Ceph » devops

However, if I add "::1 <hostname>" to my /etc/hosts, and issue getaddrinfo, I get
back v6 entries that look like:

(10, 1, 6, '', ('::1', 0, 0, 0)), (10, 2, 17, '', ('::1', 0, 0, 0)), (10, 3, 0, '', ('::1', 0, 0, 0))

so I don't know what's causing the alternate format response on the user's system.

getaddrinfo documents:

sockaddr is a tuple describing a socket address, whose format depends on the returned family (a (address, port) 2-tuple for AF_INET, a (address, port, flow info, scope id) 4-tuple for AF_INET6)

so it seems this user's return format is not as expected; I've no idea how or why.
I note this exists: https://groups.google.com/forum/#!topic/web2py/t_11PX-hYt0 so others have had similar problems on Gentoo.

I'm the user that orignally reported this issue, and after talking with the Gentoo crowd about the issue, we've discovered that this is a known bug in python when ipv6 is disabled.

You can read more here: http://bugs.python.org/issue16208

ceph-deploy can do one of many things (obviously, this is not an exaustive list, and I'm not in a position to tell the ceph-deploy team what to do, these are just suggestions...)

1) Consider this a failure condition in the environment and refuse to address the issue (perfectly valid response).
2) Work around this error by parsing the returned hexidecimal gibberish and handling it that way.
3) Warn the user that their python installation has IPv6 disabled and exit, or skip over ipv6 addresses.

To further elaborate, when I recompiled python so that IPv6 support was properly enabled, the problems that I was having with running ceph-deploy were no longer reproducable.