Changing zookeeper client port in Ambari is not reflected for all services

[#1]

Hi,

I am trying to deploy a HDP cluster via Ambari on a number of VM's that are each running a different version of zookeeper already.

As expected, I am encountering errors that the zookeeper client port (:2181) is already occupied when I try to deploy zookeeper via Ambari.

I tried changing the client port to an unoccupied port in /zoo.cfg in the *Customize Services* of Ambari deployment. Doing this I was able to successfully deploy Zookeeper, but the services that connect to zookeeper are still looking for it at :2181. An example of this is below:

"kafka-broker" : {
     "properties" : { 
	...
	"zookeeper.connect" : "%HOSTGROUP::host_group_1%:2181",
	...
	}
}

I would prefer not to manually change zookeepers client port on every single service that uses zookeeper. Is there an easier alternative?

[#2]

@John Martin

You can go to each service configuration in Ambari UI and change the port number.

For example for Kafka, see below screenshot

[#3]

@Kuldeep Kulkarni

Thanks for getting back to me!

I know that I can manually change the port number for each service, I was wondering if there was an alternative to doing that (I don't know much about Zookeeper).

The reason I ask is because I am working on an automated deployment of HDP, and manually changing the port (even if I did so within a Blueprint) gets in the way of that.

[#4]

@John Martin - If you are going to use blueprint then I believe mentioning port in the configuration is the only option.

[#5]

@Kuldeep Kulkarni each service configuration can be updated by using API. Take a look here. You can also make changes in the blueprint.

Configuration update involves the following steps:

• Identify the config type to update and note the latest version applied. When a config type is updated the whole property set needs to be updated. So copying the values from the latest and updating specific values or adding/removing properties as needed is the easiest option.
• Read the cluster resource and note the version of the type you want to update
• Read the config type with the tag and note the properties
• Edit the properties as needed and then update the config type
  • Config update requires creation of a new version (typically current timestamp is a good choice)
  • The new version of the config type must be added and applied to the cluster
• Restart affected services/components to have the config take effect

You can use APIs or a wrapper script (/var/lib/ambari-server/resources/scripts/configs.sh) to edit configurations.

Edit configuration using APIs (1.4.1/1.2.5)

Verified against releases 1.4.1/1.2.5

Starting 1.4.2/1.4.3 you will have to add -H option to the curl calls. E.g. -H "X-Requested-By: ambari"

1. Find the latest version of the config type that you need to update.

curl -u admin:admin -X GET http://AMBARI_SERVER_HOST:8080/api/v1/clusters/CLUSTER_NAME?fields=Clusters/desired_configs Sample OUTPUT { "href" : "http://AMBARI_SERVER_HOST:8080/api/v1/clusters/CLUSTER_NAME?fields=Clusters/desired_configs", "Clusters" : { "cluster_name" : "CLUSTER_NAME", "version" : "HDP-2.0.6", "desired_configs" : { ... "mapred-site" : { "user" : "admin", "tag" : "version1384716039631" } ... } } }

2. Read the config type with correct tag

curl -u admin:admin "http://AMBARI_SERVER_HOST:8080/api/v1/clusters/CLUSTER_NAME/configurations?type=mapred-site&tag=vers... Sample OUTPUT { "href" : "http://AMBARI_SERVER_HOST:8080/api/v1/clusters/CLUSTER_NAME/configurations?type=mapred-site&tag=vers..., "items" : [ { "href" : "http://AMBARI_SERVER_HOST:8080/api/v1/clusters/CLUSTER_NAME/configurations?type=mapred-site&tag=vers..., "tag" : "version1384716039631", "type" : "mapred-site", "Config" : { "cluster_name" : "CLUSTER_NAME" }, "properties" : { ... THESE ARE THE PROPERTY KEY-VALUE PAIRS ... } }] }

3a. Save a new version of the config and apply it (see 3b for doing it using one call)

curl --user admin:admin -i -X POST -d '{"type": "mapred-site", "tag": "version1384716041120", "properties" : {"mapreduce.admin.map.child.java.opts" : "-Djava.net.preferIPv4Stack=true -Dhadoop.metrics.log.level=WARN",...}}' http://AMBARI_SERVER_HOST:8080/api/v1/clusters/CLUSTER_NAME/configurations curl --user admin:admin -i -X PUT -d '{"Clusters":{"desired_config" : {"type": "mapred-site", "tag": "version1384716041120"}}}' http://AMBARI_SERVER_HOST:8080/api/v1/clusters/CLUSTER_NAME

3b. Save a new version of the config and apply it using one call

curl --user admin:admin -i -X PUT -d '{"Clusters":{"desired_config" : {"type": "mapred-site", "tag": "version1384716041120", "properties" : {...}}}}' http://AMBARI_SERVER_HOST:8080/api/v1/clusters/CLUSTER_NAME

4. Restart all components or services to have the config change take effect E.g. Stop and Start a service

curl --user admin:admin -i -X PUT -d '{"RequestInfo": {"context": "Stop HDFS"}, "ServiceInfo": {"state": "INSTALLED"}}' http://AMBARI_SERVER_HOST:8080/api/v1/clusters/CLUSTER_NAME/services/HDFS curl --user admin:admin -i -X PUT -d '{"RequestInfo": {"context": "Start HDFS"}, "ServiceInfo": {"state": "STARTED"}}' http://AMBARI_SERVER_HOST:8080/api/v1/clusters/CLUSTER_NAME/services/HDFS

[#6]

@Sunile Manjee - Thanks for the info, I'm aware that we can do this using API, as @John Martin wanted to do this while installing HDP cluster hence suggested to put in configs section of ambari blueprint.

[#7 · Accepted solution]

@John Martin

Please post it as an idea in HCC / file an enhancement request with Apache / Hortonworks.