| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
| |
hence 'python -m psutil.tests' won't work out of the box
|
| |
|
|
|
| |
cpu_freq is not always available on Linux
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
| |
* small changes
* AIX: implement num_ctx_switches
|
| |
|
| |
|
|\ |
|
| |
| |
| | |
The value may come from two different files, must mock both.
|
| | |
|
|/ |
|
| |
|
|
|
|
|
| |
Helps library users know, at a glance, what platforms are tested and
supported. This helps users know if the library is suitable for
integration in an existing project.
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
* Reap test subprocess.
The leftover child process was triggering an assert in a later test:
======================================================================
FAIL: psutil.tests.test_connections.TestConnectedSocketPairs.test_combos
----------------------------------------------------------------------
Traceback (most recent call last):
File "/home/ade/projects/psutil/psutil/tests/__init__.py", line 792, in wrapper
return fun(*args, **kwargs)
File "/home/ade/projects/psutil/psutil/tests/test_connections.py", line 330, in test_combos
self.assertEqual(len(cons), 1)
AssertionError: 0 != 1
* Inherit from Base so that reap_children() cleans up the test processes.
Fixes memory leak test asserts.
psutil/tests/test_memory_leaks.py:334:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
psutil/tests/test_memory_leaks.py:125: in execute
self.assertEqual(thisproc.children(), [])
E AssertionError: Lists differ: [<psutil.Process(pid=908, name... != []
E
E First list contains 10 additional elements.
E First extra element 0:
E <psutil.Process(pid=908, name='python2.7') at 34513317008>
E
E + []
E - [<psutil.Process(pid=908, name='python2.7') at 34513317008>,
E - <psutil.Process(pid=909, name='python2.7') at 34459063760>,
E - <psutil.Process(pid=910, name='python2.7') at 34459064464>,
E - <psutil.Process(pid=911, name='python2.7') at 34513749456>,
E - <psutil.Process(pid=912, name='python2.7') at 34513749520>,
E - <psutil.Process(pid=913, name='python2.7') at 34513751184>,
E - <psutil.Process(pid=914, name='python2.7') at 34513751120>,
E - <psutil.Process(pid=915, name='python2.7') at 34513751312>,
E - <psutil.Process(pid=916, name='python2.7') at 34513751376>,
E - <psutil.Process(pid=917, name='python2.7') at 34513748176>]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Arch Linux and Ubuntu 17.10 use a newer ifconfig version than other distributions and that changes the statistics output text formatting, causing the following tests to fail:
psutil.tests.test_linux.TestSystemNetwork.test_net_if_stats
psutil.tests.test_linux.TestSystemNetwork.test_net_io_counters
MTU becomes lower case, colons are replaced with spaces, and packets and bytes are on the same line.
Example ifconfig output:
enp2s0f0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST> mtu 1500
ether a8:20:66:04:4c:45 txqueuelen 1000 (Ethernet)
RX packets 1396351 bytes 1928499072 (1.7 GiB)
RX errors 0 dropped 0 overruns 0 frame 0
TX packets 750492 bytes 185338978 (176.7 MiB)
TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0
device interrupt 16
|
|
|
|
|
|
|
|
|
|
| |
sh() return value is a string due to Popen(universal_newlines=True).
Traceback (most recent call last):
File ".git/hooks/pre-commit", line 118, in <module>
main()
File ".git/hooks/pre-commit", line 78, in main
py_files = [x for x in out.split(b'\n') if x.endswith(b'.py') and
TypeError: must be str or None, not bytes
|
| |
|
|\ |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
* fix valid_types in memory_percent
pfullmem._fields is always added twice to valid_types so
the message about invalid memtype lists the types twice too.
pfullmem is available on all platforms and is always the
same as or a superset of pmem. We can look at its fields
only to get all valid_types.
Also we can check whether to use memory_full_info or not
by checking the fields of pfullmem vs. pmem instead of
using hard coded mem types.
* remove workaround made for Solaris on AIX
The problem described in the comment doesn't apply for AIX
* update "oneshot" table in documentation
* Removed "nice" and "ionice" which are not boosted
* Removed "~Process." prefix which was only on a few
methods and not others
* Added AIX
Fixes #1157
* small AIX additions to docs
|
|/ |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|\ |
|
| |
| |
| |
| |
| |
| |
| |
| | |
If the TerminateProcess WinAPI function is called with 0, then the
exit code of the terminated process (e.g., observed by its parent)
will be 0. However, this is usually associated with successful
execution. Any other exit code could be used to signal forced
termination, but perhaps the value of SIGTERM is the most
meaningful.
|
| |
| |
| |
| |
| |
| | |
* create_exe should use default code if c_code is True
* fix or skip remaining AIX unit tests
|
|/ |
|
| |
|
| |
|
| |
|
| |
|