ip, port) tuple or an absolute path in case of UNIX sockets. When the remote endpoint is not connected you’ll get an empty tuple (AF_INET) or None (AF_UNIX). On Linux AF_UNIX sockets will always have this set to None.
• status: represents the status of a TCP connection. The return value is one of the psutil.CONN_* constants. For UDP and UNIX sockets this is always going to be psutil.CONN_NONE.

The kind parameter is a string which filters for connections that fit the following criteria:

Kind value	Connections using
“inet”	IPv4 and IPv6
“inet4”	IPv4
“inet6”	IPv6
“tcp”	TCP
“tcp4”	TCP over IPv4
“tcp6”	TCP over IPv6
“udp”	UDP
“udp4”	UDP over IPv4
“udp6”	UDP over IPv6
“unix”	UNIX socket (both UDP and TCP protocols)
“all”	the sum of all the possible families and protocols

Example:

>>>

>>> import psutil
>>> p = psutil.Process(1694)
>>> p.name()
'firefox'
>>> p.connections()
[pconn(fd=115, family=2, type=1, laddr=('10.0.0.1', 48776), raddr=('93.186.135.91', 80), status='ESTABLISHED'),
 pconn(fd=117, family=2, type=1, laddr=('10.0.0.1', 43761), raddr=('72.14.234.100', 80), status='CLOSING'),
 pconn(fd=119, family=2, type=1, laddr=('10.0.0.1', 60759), raddr=('72.14.234.104', 80), status='ESTABLISHED'),
 pconn(fd=123, family=2, type=1, laddr=('10.0.0.1', 51314), raddr=('72.14.234.83', 443), status='SYN_SENT')]

is_running
(
)
[source]

Return whether the current process is running in the current process list. This is reliable also in case the process is gone and its PID reused by another process, therefore it must be preferred over doing psutil.pid_exists(p.pid).

Note

 

this will return True also if the process is a zombie (p.status() == psutil.STATUS_ZOMBIE).

send_signal
(
signal
)
[source]

Send a signal to process (see signal module constants) pre-emptively checking whether PID has been reused. This is the same as os.kill(pid, sig). On Windows only SIGTERM is valid and is treated as an alias for kill().

suspend
(
)
[source]

Suspend process execution with SIGSTOP signal pre-emptively checking whether PID has been reused. On UNIX this is the same as os.kill(pid, signal.SIGSTOP). On Windows this is done by suspending all process threads execution.

resume
(
)
[source]

Resume process execution with SIGCONT signal pre-emptively checking whether PID has been reused. On UNIX this is the same as os.kill(pid, signal.SIGCONT). On Windows this is done by resuming all process threads execution.

terminate
(
)
[source]

Terminate the process with SIGTERM signal pre-emptively checking whether PID has been reused. On UNIX this is the same as os.kill(pid, signal.SIGTERM). On Windows this is an alias for kill().

kill
(
)
[source]

Kill the current process by using SIGKILL signal pre-emptively checking whether PID has been reused. On UNIX this is the same as os.kill(pid, signal.SIGKILL). On Windows this is done by using TerminateProcess.

wait
(
timeout=None
)
[source]

Wait for process termination and if the process is a children of the current one also return the exit code, else None. On Windows there’s no such limitation (exit code is always returned). If the process is already terminated immediately returnNone instead of raising NoSuchProcess. If timeout is specified and process is still alive raise TimeoutExpired exception. It can also be used in a non-blocking fashion by specifying timeout=0 in which case it will either return immediately or raiseTimeoutExpired. To wait for multiple processes use psutil.wait_procs().

Popen class

class 
psutil.
Popen
(
*args, 
**kwargs
)
[source]

A more convenient interface to stdlib subprocess.Popen. It starts a sub process and deals with it exactly as when usingsubprocess.Popen but in addition it also provides all the methods of psutil.Process class in a single interface. For method names common to both classes such as send_signal(), terminate() and kill() psutil.Process implementation takes precedence. For a complete documentation refer to subprocess module documentation.

Note

 

Unlike subprocess.Popen this class pre-emptively checks wheter PID has been reused on send_signal(), terminate() andkill() so that you don’t accidentally terminate another process, fixing http://bugs.python.org/issue6973.

>>>

>>> import psutil
>>> from subprocess import PIPE
>>>
>>> p = psutil.Popen(["/usr/bin/python", "-c", "print('hello')"], stdout=PIPE)
>>> p.name()
'python'
>>> p.username()
'giampaolo'
>>> p.communicate()
('hello\n', None)
>>> p.wait(timeout=2)
0
>>>