Re: What performance to expect from Cloudera VM ?

AcharkiMed · ‎08-08-2019

Hi @Zane- This is an OOM (Out If Memory) error, it simply means that this query needs more memory to be completed; it usually occurs when the cluster is in charge. The request has exceeded the existing memory. The solution is to add more memory to your nodes or to add more nodes. Otherwise, try to optimize your queries, and you can make sure that your impala setting is optimal too. https://www.cloudera.com/documentation/enterprise/latest/topics/impala_performance.html Good Luck.

AcharkiMed · ‎07-04-2019

hi @punshi Try to read this to get more info about HDFS caching in Impala: https://www.cloudera.com/documentation/enterprise/5-16-x/topics/impala_perf_hdfs_caching.html

AcharkiMed · ‎07-04-2019

Have you made any manual changes to the metastore user or database permissions? because it looks like the DBS table is not found in metastore DB! Remark: check if the CM point to the correct metastore DB with the pertinent user.

AcharkiMed · ‎07-04-2019

Hi @andreas It looks like you have a connectivity issue with hive metastore.. try to put off the firewall and test. Else, please share with us those log files: /var/log/hive/hadoop-cmf-hive-HIVEMETASTORE-XXXX.log.out /var/log/impalad/impalad.hdm1.emd.impala.log.INFO.YYYYYYY Good luck

AcharkiMed · ‎07-04-2019

Hi @punshi Yes you can change it by editing this parameter: Maximum Memory Used for Caching dfs.datanode.max.locked.memory But you need to know that data caching has moved from HDFS to memory (RAM), so you can not increase it considerably!

AcharkiMed · ‎07-03-2019

Hi @punshi How much cache space you have configured? please try this hdfs command to display the details of cache configured and used. hdfs dfsadmin -report

AcharkiMed · ‎07-03-2019

HI @punshi Do you use the QuickStart VM for CDH 5.13? Indeed, the VM is just for testing !! To unlock the real power of the impala (CDH), you should have a cluster so that you can benefit from the senergy of several nodes. Anyway, you can improve your query time by: 1- Set PARQUET_FILE_SIZE = 256MB instead of 512MB. 2- Try to minimize the number of partitions (in fact, I think in your case, the year is sufficient). NB: I think that in a cluster that has more than 10 nodes, this request will not exceed 2 to 4 seconds. Good luck.

AcharkiMed · ‎05-06-2019

Hi @anis447 Can you bring us this queries results in SQL server and Impala: Select avg(tagno) from tag; Select avg(tagno) from has_tag; Select count(*) from tag where tagno is null; Select count(*) from has_tag where tagno is null; Also try to add this on Impala query, and let us know if there is any change: ... Inner join has_tags hit on (s.tagno = hit.tagno and s.categorycode = hit.categorycode) ... Good luck.

AcharkiMed · ‎04-28-2019

Hi @bridgor It looks like you have a firewall issue between your nodes.. try to check iptables. Please share with us the impalad log files. Good luck.

AcharkiMed · ‎04-25-2019

Hi @Ravikiran Normaly, most softwares on top of HDFS reads only one replication in one time, so the memory will be charged with the real size of the data, not size*replication_factor. I think YARN service that concerned by the resource management take this role, and always the most free and fast datanode will be chosed to pull his data to memory to be treated by impalad. Good luck.

AcharkiMed · ‎04-22-2019

Hi, the unix_timestamp function return a number of seconds but it seems that ipid.bk_eff_strt_dt column is inserted with the milliseconds number!? About your query try this: SELECT COUNT(*) FROM ipid WHERE unix_timestamp('20190124',"yyyyMMdd")*1000 BETWEEN ipid.BK_EFF_STRT_DT AND ipid.BK_EFF_END_DT; Good luck.

AcharkiMed · ‎04-21-2019

Hi, The "thread_network_send_wait_time" is the sum of the time spent waiting to send data over the network by all threads of the query, so it depends mainly to your nodes and/or client network, try to use "profile" to see who is the slowest node network that made the problem or maybe it's just a load issue..

AcharkiMed · ‎04-21-2019

Hi, Optimization is not an easy task, please try to read this document and explore the reel Impala query optimization methods. https://www.cloudera.com/documentation/enterprise/5-15-x/topics/impala_performance.html Good luck.

AcharkiMed · ‎02-09-2019

Hi @Tim Armstrong While IMPALA-1618 steel open and unresolved, I confirmed that this "workaround" is safe and efficient (I'm using it on a large scope and during more than 9 months) so that this is the only solution I find to solve or -get around- this big problem. Hope that the main problem will be fixed ASAP. Thanks for the remark.

AcharkiMed · ‎02-09-2019

Hi @Bishnup If you still have the same problem please try to share with us your URL string.

AcharkiMed · ‎02-09-2019

Hi @zeni86cit try to change the file content by this SELECT CONCAT('invalidate metadata ', trim(table_name), '; refresh ', trim(table_name) ,';') from my_table; or SELECT CONCAT("invalidate metadata ", trim(table_name), "; refresh ", trim(table_name) ,";") from my_table; Good luck.

AcharkiMed · ‎02-02-2019

Hi @kampith The first step of troubleshooting a cloudera service is to look at it's log file, so share it with us please. the log files paths : scm agent: /var/log/cloudera-scm-server/cloudera-scm-agent.log scm-server /var/log/cloudera-scm-server/cloudera-scm-server.log impala deamon: /var/log/impalad Good luck.

AcharkiMed · ‎01-30-2019

Hi @Rr, Please give us more details, errors messages or screenshots so we can help you.

AcharkiMed · ‎10-01-2018

Hi, Please try to change all these 3 params: TSaslTransportBufSize=4000; RowsFetchedPerBlock=60536; SSP_BATCH_SIZE=60536;

AcharkiMed · ‎09-29-2018

Hi @Bishnup ConfiguringServer-SideProperties When connecting to a server that is running Impala 2.0 or later, you can use the driver to apply configuration properties to the server by setting the properties in the connection URL. https://www.cloudera.com/documentation/other/connectors/impala-jdbc/latest/Cloudera-JDBC-Driver-for-Impala-Install-Guide.pdf Good luck.

AcharkiMed · ‎07-26-2018

Hi @lonetiger You can do it by two kind of scripts: 1- show all tables: SHOW TABLES; get the list and run this query on all tables: DESCRIBE FORMATTED tableX; Then you can get the results and extract the owner, if it's the desired one, drop the table. 2- Connect with your hive metastore DB and get the table list of the owner you want SELECT "TBL_NAME" FROM "TBLS" WHERE "OWNER" = 'ownerX'; Then drop them. Good luck.

AcharkiMed · ‎07-25-2018

Hi @Tomas79 Check the value of your params SSL= UseSASL= in odbc.ini

AcharkiMed · ‎07-25-2018

Hi @lonetiger What do you mean by table owner ? If you use apache sentry just login to impala-shell by the user concerned (owner) and execute show tables; and you'll see just this user tables Good luck.

AcharkiMed · ‎07-25-2018

@saranvisa Thank you, I think your idea is very good to apply in several use cases to define the modified hive/impala tables. . But I thing there is a misunderstanding, because I mean by the 56 tables the metastore (postgres) database tables! And the backup approach that I used based on 2 steps: 1- Backup the updated HDFS tables by DistCp (done). 2- Backup the metastore (postgres) datastores (current question ).

AcharkiMed · ‎07-25-2018

Hi, Thanks @saranvisa to you response, Yes, but the question is how can I define the delta/impacted tables backup within about 56 tables in the metastore database. Thanks again.

AcharkiMed · ‎07-25-2018

Hi all, I use DistCp every X minutes to transfert the HDFS data to a hot bachup cluster, should I replicate the whole hive metastore database (manualy or using DB HA..) to accomplish the backup/restore ? or I need just to import/export only some specefic hive metastore tables ? Thanks in advance.

AcharkiMed · ‎07-25-2018

I'm not using the BDR. I want to make a hot backup cluster, so I'll test to replicate the whole metastore database every X minutes and see the results. Thanks for your replies.

AcharkiMed · ‎07-25-2018

Thanks @alexmc6 for your reply, The note in the documentation is clear: "you must run the Impala INVALIDATE METADATA statement on the destination cluster to prevent queries from failing". I have some tables with a big partition number, and I use it in real-time cases, so there is no time to execute the first query and wait a minutes ..

AcharkiMed · ‎07-20-2018

HI @Derek Try this workaround: 1- Create new table with the original table's data. 2- Drop all data from old table (using delete). 3- Insert data from new table into old table. 4- Drop new table . Good luck.

AcharkiMed · ‎07-18-2018

Hi, My question is, after using DistCp to transfert all the HDFS data to a second cluster, can I replicate the whole hive metastore database (manualy or using DB HA..) to accomplish the backup/restore ?or I should import/export only some specefic hive metastore tables ? Thanks.

Online	Offline
Last Visited	‎04-06-2022 03:13 PM

Date Registered	‎07-17-2017 07:15 AM
Last Visited	‎04-06-2022 03:13 PM
Total Messages Posted	143
Total Kudos Received	16

Re: What performance to expect from Cloudera VM ?

Re: Impala date

Re: Error 1107

Re: Cannot connect to Impala via ODBC

Re: Getting improper "Unexpected character" using ...

Re: ExecQueryFInstances rpc query_id=e74ef8d9b9215...

Re: Unable to cache full data in cache pool ?

Re: Error starting Impala

Re: Error starting Impala

Re: Unable to cache full data in cache pool ?

Re: Unable to cache full data in cache pool ?

Re: What performance to expect from Cloudera VM ?

Re: Same query, same data, different results betwe...

Re: Impala daemons constantly connecting/disconnec...

Re: Impala - Table block loading in memory

Re: Impala date

Re: How to reduce Impala thread_network_send_wait_...

Re: Increase IMPALA query performance/optimization...

Re: Impala ODBC/JDBC bad performance - rows fetch ...

Re: Impala ODBC/JDBC bad performance - rows fetch ...

Re: concat not working in beeline

Re: Troubleshooting Impala health test troubleshoo...

Re: Error 1107

Re: Impala ODBC/JDBC bad performance - rows fetch ...

Re: Impala ODBC/JDBC bad performance - rows fetch ...

Re: How to show tables based on the table owner?

Re: Cannot connect to Impala via ODBC

Re: How to show tables based on the table owner?

Re: Manual Hive replication - metastore database!

Re: Manual Hive replication - metastore database!

Manual Hive replication - metastore database!

Re: Hive metastore database (or tables) manual rep...

Re: BDR - problem of using invalidate metadata in ...

Re: How to rename kudu table name on impala versio...

Hive metastore database (or tables) manual replica...