Tez session hasn't been created yet. Opening session

[#1]

Hello,

I'm trying to do an export table to a file using beeline.

jdbc:hive2://servername> INSERT OVERWRITE LOCAL DIRECTORY '/tmp/custorder' select * from custorder;

But the operation is stop with : Tez session hasn't been created yet. Opening session for a very long time.

Actually, nothing happend even after one hour.

Can someone help me on this issue?

Thank very much.

Best regards,

Daniela

[#2]

If YARN cluster dont have capacity to launch AM then Hive CLI waits indefinitely to launch an AM. Tez have its own optimization step where it pre-allocate resources saving the user the overhead of launching a AM everytime a new query launch from same session. But when it lanches using mr it dont have any condition called preallocate and works successfully.

[#3]

HI Rodgers,

Please find bellow the result of the command :

[spark@servername ~]$ yarn application -list 16/07/08 07:04:34 INFO impl.TimelineClientImpl: Timeline service address: http://servername:8188/ws/v1/timeline/ 16/07/08 07:04:34 INFO client.RMProxy: Connecting to ResourceManager at servername/ip:8050 Total number of applications (application-types: [] and states: [SUBMITTED, ACCEPTED, RUNNING]):5 Application-Id Application-Name Application-Type User Queue State Final-State Progress Tracking-URL application_1467377616814_0001 SparkSQL::servername SPARK spark default RUNNING UNDEFINED 10% http://servername:4040 application_1467377616814_0007 HIVE-701d4281-08f4-4bc4-aab3-724c931fc606 TEZ spark default ACCEPTED UNDEFINED 0% N/A application_1467377616814_0005 HIVE-55924703-c794-413f-8df0-fbb464b48885 TEZ spark default ACCEPTED UNDEFINED 0% N/A application_1467377616814_0010 HIVE-9ef8ceda-b31f-4311-8b4a-288b5f476e93 TEZ hive2 default ACCEPTED UNDEFINED 0% N/A application_1467702596252_0002 HIVE-a549f4eb-ab4c-4f57-a180-3e573160b9d9 TEZ hive2 default ACCEPTED UNDEFINED 0% N/A

thank you very much.

Best regards,

daniela

,

Hi Rodgers,

please find bellow the result of the yarn application -list

[spark@servername ~]$ yarn application -list 16/07/08 07:04:34 INFO impl.TimelineClientImpl: Timeline service address: http://servername:8188/ws/v1/timeline/ 16/07/08 07:04:34 INFO client.RMProxy: Connecting to ResourceManager at servername/10.208.159.170:8050 Total number of applications (application-types: [] and states: [SUBMITTED, ACCEPTED, RUNNING]):5 Application-Id Application-Name Application-Type User Queue State Final-State Progress Tracking-URL application_1467377616814_0001 SparkSQL::servername SPARK spark default RUNNING UNDEFINED 10% http://servername:4040 application_1467377616814_0007 HIVE-701d4281-08f4-4bc4-aab3-724c931fc606 TEZ spark default ACCEPTED UNDEFINED 0% N/A application_1467377616814_0005 HIVE-55924703-c794-413f-8df0-fbb464b48885 TEZ spark default ACCEPTED UNDEFINED 0% N/A application_1467377616814_0010 HIVE-9ef8ceda-b31f-4311-8b4a-288b5f476e93 TEZ hive2 default ACCEPTED UNDEFINED 0% N/A application_1467702596252_0002 HIVE-a549f4eb-ab4c-4f57-a180-3e573160b9d9 TEZ hive2 default ACCEPTED UNDEFINED 0% N/A

Thank you,

Daniela

[#4]

If you could kill all of those processes (if your cluster will allow) and try running again. It looks like a lot of Tez resources are busy and thus could be why you are experiencing the hanging. If not you may need to look into YARN Queue Manager

[#5]

I have the same issue with you.

I solved it by modifying the following..

reduce Tez config : tez.am.resource.memory.mb =256

default 2048 .

and Increase Yarn ： Memory allocated for all YARN containers on a node 。

I guess on hive cli use Tez need {tez.am.resource.memory.mb} memory . when {Memory allocated for all YARN containers on a node} no have enough memory . Then Application-id is hang.