esbench throws KeyError:'nodes' due to cluster_stats not lining up.

[TimBanks]

INFO:esbench.data:downloading 'https://s3-us-west-1.amazonaws.com/esbench/appl_2005_aa.gz' to '/tmp/appl_2005_aa.gz'...
INFO:esbench.data:done downloading https://s3-us-west-1.amazonaws.com/esbench/appl_2005_aa.gz, time: 18.81s
INFO:esbench.bench:loaded 6 lines into index 'esbench_test', size: 120968 (0.00MB)
INFO:esbench.bench:beginning observation no: 1, 2015-06-23T17:00:23Z
INFO:esbench.bench:ran query 'match_description_facet_date_histogram' 100 times in 5.35s
INFO:esbench.bench:ran query 'match_description_sorted_abstract' 100 times in 4.67s
INFO:esbench.bench:ran query 'match_description' 100 times in 4.62s
INFO:esbench.bench:finished observation no: 1, id: 052cb9af, time: 14.635
ERROR:esbench.client:'nodes'
Traceback (most recent call last):
File "/root/tempdir/env/lib/python2.7/site-packages/esbench-0.2.3-py2.7.egg/esbench/client.py", line 174, in main
benchmark.run(batches)
File "/root/tempdir/env/lib/python2.7/site-packages/esbench-0.2.3-py2.7.egg/esbench/bench.py", line 370, in run
self.observe()
File "/root/tempdir/env/lib/python2.7/site-packages/esbench-0.2.3-py2.7.egg/esbench/bench.py", line 329, in observe
observation.record()
File "/root/tempdir/env/lib/python2.7/site-packages/esbench-0.2.3-py2.7.egg/esbench/bench.py", line 279, in record
'cluster': self._cluster_stats(),
File "/root/tempdir/env/lib/python2.7/site-packages/esbench-0.2.3-py2.7.egg/esbench/bench.py", line 237, in _cluster_stats
cluster_stats['node_count'] = len(cluster_stats['nodes'].keys())
KeyError: 'nodes'

[TimBanks]

Output of clusters_stats.nodes is looks like:

"nodes" : {
"count" : {
"total" : ,
"master_only" : ,
"data_only" : ,
"master_data" : ,
"client" :
},

[alanbbr]

I'm seeing this same issue as well. OS X 10.10.4, Elasticsearch 1.7.1 via Homebrew.

[mcortinas]

I'm seeing this same issue as well. Ubuntu wheezy/sid with Elasticsearch 1.5.2. via dpkg and Python 2.7.3

[alanbbr]

My fork has handlers to add support for newer versions of Elasticsearch, although not elegantly.

[alanbbr]

It looks like Tim has a pull request to handle the changed Elasticsearch API too.

[rituparnab]

I am trying to use esbench. I installed the dev version. I am using Centos 7 and 2 node ES cluster (version 1.7.1) and python 2.7.5. I see this Error. Is there a fix or workaround?

[badpianoman]

Just in case anyone else find this, this error is caused because the APIs have changed in the newer versions of ElasticSearch as alanbbr has pointed out. You can check /opt/elasticsearch/logs to see that ElasticSearch doesn't like the API call. You could remove bench.py line 237 and not get the metrics or do as alanbbr has done and try and update the code to handle the new APIs...