Can not modify Hive view in Ambari

[#1]

In a freshly installed Ambari server most views did not point to a cluster name (the cluster name was empty). This was strange itself, but even more disappointing was the fact that also I was not able to modify Hive view. The pop came with the message:

Can't save settings. Server error

which was not very informative...

The attempts to find any relevant information in Ambari server logs also were unsuccessful.

So where should I look further?

[#2]

@Dmitry Otblesk

Which version of ambari server is it ?

- From Ambari 2.4 onwards we have separate logs files for different views. So if you are not finding any error message in the ambari-server.log and if you are using Ambari 2.4 or above then better to check the "/var/log/ambari-server/hive-view/hive-view.log" log file.

- Also can you please attach the error screenshot. And the config changes that you are planning to make.

[#3]

I am using Ambari 2.4, the one currently available for download.

Log file "/var/log/ambari-server/hive-view/hive-view.log is empty.

The image below shows that no cluster was assigned to the view (this view was created automatically). This was true for all views, not only Hive view.

However only Hive view did not allow me to assign the cluster. Any attempt to modify the view throws the pop up shown below

[#4]

@jss

Also it was not possible to delete this view

[#5]

@Dmitry Otblesk

try the Ambari Views API to update properties, here's an example

curl --user admin:admin -i -H 'X-Requested-By: ambari' -X PUT http://localhost:8080/api/v1/views/FILES/versions/1.0.0/instances/FILES_NEW_INSTANCE \
--data '{
  "ViewInstanceInfo" : {
      "description" : "Files API",
      "label" : "Files View",
      "properties" : {
      "webhdfs.client.failover.proxy.provider" : "org.apache.hadoop.hdfs.server.namenode.ha.ConfiguredFailoverProxyProvider",
      "webhdfs.ha.namenode.http-address.nn1" : "u1201.ambari.apache.org:50070",
      "webhdfs.ha.namenode.http-address.nn2" : "u1201.ambari.apache.org:50070",
      "webhdfs.ha.namenode.https-address.nn1" : "u1201.ambari.apache.org:50470",
      "webhdfs.ha.namenode.https-address.nn2" : "u1202.ambari.apache.org:50470",
      "webhdfs.ha.namenode.rpc-address.nn1" : "u1201.ambari.apache.org:8020",
      "webhdfs.ha.namenode.rpc-address.nn2" : "u1202.ambari.apache.org:8020",
      "webhdfs.ha.namenodes.list" : "nn1,nn2",
      "webhdfs.nameservices" : "hacluster",
      "webhdfs.url" : "webhdfs://hacluster"
      }
    }
}'

I have more examples published here https://github.com/dbist/ambari-chef/blob/master/notes

to get a specific instance of a view and delete it

# get specific instance of FILES view

curl --user admin:admin -i -H 'X-Requested-By: ambari' -X GET http://localhost:8080/api/v1/views/FILES/versions/1.0.0/instances/FILES_NEW_INSTANCE

# delete an instance of FILES view

curl --user admin:admin -i -H 'X-Requested-By: ambari' -X DELETE http://localhost:8080/api/v1/views/FILES/versions/1.0.0/instances/FILES_NEW_INSTANCE

[#6 · Accepted solution]

@Artem Ervits

Thank you very much for your help.

I actually already found a much easier workaround, which was to create a new Hive view and use it instead of a broken one.

But at this point I am still wondering why these things are happening? I.e.

- why automatically created view was not pointing to any cluster?

- why it was not possible to delete it or modify it via Ambari?

- why all this was happening "silently", meaning that I was not able to find any error messages in relevant logs?

[#7]

@Dmitry Otblesk

Did you check both the logs ?

1. "/var/log/ambari-server/hive-view/hive-view.log" (from ambari 2.4 onwards)

2. "/var/log/ambari-server/ambari-server.log"

[#8]

Yes I did. Both were empty at the time when problem occured