Implement EBS volume mounting

This can either be invoked by instrumenting the user-data with a mime
part with content-type 'text/x-ebs-mount-description' with a body like
so:

device=/dev/sde:/var/lib/mysql,/etc/alfresco
device=/dev/sdf:/other/things

or by using the appliance config XML format like so:

<appliance>
  <storage device="/dev/sde">
	<path>/var/lib/mysql</path>
    <path>/etc/alfresco</path>
	</storage>
  <storage device="/dev/sdf">
    <path>/other/things</path>
  </appliance>
</appliance>

In either case, if the volume does not yet have a filesystem, one will be created.

For each path that is to live on the volume, a directory is created, and
populated with the data currently in the target directory (e.g.
/var/lib/mysql is copied to ${ebs_volume_path}/_var_lib_mysql). Once
this is done, the directories are bind-mounted to the relevant paths.
If the directories in question already exist, they will just be bind-mounted.

4 files changed, 108 insertions, 4 deletions

diff --git a/ec2-init-appliance-ebs-volume-mount.sh b/ec2-init-appliance-ebs-volume-mount.sh
new file mode 100755
index 00000000..7106b353
--- /dev/null
+++ b/ec2-init-appliance-ebs-volume-mount.sh
@@ -0,0 +1,48 @@
+#!/bin/sh
+
+if [ -n "$EBSMOUNT_DEBUG" ]
+then
+    do="echo"
+    mktemp_args="-u"
+else
+    do=""
+    mktemp_args=""
+fi
+
+if [ "$#" -lt 2 ]
+then
+	echo "Usage: $0 <EBS device> <path> [<path> [<path>...]]"
+	exit 1
+fi
+
+ebs_volume_device="$1"
+shift
+
+canonicalise_dir() {
+    dirname="$1"
+    echo "${dirname}" | sed -e 's/[^a-zA-Z0-9]/_/g'
+}
+
+# The blkid call will detect whether there's already a filesystem on the EBS volume
+if [ -n "$(blkid -p -o udev "${ebs_volume_device}")" ]
+then
+    $do mkfs.ext3 "${ebs_volume_device}"
+fi
+
+tmpdir="$(mktemp -d $mktemp_args --tmpdir=/var/run/ec2-init)"
+$do mount ${ebs_volume_device} ${tmpdir}
+
+for dir in "$@"
+do
+    ebsdir="${tmpdir}/$(canonicalise_dir "${dir}")"
+    if [ ! -d "${ebsdir}" ]
+    then
+        # We bootstrap the storage with the existing data
+        $do mkdir "${ebsdir}"
+        $do cp -a ${dir} "${ebsdir}"
+        $do chown --reference "${dir}" "${ebsdir}"
+        $do chmod --reference "${dir}" "${ebsdir}"
+    fi
+    # Finally, we mount it on top of the old directory.
+    $do mount --bind "${ebsdir}" "${dir}"
+done
diff --git a/ec2-run-user-data.py b/ec2-run-user-data.py
index 67ecf219..d7e8e632 100755
--- a/ec2-run-user-data.py
+++ b/ec2-run-user-data.py
@@ -61,7 +61,7 @@ def handle_appliance_config(payload):
 
 @handler('text/x-ebs-mount-description')
 def handle_ebs_mount_description(payload):
-    (volume_description, path) = payload.split(':')
+    (volume_description, paths) = payload.split(':')
     (identifier_type, identifier) = volume_description.split('=')
 
     if identifier_type == 'device':
@@ -73,6 +73,17 @@ def handle_ebs_mount_description(payload):
     else:
         return
 
+    mount_ebs_volume(device, paths.split(','))
+
+def mount_ebs_volume(device, paths):
+    if os.path.exists('ec2-init-appliance-ebs-volume-mount.sh'):
+        helper = './ec2-init-appliance-ebs-volume-mount.sh'
+    else:
+        helper = '/usr/share/ec2-init/ec2-init-appliance-ebs-volume-mount.sh'
+    helper = subprocess.Popen([helper, device] + paths, stdout=subprocess.PIPE)
+    stdout, stderr = helper.communicate()
+    return stdout
+
 @handler('text/x-shellscript')
 def handle_shell_script(payload):
     (fd, path) = tempfile.mkstemp()
@@ -129,6 +140,16 @@ class ApplianceConfig(object):
                     # An empty script?
                     continue
                 content_type_handlers['text/x-shellscript'](script)
+            elif node.tagName == 'storage':
+                paths = []
+                device = node.getAttribute('device')
+                for subnode in node.childNodes:
+                    if subnode.tagName == 'path':
+                        for subsubnode in subnode.childNodes:
+                            if subsubnode.nodeType == root.TEXT_NODE:
+                                paths += [subsubnode.nodeValue.strip()]
+                                break
+                mount_ebs_volume(device, paths)
 
 def main():
     ec2 = ec2init.EC2Init()
diff --git a/setup.py b/setup.py
index fa855bd6..dab57103 100755
--- a/setup.py
+++ b/setup.py
@@ -37,5 +37,7 @@ setup(name='EC2-init',
                'ec2-wait-for-meta-data-service.py'],
       data_files=[('/etc/ec2-init', ['ec2-config.cfg']),
                   ('/etc/ec2-init/templates', glob('templates/*')),
-                  ('/etc/init.d', ['ec2-init'])],
+                  ('/etc/init.d', ['ec2-init']),
+                  ('/usr/share/ec2-init', ['ec2-init-appliance-ebs-volume-mount.sh']),
+                  ],
       )
diff --git a/tests.py b/tests.py
index 33c45ec7..84103745 100644
--- a/tests.py
+++ b/tests.py
@@ -18,9 +18,42 @@
 #    along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 
+import re
+import os
 import unittest
 
-class RunUserDataApplianceConfigScript(unittest.TestCase):
+class RunUserDataApplianceTestCase(unittest.TestCase):
+    def handle_xml(self, xml):
+        msg = self.ec2_run_user_data.parse_user_data(xml)
+        self.ec2_run_user_data.handle_part(msg)
+
+class RunUserDataApplianceConfigEBS(RunUserDataApplianceTestCase):
+    def setUp(self):
+        self.ec2_run_user_data = __import__('ec2-run-user-data')
+        reload(self.ec2_run_user_data) 
+        self.real_mount_ebs_volume = self.ec2_run_user_data.mount_ebs_volume
+        self.ec2_run_user_data.mount_ebs_volume  = self.fake_mount_ebs_volume
+    
+    def fake_mount_ebs_volume(self, device, paths):
+        self.assertEqual(device, '/dev/sdc')
+        self.assertEqual(paths, ['/etc/alfresco', '/var/lib/mysql'])
+
+    def testApplianceConfigEBS(self):
+        os.environ['EBSMOUNT_DEBUG'] = 'yes, please'
+        xml = '<appliance><storage device="/dev/sdc"><path>/etc/alfresco</path><path>/var/lib/mysql</path></storage></appliance>'
+        self.handle_xml(xml)
+
+    def testMountEBSVolume(self):
+        output = self.real_mount_ebs_volume('/dev/sdh', ['/foo', '/bar'])
+        lines = output.strip().split('\n')
+        self.assertEqual(len(lines), 11)
+        match = re.match('mount /dev/sdh (/var/run/ec2-init/tmp.[a-zA-Z0-9]+)', lines[0])
+        self.assertNotEqual(match, None)
+        tmpdir = match.group(1)
+        for (i, s) in zip(range(10), ['mkdir %s/_foo', 'cp -a /foo %s/_foo', 'chown --reference /foo %s/_foo', 'chmod --reference /foo %s/_foo', 'mount --bind %s/_foo /foo', 'mkdir %s/_bar', 'cp -a /bar %s/_bar', 'chown --reference /bar %s/_bar', 'chmod --reference /bar %s/_bar', 'mount --bind %s/_bar /bar']):
+            self.assertEqual(s % tmpdir, lines[i+1])
+
+class RunUserDataApplianceConfigScript(RunUserDataApplianceTestCase):
     def setUp(self):
         self.ec2_run_user_data = __import__('ec2-run-user-data')
         self.fake_handle_shell_script_counter = 0
@@ -62,7 +95,7 @@ echo hey'''
         self.handle_xml(xml)
         self.assertEqual(self.fake_handle_shell_script_counter, 1) 
 
-class RunUserDataApplianceConfigPackageHandling(unittest.TestCase):
+class RunUserDataApplianceConfigPackageHandling(RunUserDataApplianceTestCase):
     def setUp(self):
         self.fake_install_remove_package_counter = 0