[scm-user] Re: Hive Metastore service health check

    I have found same issue in my cluster,we have 2 cluster , one is CM 5.0 and other is 5.1 version, This issue existing in 5.1 version CM. 

   can you provide the bug link?

在 2014年9月9日星期二UTC+8上午12时44分20秒，Gilad Wolff写道：

Hi Guy,

There is a known bug with the hive canary that may cause it to fail constantly - the client-configs we are using to connect to the hive metastore are partial. This may be the root cause of this failure. Do you have security enabled on the cluster? Did you change the hadoop.rpc.protection configuration option? A fix is going to be available very soon with the next release of cloudera manager (5.1.3) and hopefully it will solve the problem. In the meantime you can disable the hive metastore canary.

G.

On Mon, Sep 8, 2014 at 7:32 AM, Guy Doulberg <guy...@gmail.com>wrote:

Hi guys,

In our CM there is a health check called Hive Metastore Canary Health Test

This check is constantly failing,

A little investigation  shows that the check creates a DB (cloudera_manager_metastore_canary_test_db_hive2_hivemetastoreXXX) and a table in it (cm_test_table). 

The check then try to delete the table and destroy the DB, the check fails while trying to do that. 

As a result we end up with many databases in HIVE that contain one table. 

Furthermore, we are not able to delete the table becasue the hive complains the serde is null... we looked in hive metastore SDS table, and it creates the table with null SERDE, null outputformant and null inputformat

Did any of you encountered such a thing?

Thanks 

To unsubscribe from this group and stop receiving emails from it, send an email to scm-users+...@cloudera.org.

在 2014年9月9日星期二UTC+8上午12时44分20秒，Gilad Wolff写道：

Hi Guy,

There is a known bug with the hive canary that may cause it to fail constantly - the client-configs we are using to connect to the hive metastore are partial. This may be the root cause of this failure. Do you have security enabled on the cluster? Did you change the hadoop.rpc.protection configuration option? A fix is going to be available very soon with the next release of cloudera manager (5.1.3) and hopefully it will solve the problem. In the meantime you can disable the hive metastore canary.

G.

On Mon, Sep 8, 2014 at 7:32 AM, Guy Doulberg <guy...@gmail.com>wrote:

Hi guys,

In our CM there is a health check called Hive Metastore Canary Health Test

This check is constantly failing,

A little investigation  shows that the check creates a DB (cloudera_manager_metastore_canary_test_db_hive2_hivemetastoreXXX) and a table in it (cm_test_table). 

The check then try to delete the table and destroy the DB, the check fails while trying to do that. 

As a result we end up with many databases in HIVE that contain one table. 

Furthermore, we are not able to delete the table becasue the hive complains the serde is null... we looked in hive metastore SDS table, and it creates the table with null SERDE, null outputformant and null inputformat

Did any of you encountered such a thing?

Thanks 

The fix I mentioned above is in cm 5.1.2. What version of cm 5.1 are you running?