Python pip raising NewConnectionError while installing libraries

Question

I've Python 3 running in a linux server. I need to install some libraries (obviously) so I'm trying :

pip3 install numpy

Which, is resulting in the following error:

Collecting numpy
  Retrying (Retry(total=4, connect=None, read=None, redirect=None, status=None)) after connection broken by 'NewConnectionError('<urllib3.connection.VerifiedHTTPSConnection object at 0x7f7542572828>: Failed to establish a new connection: [Errno -2] Name or service not known',)': /simple/numpy/
  Retrying (Retry(total=3, connect=None, read=None, redirect=None, status=None)) after connection broken by 'NewConnectionError('<urllib3.connection.VerifiedHTTPSConnection object at 0x7f7542572eb8>: Failed to establish a new connection: [Errno -2] Name or service not known',)': /simple/numpy/
  Retrying (Retry(total=2, connect=None, read=None, redirect=None, status=None)) after connection broken by 'NewConnectionError('<urllib3.connection.VerifiedHTTPSConnection object at 0x7f7542572be0>: Failed to establish a new connection: [Errno -2] Name or service not known',)': /simple/numpy/
  Retrying (Retry(total=1, connect=None, read=None, redirect=None, status=None)) after connection broken by 'NewConnectionError('<urllib3.connection.VerifiedHTTPSConnection object at 0x7f7542572d30>: Failed to establish a new connection: [Errno -2] Name or service not known',)': /simple/numpy/
  Retrying (Retry(total=0, connect=None, read=None, redirect=None, status=None)) after connection broken by 'NewConnectionError('<urllib3.connection.VerifiedHTTPSConnection object at 0x7f7542572860>: Failed to establish a new connection: [Errno -2] Name or service not known',)': /simple/numpy/
  Could not find a version that satisfies the requirement numpy (from versions: )
No matching distribution found for numpy

Questions:

1. What could be the problem? Why is this error being raised?
2. How can this be avoided in the future?

Feel free to ask for more details.

UPDATE: I tried ping google.com and got the error:

ping: google.com: Name or service not known

But when I tried ping 8.8.8.8, I got:

PING 8.8.8.8 (8.8.8.8) 56(84) bytes of data.
64 bytes from 8.8.8.8: icmp_seq=1 ttl=118 time=10.8 ms
64 bytes from 8.8.8.8: icmp_seq=2 ttl=118 time=10.6 ms
64 bytes from 8.8.8.8: icmp_seq=3 ttl=118 time=10.6 ms
64 bytes from 8.8.8.8: icmp_seq=4 ttl=118 time=10.7 ms
64 bytes from 8.8.8.8: icmp_seq=5 ttl=118 time=10.6 ms
64 bytes from 8.8.8.8: icmp_seq=6 ttl=118 time=10.7 ms
64 bytes from 8.8.8.8: icmp_seq=7 ttl=118 time=10.7 ms
64 bytes from 8.8.8.8: icmp_seq=8 ttl=118 time=10.7 ms
64 bytes from 8.8.8.8: icmp_seq=9 ttl=118 time=10.7 ms
64 bytes from 8.8.8.8: icmp_seq=10 ttl=118 time=10.7 ms
64 bytes from 8.8.8.8: icmp_seq=11 ttl=118 time=10.7 ms
64 bytes from 8.8.8.8: icmp_seq=12 ttl=118 time=10.8 ms
64 bytes from 8.8.8.8: icmp_seq=13 ttl=118 time=10.7 ms
64 bytes from 8.8.8.8: icmp_seq=14 ttl=118 time=10.7 ms
64 bytes from 8.8.8.8: icmp_seq=15 ttl=118 time=10.7 ms
64 bytes from 8.8.8.8: icmp_seq=16 ttl=118 time=10.7 ms
64 bytes from 8.8.8.8: icmp_seq=17 ttl=118 time=10.6 ms
64 bytes from 8.8.8.8: icmp_seq=18 ttl=118 time=10.7 ms
64 bytes from 8.8.8.8: icmp_seq=19 ttl=118 time=10.6 ms
64 bytes from 8.8.8.8: icmp_seq=20 ttl=118 time=10.7 ms
64 bytes from 8.8.8.8: icmp_seq=21 ttl=118 time=10.7 ms
64 bytes from 8.8.8.8: icmp_seq=22 ttl=118 time=10.7 ms
64 bytes from 8.8.8.8: icmp_seq=23 ttl=118 time=10.7 ms
64 bytes from 8.8.8.8: icmp_seq=24 ttl=118 time=10.7 ms
64 bytes from 8.8.8.8: icmp_seq=25 ttl=118 time=10.7 ms
64 bytes from 8.8.8.8: icmp_seq=26 ttl=118 time=10.7 ms
^C
--- 8.8.8.8 ping statistics ---
26 packets transmitted, 26 received, 0% packet loss, time 25046ms
rtt min/avg/max/mdev = 10.655/10.731/10.827/0.073 ms

A problem with DNS maybe? What should I do?

Ministry

I am facing this problem on Kaggle not locally. And it's specific to the installation of NLPAUG model. WARNING: Retrying (Retry(total=0, connect=None, read=None, redirect=None, status=None)) after connection broken by 'NewConnectionError('<pip._vendor.urllib3.connection.HTTPSConnection object at 0x7a25f260eb60>: Failed to establish a new connection: [Errno -3] Temporary failure in name resolution')': /simple/nlpaug/ ERROR: Could not find a version that satisfies the requirement nlpaug (from versions: none) ERROR: No matching distribution found for nlpaug – Starobin

Answer 1

All I had to do was add nameserver 8.8.8.8 to the /etc/resolv.conf file. 8.8.8.8 is Google's public free DNS server's IP. So, since a system must know the IP address of a domain name, we have to give it a way to do just that. A public DNS server provides this binding of a domain name to the IP address.

Ministry

Had this exact same issue with WSL and Virtualenv so thank you! – Carthusian

Neat solution!!! – Soundproof

Also had to restart WSL :) – Blowing

This is magic! I have suffered for months with this. Thanks a lot – Imitation

How to fix this for Windows container? – Cabot

1) I actually had to remove the nameserver lines. 2) Just exiting the WSL was not enough for me. I had to use a powershell and do wsl --shutdown to really quit it. – Zeeba

@AndreasStorvikStrauman how did you restart wsl? – Redcoat

Had to do wsl --shutdown for changes in /etc/resolv.conf file to take effect, as said by @Zeeba – Worked

Answer 2

For Windows 10 User, Go to Settings -> Network and Internet -> Proxy

Disable Automatically detect settings

Jolley

Classic windows! I had to restart the computer to make this work. – Ellga

I just wanted to say that I love you. – Taveras

Thanks for the tip. It worked. I didn't have to restart after making this change. – Testosterone

This isn't worked for me. Any other suggestion where I might be wrong? – Cabot

Unfortunately I don't feel this is much of a solution, as on a company issued laptop proxy settings can be controlled via group policy thus meaning the user doesn't have the ability to just turn off various Windows settings – Seeder

Thank you! I use WSL. I disabled the "Automatically detect settings" and then shut down WSL by running wsl --shutdown then started it by running bash, and it worked! – Toga

Answer 3

If using proxy network and facing same issue then use

pip install --proxy="proxy:portno" numpy

Foe example:

C:\Users\Dell>pip install --proxy="10.50.225.80:3128" numpy

Maryannemarybella

Thanks, such a life saver ! My Windows automatic proxy set up was already turned off. This finally triggered the installation. – Joyance

Thank you for the solution, now it works. Previously I had tried to set the proxy globally, by modifying ~/.bashrc, as shown bellow. However it didn't work. export http_proxy="proxy_url:proxy_port" export HTTP_PROXY="proxy_url:proxy_port" export https_proxy="proxy_url:proxy_port" export HTTPS_PROXY="proxy_url:proxy_port" export ftp_proxy="proxy_url:proxy_port" export FTP_PROXY="proxy_url:proxy_port" export all_proxy="proxy_url:proxy_port" export ALL_PROXY="proxy_url:proxy_port" – Metaxylem

It worked on my company laptop! To find the proxy on Windows I guessed that the port number was 8080 (most of the internal proxies work on port 8080), then I opened cmd and run netstat -a -n and I searched for 8080 with CTRL+F and I looked for an address with state "ESTABLISHED". Then I run pip install --proxy="proxy_found_in_cmd:8080" numpy and it worked! – Roseliaroselin

Answer 4

I just restarted the network services and then did docker restart. Commands used:

systemctl restart NetworkManager.service
sudo service docker restart

Fawcette

This solution worked for me. I was using Ubuntu 22.04 virtual machine in VMWare. – Wholewheat

This worked for me without restarting NetworkManager (so, I only restarted docker service). Maybe it was caused by turning my VPN off and back on again some time before running docker-compose up. – Adam

Answer 5

if you are trying to build a docker image and see this error use --network=host at the end of your command.

docker build -t {name} . --network=host

Rabelais

this helped me ,i was running the same command but inside the WSL (Ubuntu), thank you @mohammad – Lemmy

Answer 6

Check the proxy configuration inside proxy settings. Then use :

 pip install --proxy=http://yourproxyurl:port numpy(Python package name)

Britneybritni

I found it better to set environment vars for proxy to avoid errors with setup.py for modules. export http_proxy=<your-proxy:port>; export https_proxy=<your-https-proxy:port> – Hightower

This works but for some reason, proxy does not apply to downloading dependencies, only main package is downloaded. For dependencies, I get same error – Bruxelles

Answer 7

Not an answer to the original question.

Retrying (Retry(total=4, connect=None, read=None, redirect=None, status=None)) after connection broken by 'NewConnectionError('<urllib3.connection.VerifiedHTTPSConnection object at 0x7f7542572828>: Failed to establish a new connection: [Errno -2] Name or service not known',)':

The same warning occurs to me when I was writing Dockerfile.
Obviously, it's the network problem.
However, the result of DNS and ping from host to pip source is fine.
Eventually, I deleted the Dockerfile from the start of the pip part to the end.
After I exec into the container, the network is unable to get the DNS server.

Remind by my coworker, there was a change about iptables.
When a change made to iptables, the server needs to restart the docker engine to make it work.
It could be but not limit to sudo service docker restart
However, hope this answer could save some others' time.

Trejo

Answer 8

I have this problem when I use VPN. Not sure what's the reason for this, but switching it off solves the problem.

Thirddegree

Restarting docker (see Romita's answer above) seems to fix this. – Adam

I have the same issue. Did you find out why it would not allow connection on VPN? Are there security vulnerabilities that the network admin, domain admin, or vpn admin would not allow connection for pip or python installation? – Labradorite

Answer 9

I resolved a similar issue by updating the DNS configuration to a more reliable one (Google nameservers). Follow the below steps to append it

sudo nano /etc/resolv.conf

Move to the end of the file and add below

nameserver 8.8.8.8
nameserver 8.8.4.4

Save and exit

Note: I was using Ubuntu

Sidero

Answer 10

You can close all your proxy programs and leave the Windows proxy section blank. I had this problem with Lantern and Psiphon.

Lyndsaylyndsey

Answer 11

I had this problem on Raspbian 10 after intalling network-manager following the instructions in this link:

https://www.giovanni.panozzo.it/guides/raspbian-stretch-network-manager-openvpn.html

After undoing following step, DNS works again!

sudo ln -sf /run/resolvconf/resolv.conf /etc/resolv.conf

(luckily I saved the original file!)

Suctorial

Answer 12

It's purely network issue. so you just check your network speed for the pip package installations or restart your network and check again with pip install {{package_name}}

Fiducial

Answer 13

I ran into this issue using docker build. What is your docker version? Upgrading to the latest version helped me solve the issue!

Docker version 23.0.3

Quillen

Answer 14

If you are on window use this :

py -m pip install --proxy="http://username:passwd@proxy-ip:port" package_name

Palmation

Answer 15

In my case i needed to add hosts 'pypi.python.org' and 'pypi.org' to firewall' whitelist

Makepeace

Answer 16

I know that some other people have pointed out the fact that you can run pip with a proxy, but I'd like to point out that you can do this "globally" too. I am developing in an Ubuntu 20 VM, and for me, running

export http_proxy=[proxy info]
export https_proxy=[proxy info]
pip install flask

worked.

Unleash