4 files changed, 74 insertions, 8 deletions
diff --git a/cloudinit/CloudConfig/cc_apt_update_upgrade.py b/cloudinit/CloudConfig/cc_apt_update_upgrade.py
index 495b8522..10707c77 100644
--- a/cloudinit/CloudConfig/cc_apt_update_upgrade.py
+++ b/cloudinit/CloudConfig/cc_apt_update_upgrade.py
@@ -27,10 +27,8 @@ def handle(name,cfg,cloud,log,args):
     upgrade = util.get_cfg_option_bool(cfg, 'apt_upgrade', False)
 
     release = get_release()
-    if cfg.has_key("apt_mirror"):
-        mirror = cfg["apt_mirror"]
-    else:
-        mirror = cloud.get_mirror()
+
+    mirror = find_apt_mirror(cloud, cfg)
 
     if not util.get_cfg_option_bool(cfg, \
         'apt_preserve_sources_list', False):
@@ -162,3 +160,46 @@ def add_sources(srclist, searchList={ }):
     return(elst)
 
 
+def find_apt_mirror(cloud, cfg):
+    """ find an apt_mirror given the cloud and cfg provided """
+
+    # TODO: distro and defaults should be configurable
+    distro = "ubuntu"
+    defaults = {
+        'ubuntu': "http://archive.ubuntu.com/ubuntu",
+        'debian': "http://archive.debian.org/debian",
+    }
+    mirror = None
+
+    if cfg.has_key("apt_mirror"):
+        mirror = [cfg["apt_mirror"],]
+    elif cfg.has_key("apt_mirror_search"):
+        mirror = util.search_for_mirror(cfg['apt_mirror_search'])
+    else:
+        if cloud:
+            mirror = cloud.get_mirror()
+
+        mydom = ""
+
+        doms = []
+
+        if not mirror and cloud:
+            # if we have a fqdn, then search its domain portion first
+            ( hostname, fqdn ) = util.get_hostname_fqdn(cfg, cloud)
+            mydom = ".".join(fqdn.split(".")[1:])
+            if mydom:
+                doms.append(".%s" % mydom)
+
+        doms.extend((".localdomain", "",))
+
+        mirror_list = []
+        for post in doms:
+            mirror_list.append("http://ubuntu-mirror%s/ubuntu" % post)
+
+        print "searching %s" % mirror_list
+        mirror = util.search_for_mirror(mirror_list)
+
+    if not mirror:
+        mirror = defaults[distro]
+
+    return mirror
diff --git a/cloudinit/DataSource.py b/cloudinit/DataSource.py
index 597760fa..06061a2e 100644
--- a/cloudinit/DataSource.py
+++ b/cloudinit/DataSource.py
@@ -94,7 +94,7 @@ class DataSource:
         return('en_US.UTF-8')
 
     def get_local_mirror(self):
-        return('http://archive.ubuntu.com/ubuntu/')
+        return None
 
     def get_instance_id(self):
         if 'instance-id' not in self.metadata:
diff --git a/cloudinit/DataSourceEc2.py b/cloudinit/DataSourceEc2.py
index 0d8f358f..908a29cd 100644
--- a/cloudinit/DataSourceEc2.py
+++ b/cloudinit/DataSourceEc2.py
@@ -69,7 +69,7 @@ class DataSourceEc2(DataSource.DataSource):
         if availability_zone == None:
             availability_zone = self.get_availability_zone()
 
-        fallback = 'http://archive.ubuntu.com/ubuntu/'
+        fallback = None
 
         if self.is_vpc():
             return fallback
diff --git a/cloudinit/util.py b/cloudinit/util.py
index 0c457128..dc461f6c 100644
--- a/cloudinit/util.py
+++ b/cloudinit/util.py
@@ -24,9 +24,11 @@ from Cheetah.Template import Template
 import urllib2
 import urllib
 import logging
+import re
+import socket
 import time
 import traceback
-import re
+import urlparse
 
 try:
     import selinux
@@ -97,7 +99,7 @@ def get_cfg_by_path(yobj,keyp,default=None):
         cur = cur[tok]
     return(cur)
 
-# merge values from src into cand.
+# merge values from cand into source
 # if src has a key, cand will not override
 def mergedict(src,cand):
     if isinstance(src,dict) and isinstance(cand,dict):
@@ -498,3 +500,26 @@ def get_fqdn_from_hosts(hostname, filename="/etc/hosts"):
             pass
 
     return fqdn
+
+def is_resolvable(name):
+    """ determine if a url is resolvable, return a boolean """
+    try:
+        socket.getaddrinfo(name, None)
+        return True
+    except socket.gaierror as e:
+        return False
+
+def is_resolvable_url(url):
+    """ determine if this url is resolvable (existing or ip) """
+    return(is_resolvable(urlparse.urlparse(url).hostname))
+
+def search_for_mirror(candidates):
+    """ Search through a list of mirror urls for one that works """
+    for cand in candidates:
+        try:
+            if is_resolvable_url(cand):
+                return cand
+        except Exception as e:
+            raise
+
+    return None