[16:55:19] <Guest19490> Hello, I am facing an issue with pip after Ubuntu upgrade 14.10 -> 15.04, but it does look like a generic problem.
[16:55:31] <Guest19490> File "/usr/local/lib/python2.7/dist-packages/pip-7.1.0.dev0-py2.7.egg/pip/_vendor/requests/packages/urllib3/connectionpool.py", line 314, in _raise_timeout if 'timed out' in str(err) or 'did not complete (read)' in str(err): # Python 2.6 TypeError: __str__ returned non-string (type Error)
[16:56:22] <Guest19490> And it is about timing out in /connectionpool.py
[16:58:04] <Guest19490> Does any one would have recommendations on how to overcome this issue?
[16:59:47] <Guest19490> The pip version is 7.1.0, but same was with original pip version as well. A broken dependency?
[20:36:51] <ronny> Guest19490: did you easy_install pip ?!
[20:37:20] <ronny> thats in general the worst way ever to install pip
[20:57:02] <gtback> Guest19490: What are you trying to do with pip? The best I could do to try to reproduce it was run netcat and then try to install a nonexistant "file" from that "webserver" on my local machine. (nc -l 8000; pip install http://localhost:8000/foo.tar.gz) This gave the expected connection timeout message after waiting.
[20:57:50] <Guest19490> Just humble : pip search numpy
[20:59:34] <Guest19490> ronny: what I tried so far ---> 1. Remove/Install python-pip (the original Ubuntu version) 2. easy_install pip 3. pip from github - the lattest dev version
[21:00:30] <Guest19490> each variation gave a list of errors, and they were different once - but they all had in common the latest part - the connect.py failing / timing out
[21:00:46] <gtback> anything crazy about the network you're on? proxies/firewalls/etc.
[21:01:05] <Guest19490> no, jus plain old school connection
[21:01:17] <gtback> I can't reproduce unless I can actually get it to time out at the right spot :-(
[21:02:08] <Guest19490> by the way, since I have certain powers related to my work - I even created Ubuntu 14.10 in a cloud -- pip worked as a charm, then did upgraded to Ubuntu 15.10 -- and got exactly same issue
[21:03:06] <Guest19490> local install and remote were complaining in a same way - a continent apart
[21:03:32] <gtback> if you actually have the source from github, you could try adding some "print(type(err))" or "print(type(err.__str__()))" before that line in urllib3/connectionpool.py
[21:04:17] <gtback> I'm building an Ubuntu 15.04 VM to test on (previously this was on a 14.04 machine I had available)
[21:04:20] <Guest19490> Oh! Perfect - that what I looking for - please give me a couple of mins here.... Thank you for the tip!
[21:11:46] <gtback> sorry, I can't reproduce to help you test. If you get a larger traceback, could you share it as a pastebin? That might help narrow down which request is timing out, too.
[21:20:27] <Guest19490> hm, looks like some type of SSL error:
[22:07:24] <Guest19490> The trace creates 4.3M of data and pastbin won't take it - can I past just last 512k of data? OR use some special flags for strace to make it more informative and smaller?
[22:26:14] <Guest19490> <gtback> In a source code the "traceback" module is used only in basecommand.py - should I reuse "format_exec()" to create traceback for connectionpool.py ?
[22:26:15] <ionelmc> Guest19490: what sort of trace have you made?
[22:37:54] <gtback> Sorry, Guest19490. I didn't mean an strace, but a python stack trace, showing all of the functions executing when the error occurred. I'm not sure if that will show up in the pip output or if you have to pass -v.
[23:00:22] <Guest19490> The additional info "pip -vvv search numpy" brings ---> Starting new HTTPS connection (1): pypi.python.org
[23:01:18] <Guest19490> the rest is previously seen parade of errors
[23:03:02] <Guest19490> <gtback> ... and I can happily ping and nmap pypi.python.org
[23:03:03] <gtback> the parade of errors might be useful, could you pastebin those?