From 6ac1cc485043765cf110f4e3b4aaa1769459afd4 Mon Sep 17 00:00:00 2001
From: Shivaram Venkataraman <shivaram@eecs.berkeley.edu>
Date: Fri, 16 Aug 2013 11:28:07 -0700
Subject: [PATCH] Add scripts to print master hostnames and copy progress

---
 check-copy-progess.sh |  5 +++++
 get-masters           |  2 ++
 get_masters.py        | 37 +++++++++++++++++++++++++++++++++++++
 3 files changed, 44 insertions(+)
 create mode 100644 check-copy-progess.sh
 create mode 100755 get-masters
 create mode 100644 get_masters.py

diff --git a/check-copy-progess.sh b/check-copy-progess.sh
new file mode 100644
index 0000000..16d23bc
--- /dev/null
+++ b/check-copy-progess.sh
@@ -0,0 +1,5 @@
+#!/bin/bash
+# Usage
+# check-copy-progress <private_key_path> <file-with-hostnames>
+
+pssh -i -x "-i $1" -l root -h $2 '~/ephemeral-hdfs/bin/hadoop fs -dus /'
diff --git a/get-masters b/get-masters
new file mode 100755
index 0000000..97460bd
--- /dev/null
+++ b/get-masters
@@ -0,0 +1,2 @@
+#!/bin/bash
+PYTHONPATH="./third_party/boto-2.4.1.zip/boto-2.4.1:$PYTHONPATH" python get_masters.py $@
diff --git a/get_masters.py b/get_masters.py
new file mode 100644
index 0000000..9e810aa
--- /dev/null
+++ b/get_masters.py
@@ -0,0 +1,37 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+
+from __future__ import print_function
+import json
+import urllib2
+import boto
+from optparse import OptionParser
+from boto import *
+
+def is_active(instance):
+  return (instance.state in ['pending', 'running', 'stopping', 'stopped'])
+
+def main():
+  parser = OptionParser(usage="get_masters [cluster_prefix]",
+      add_help_option=True)
+  (opts, args) = parser.parse_args()
+  if len(args) != 1:
+    get_cluster_masters()
+  else:
+    get_cluster_masters(args[0])
+
+def get_cluster_masters(prefix=""):
+  conn = connect_ec2()
+  res = conn.get_all_instances()
+  # master reservations have only one node
+  masters = [i for i in res if len(i.instances) == 1]
+  master_instances = [m.instances[0] for m in masters if is_active(m.instances[0])]
+
+  name_host = [(m.tags['cluster'], m.public_dns_name) for m in master_instances]
+
+  for (name, host) in name_host:
+    if prefix in name:
+      print(name + " " + host)
+
+if __name__ == "__main__":
+  main()