The following bug details the type of broken http server that causes a problem: https://fedorahosted.org/fedora-infrastructure/ticket/3859 In this case, where most download clients (curl, chrome, firefox, etc) will download a gzipped file and leave it compressed, the "requests" library follows the letter of RFC2616 and unzips the file. This causes a problem for us, because the code requesting a download expects to find a file that is still gzipped. This caused a complete failure to sync a fedora 18 repo. (link in the bug listed above) Nectar should thus make some intelligent decisions about ignoring the content-encoding header when its value is "x-gzip" and the file ends in '.gz'.
https://github.com/pulp/nectar/pull/5
build: 2.2.0-18.beta
Correction: build: 2.2.0-0.18.beta
verified [root@ibm-x3550m3-11 ~]# rpm -qa |grep pulp-server pulp-server-2.2.0-0.20.beta.fc18.noarch [root@ibm-x3550m3-11 ~]# [root@ibm-x3550m3-11 ~]# pulp-admin repo list +----------------------------------------------------------------------+ Repositories +----------------------------------------------------------------------+ Id: f18 Display Name: f18 Description: None Content Unit Counts: Package Category: 12 Package Group: 151 Rpm: 33863 Yum Repo Metadata File: 1
2.2 released http://repos.fedorapeople.org/repos/pulp/pulp/stable/2.2/