pip fails with HTTP error 503 while getting https://pypi.python

Question

I'm trying to set up a build machine using Jenkins on an Amazon EC2 instance. It's building python, and using the shiningpanda plugin to set up a virtualenv for the build.

Every time I run the build I run:

pip install --use-mirrors --force-reinstall -r requirements.txt 

I've been making builds all day trying to get my coverage and pylint settings right.

Now, at the end of the day, I'm getting these types of errors for a few of the projects:

HTTP error 503 while getting https://pypi.python.org/packages/source/c/coverage/coverage-3.6.tar.gz#md5=67d4e393f4c6a5ffc18605409d2aa1ac (from https://pypi.python.org/simple/coverage/)

Could not install requirement coverage==3.6 (from -r requirements.txt (line 11)) because of error HTTP Error 503: Service Unavailable

If I visit the link in the browser it loads fine.

Why is this happening? Is there an api limit on the pypi api that I'm exceeding? This has been working all day.

One more note, each time i run pip, it fails on a different package. The subsequent build after the error message above coverage downloaded successfully, but I got a 503 error three packages later.

Answer 1

503 usually means a temporary error -- the webserver is not able to service the request due to, for example, temporary overloading.

The fact that it's a different package each time would indicate this kind of transient error. The overloading is probably just a result of lots of other calls coming in at the same time as you.