Every single command from pip runs super slow

[peidaqi]

Environment

• pip version: pip 20.0.2 from /usr/lib/python3/dist-packages/pip (python 3.8)
• Python version: 3.8.2
• OS: Ubuntu 20.04 (Windows WSL2 - kernel 4.19.104-microsoft-standard) on Windows 10 (19041)

Description

Any commands on pip3 runs super slow including simple ones like:
pip3 list

It used to take 1~2 seconds and now it's like a minute.

Expected behavior

How to Reproduce

Tried to clean cache dir and it didn't work.
Tried to purge python3-pip package and reinstall and didn't work.

Not sure if it's linked to the recent Windows 10 19041 update.

[uranusjr]

What are you referring by “it used to”? Does this slowdown happen because of a pip upgrade or a system upgrade? If it happens out of nowhere, it is very likely not a pip problem, but something happening on your particular machine, which pip does not have control over.

[peidaqi]

No, I didn't do any pip upgrade. Windows 10 does make frequent updates but I cannot fathom how this impacts performance with pip inside WSL2.

I'm hoping someone could point me a direction on how I can even trace this problem. Right now pip doesn't output any error messages so no way to know what's going on.

[pradyunsg]

Could you share how many packages are already installed (i.e. the output of pip list)? It might be related to pip's internal logic [1] that tried to look at all the installed packages before any output processing.

[1] pkg_resources's WorkingSet construction on load, for those wondering what I'm talking about

[piyushchauhan2011]

Same issue with me too. I ran pip3 list command and it took around 10+ seconds to list out packages. For now I create virtual environment using pipenv which removes the problem. I think it might be interfering with the sharing of executables between wsl2 linux and windows. I'm not sure what is the source of problem though!

[uranusjr]

I think it might be interfering with the sharing of executables between wsl2 linux and windows

This sounds plausible. WSL2’s filesystem performance is terrible if you access the Windows filesystem at the Linux side. What is the Python associated to your pip3 command? Can you provide its sys.path? Does this happen if you run pip3 at a different location? Does it matter if the location is in the Linux filesystem or the Windows side?

[piyushchauhan2011]

If we run same pip3 list on windows powershell, it's instant and the issue doesn't happen.

Sys path without pipenv environment activated

['', '/usr/lib/python38.zip', '/usr/lib/python3.8', '/usr/lib/python3.8/lib-dynload', '/home/<user>/.local/lib/python3.8/site-packages', '/usr/local/lib/python3.8/dist-packages', '/usr/lib/python3/dist-packages']

Sys path with pipenv environment activated

['', '/usr/lib/python38.zip', '/usr/lib/python3.8', '/usr/lib/python3.8/lib-dynload', '/home/<user>/.local/share/virtualenvs/myproj-SiazyaGz/lib/python3.8/site-packages']

It could be possible if sys path is changed for global, then pip3 won't interfere with windows one. I haven't tried it out!

[uranusjr]

Hmm, none of the directories seem out of ordinary. Do you happen to mount any of the Windows directories? For example, do you symlink your home directory (or anything listed in sys.path) to a Windows directory? Or do you run the command in a directory under /mnt?

I would probably try to temporarily move some of the entries listed in sys.path away (espacially the ~/.local/lib one) and see if that changes anything. Or maybe you can just insert a few profilers into the Python runtime and see what exactly are slowing things down. There are many tricks you can do to isolate the problem. That would be immensely more useful than me (or anyone without physical access to your machine) trying to troubleshoot against thin air.

[ngraymon]

Edit: This may somehow be linked to the display? The slowdown largely goes away (takes 0.5 seconds) when I start up an X11 server (using MobaXterm). The reason I found this issue was that matplotlib was super slow so I tried to use pip to reinstall. I forgot that I needed to run an Xterm to use matplotlib.

I have also encountered this problem and have the same output from my python's sys.path as piyushchauhan2011. I have a symlink in my home directory to a Windows directory like so
test -> /mnt/c/Users/<user>/Documents/<git_project_folder>/

I'm developing a python package whose results need to be bench marked against a program that can only run in a unix environment. I use sublime test to edit the files in my windows directory and use the WSL2 to run the benchmark code on those files through the symlink.

I can run the following commands without any slowdown: check, show, config
I'm using pip 20.0.2 from /usr/lib/python3/dist-packages/pip (python 3.8)

Running pip3 list does return a result but it takes ~ 30 seconds.
I have about 100 packages installed.

I tried the following without any success
python3 -m pip --retries 2 --timeout 5 --no-cache-dir --isolated --verbose list
I have tried running pip both inside the WSL2 filesystem and at the Windows side each of them suffering the same slowdown.

I'm not 100% what the best way of changing the sys.path would be but here is my attempt:
I launched ipython3 which starts with a sys.path of:

'/usr/bin',
 '/usr/lib/python38.zip',
 '/usr/lib/python3.8',
 '/usr/lib/python3.8/lib-dynload',
 '',
 '/home/<user>/.local/lib/python3.8/site-packages',
 '/usr/local/lib/python3.8/dist-packages',
 '/usr/lib/python3/dist-packages',
 '/usr/lib/python3/dist-packages/IPython/extensions',
 '/home/<user>/.ipython']

Which I backup as follows backup = sys.path.copy()
At this point I confirmed that if I use run '/usr/bin/pip3' list I still encounter the slowdown.
However, when after that I set sys.path = [] and run again I get a ModuleNotFoundError: No module named 'pyparsing'. This result repeats each time I run again. BUT! Once I set sys.path = backup now run '/usr/bin/pip3' list miraculously works!
The output of using time:

CPU times: user 12.2 ms, sys: 426 µs, total: 12.6 ms
Wall time: 11.8 ms

So clearly there is something wrong with the paths.
Afterwards I can again set sys.path =[] and run '/usr/bin/pip3' list still works for some reason.

I'm not sure if this is relevant but I figured I'd mention it:
After using the run command my sys.path gets populated as follows (after having been set to an empty list)

['/usr/share/python-wheels/idna-2.8-py2.py3-none-any.whl',
 '/usr/share/python-wheels/distlib-0.3.0-py2.py3-none-any.whl',
 '/usr/share/python-wheels/msgpack-0.6.2-py2.py3-none-any.whl',
 '/usr/share/python-wheels/lockfile-0.12.2-py2.py3-none-any.whl',
 '/usr/share/python-wheels/pytoml-0.1.21-py2.py3-none-any.whl',
 '/usr/share/python-wheels/retrying-1.3.3-py2.py3-none-any.whl',
 '/usr/share/python-wheels/requests-2.22.0-py2.py3-none-any.whl',
 '/usr/share/python-wheels/setuptools-44.0.0-py2.py3-none-any.whl',
 '/usr/share/python-wheels/pep517-0.8.2-py2.py3-none-any.whl',
 '/usr/share/python-wheels/chardet-3.0.4-py2.py3-none-any.whl',
 '/usr/share/python-wheels/webencodings-0.5.1-py2.py3-none-any.whl',
 '/usr/share/python-wheels/CacheControl-0.12.6-py2.py3-none-any.whl',
 '/usr/share/python-wheels/ipaddr-2.2.0-py2.py3-none-any.whl',
 '/usr/share/python-wheels/certifi-2019.11.28-py2.py3-none-any.whl',
 '/usr/share/python-wheels/urllib3-1.25.8-py2.py3-none-any.whl',
 '/usr/share/python-wheels/wheel-0.34.2-py2.py3-none-any.whl',
 '/usr/share/python-wheels/appdirs-1.4.3-py2.py3-none-any.whl',
 '/usr/share/python-wheels/packaging-20.3-py2.py3-none-any.whl',
 '/usr/share/python-wheels/html5lib-1.0.1-py2.py3-none-any.whl',
 '/usr/share/python-wheels/six-1.14.0-py2.py3-none-any.whl',
 '/usr/share/python-wheels/pip-20.0.2-py2.py3-none-any.whl',
 '/usr/share/python-wheels/colorama-0.4.3-py2.py3-none-any.whl',
 '/usr/share/python-wheels/progress-1.5-py2.py3-none-any.whl',
 '/usr/share/python-wheels/pkg_resources-0.0.0-py2.py3-none-any.whl',
 '/usr/share/python-wheels/pyparsing-2.4.6-py2.py3-none-any.whl',
 '/usr/share/python-wheels/contextlib2-0.6.0-py2.py3-none-any.whl',
 '/usr/share/python-wheels/distro-1.4.0-py2.py3-none-any.whl',
 '/usr/bin',
 '/usr/lib/python38.zip',
 '/usr/lib/python3.8',
 '/usr/lib/python3.8/lib-dynload',
 '',
 '/home/<user>/.local/lib/python3.8/site-packages',
 '/usr/local/lib/python3.8/dist-packages',
 '/usr/lib/python3/dist-packages',
 '/usr/lib/python3/dist-packages/IPython/extensions',
 '/home/<user>/.ipython']

Which still suffers the slowdown until the sys.path is set to an empty list and then reset back to the original list or this list.

[uranusjr]

This may somehow be linked to the display? The slowdown largely goes away (takes 0.5 seconds) when I start up an X11 server (using MobaXterm). The reason I found this issue was that matplotlib was super slow so I tried to use pip to reinstall. I forgot that I needed to run an Xterm to use matplotlib.

Maybe…? The entirely issue is very weird to me. If this is a sys.path issue, wouldn’t the same slowdown happen for all Python imports, not just pip? I am quite stumped 😞

[MattiaFailla]

Hello, my environment is:

DISTRIB_ID=Ubuntu
DISTRIB_RELEASE=18.04
DISTRIB_CODENAME=bionic
DISTRIB_DESCRIPTION="Ubuntu 18.04.4 LTS"
NAME="Ubuntu"
VERSION="18.04.4 LTS (Bionic Beaver)"
ID=ubuntu
ID_LIKE=debian
PRETTY_NAME="Ubuntu 18.04.4 LTS"
VERSION_ID="18.04"
HOME_URL="https://www.ubuntu.com/"
SUPPORT_URL="https://help.ubuntu.com/"
BUG_REPORT_URL="https://bugs.launchpad.net/ubuntu/"
PRIVACY_POLICY_URL="https://www.ubuntu.com/legal/terms-and-policies/privacy-policy"
VERSION_CODENAME=bionic
UBUNTU_CODENAME=bionic

With Python 3.6.9 and pip 9.0.1 and I can confirm that pip is extremely slow with every command (especially with install)

My pip3 list is

asn1crypto (0.24.0)
attrs (17.4.0)
Automat (0.6.0)
chardet (3.0.4)
configobj (5.0.6)
constantly (15.1.0)
cryptography (2.1.4)
distro-info (0.18ubuntu0.18.04.1)
hyperlink (17.3.1)
idna (2.6)
incremental (16.10.1)
keyring (10.6.0)
keyrings.alt (3.0)
netifaces (0.10.4)
pip (9.0.1)
pyasn1 (0.4.2)
pyasn1-modules (0.2.1)
pycrypto (2.6.1)
pygobject (3.26.1)
pyOpenSSL (17.5.0)
python-apt (1.6.5+ubuntu0.3)
python-debian (0.1.32)
pyxdg (0.25)
PyYAML (3.12)
SecretStorage (2.3.1)
service-identity (16.0.0)
setuptools (39.0.1)
six (1.11.0)
Twisted (17.9.0)
ufw (0.36)
unattended-upgrades (0.1)
wheel (0.30.0)

[ngraymon]

@MattiaFailla If you want a temporary fix I recommend running an X11 server such as MobaXterm(or some other equivalent). I don't understand why, but it did fix my slowdown issue for all commands.

[MattiaFailla]

@ngraymon that’s odd but I will try this temp-fix.
Thanks!
I'll update this issue once tried.

[ngraymon]

Just to confirm that the behavior is still the same I checked just now:

Running time pip3 list inside a Windows Terminal on WSL2:

After starting MobaXterm and running time pip3 list in the same terminal:

[MattiaFailla]

@ngraymon Hello,
I have resolved the issue, please try the following steps:

• Do not run pip commands with sudo
• apt-update && apt-upgrade
• Reboot the server/computer
• Keep close attention to docker, last night I noticed that the python3 process was being intensely used by the swarm