traf_*_mods* hard-coded values

[Amanda Moran]

traf_*_mods* scripts have hard coded values for hdfs. Need a new way to look for these fields without hardcoded values.

Possible tool to help:

“jq” is the tool that my script uses to parse out the JSON stuff: http://stedolan.github.io/jq/manual/
(though it does require an extra “jq” package be installed)
$ curl -su admin:admin http://localhost:7180/api/v1/clusters/trafcluster/services | jq -r '.items[]|{(.type):.name}' {

[Steve Varnau (steve-varnau)]

Similar problem with hbase service name. Scripts query CM for the name, but find it by grepping for "hbase". There is no guarantee that "hbase" will be in the name.

[Marvin Anderson (marvin-anderson)]

Additional details:

This is specific to Cloudera distro which has names for the hadoop types and the names may not match the type:

$ curl -u admin:admin http://16.235.158.33:7180/api/v1/clusters/amethyst18/services
{
  "items" : [ {
    "name" : "zookeeper",
    "type" : "ZOOKEEPER",
    "clusterRef" : {
      "clusterName" : "cluster"
. . .
    "name" : "hbase",
    "type" : "HBASE",
    "clusterRef" : {
      "clusterName" : "cluster"

Ambari does not have names, only types. i.e. the service name for HBASE will always be "HBASE".

$ curl -u admin:admin http://sea-nodepool.hpl.hp.com:8080/api/v1/clusters/sea_nodepool/sevices
{
  "href" : "http://sea-nodepool.hpl.hp.com:8080/api/v1/clusters/sea_nodepool/services",
  "items" : [
    {
      "href" : "http://sea-nodepool.hpl.hp.com:8080/api/v1/clusters/sea_nodepool/services/HBASE",
      "ServiceInfo" : {
        "cluster_name" : "sea_nodepool",
        "service_name" : "HBASE"
      }
    },
    {
      "href" : "http://sea-nodepool.hpl.hp.com:8080/api/v1/clusters/sea_nodepool/services/HDFS",
      "ServiceInfo" : {
        "cluster_name" : "sea_nodepool",
        "service_name" : "HDFS"
      }