Re: Hive migration to a new cluster & Hive metasto...

PranayV · ‎10-11-2018

Can u check for any hung process? ps -ef | grep registry ps -ef | grep nifi You can also create the empty pid file with right permissions for the service account.

PranayV · ‎04-10-2018

Not really. you also need AM containers to spawn new AM for your jobs. You can configure % of memory reserved for AM containers in Yarn capacity scheduler view.

PranayV · ‎02-23-2018

@Alex PQ it is possible that the job had failed before log could be created. Navigate to Resource Manager UI and look for the failed job. Once you click on your application ID, it will show the status of log aggregation and possible failure reason. you can also open AM log from here.

PranayV · ‎02-23-2018

yes, you can. please look at yarn capacity scheduler settings. Please read here - https://docs.hortonworks.com/HDPDocuments/HDP2/HDP-2.6.4/bk_yarn-resource-management/content/setting_user_limits.html

PranayV · ‎02-23-2018

By Default CPU is shared unless you enable CPU based scheduling. So you will be able to run (total yarn memory/container size) on your cluster.

PranayV · ‎02-23-2018

@Raj B If you have both the clusters up and running, you can export tables from one cluster to another using Hive import export commands if the databases are not very large. https://cwiki.apache.org/confluence/display/Hive/LanguageManual+ImportExport

PranayV · ‎02-23-2018

@kotesh banoth For HDFS disk space alter HDFS configurations make sure all new data disks are provided under datanode directories. For computation You will have to alter yarn configuration - yarn.scheduler.maximum-allocation-mb yarn.scheduler.maximum-allocation-vcores yarn.nodemanager.resource.memory-mb

PranayV · ‎02-23-2018

@Daniel Müller HDFS will replicate once it scans an under-replicated block. alternatively running hdfs balancer may also trigger the same.

PranayV · ‎02-23-2018

@kishore sanchina You have to setup your browser for kerberos. You can look at here for steps - https://docs.hortonworks.com/HDPDocuments/HDP2/HDP-2.6.0/bk_security/content/enabling_browser_access_spnego_web_ui.html

PranayV · ‎02-23-2018

@Giuseppe D'Agostino All HDP packages primarily gets installed in /usr/hdp location. You can create a mount pointing to disk named /usr/hdp.

PranayV · ‎01-16-2018

@Chen Ran Open NameNode UI and go to "Datanode Volume Failures" tab. Let us know if you see any volume failures.

PranayV · ‎01-16-2018

@Nicolas Tobias Please look at hive.fetch.task.conversion property. This property gathers simple queries data without spinning up a job. Setting this to none should trigger data querying via job.

PranayV · ‎01-16-2018

answer depends on what kind of job. For e.g. a MR job, you check the running task log in Yarn Resource Manager. Depending upon what is being processed, you can check when the last activity or what activity is currently being carried out and for how long. for e.g. you can have fetcher class where reducer tasks is trying to fetch data from all map tasks. this could be time consuming if there is large amount of data, or intermediate compression is not enabled.

PranayV · ‎01-16-2018

Are you able to run queries via Hive Cli? if yes, then login to Ambari - HDFS. Perform comparison of configurations prior to adding host and after new adding host. Do similarly with HIVE and YARN services. It is possible that Ambari has reverted to default configuration during addition of new node. Looking at error, It appears that ambari is not able to write the file as user 'rene' which could be due to impersonation failure. (proxyuser settings)

PranayV · ‎10-02-2017

Can you share more details - Where do you try to do this? and possibly provide a screen shot?

PranayV · ‎09-29-2017

@ryan xia You can click on "Show advanced option" on "create cluster" wizard this with list you options to provide HDP and Ambari repositories URL.

PranayV · ‎09-29-2017

@Yair Ogen As per the log "message from server:"Host 'il-dev-05' is not allowed to connect to this MySQL server", user "hive"@"metastorehost" does not have a valid userid on mysql server at il-dev-05. Create a user on mysql with "hive"@"metastorehost" and provide full access to hive database. You can refer here - https://docs.hortonworks.com/HDPDocuments/Ambari-2.5.2.0/bk_ambari-administration/content/using_hive_with_mysql.html

PranayV · ‎09-29-2017

check this https://oozie.apache.org/docs/4.0.0/WebServicesAPI.html#Job_Log You can use curl to run the rest api.

PranayV · ‎09-29-2017

Since you are talking about secondary NameNode, a Secondary NameNode will never act as a Name/meta data service provider even if you shutdown primary Namenode. You will have to switch Secondary Namenode to standby NameNode using HA. You can read here - https://hadoop.apache.org/docs/r2.7.2/hadoop-project-dist/hadoop-hdfs/HdfsUserGuide.html#Secondary_NameNode. One way to check if your secondary NN has all latest fs image is by checking the size of CURRENT directory of NN and SNN.

PranayV · ‎09-29-2017

1) With Oozie UI you can look at the status of all past workflows. 2) If you have smartsense activity explorer and analyzer setup, you can query all the job that ran from activity.job table within given specified period and job name /type containing "oozie" in it.

PranayV · ‎09-27-2017

@Br Hmedna You are trying to export ORC data into mysql without converting it to text. You should use sqoop hive export to do this. Look at this link https://community.hortonworks.com/questions/22425/sqoop-export-from-hive-table-specifying-delimiters.html

PranayV · ‎09-27-2017

@Kumar Veerappan You will have to create an watcher/alert scripts that identifies which NN is active and alert/email if NN flips. Namenode service provides JMX which has information on which NameNode is active. Your watcher script can query this data to identify if NN failovers. name" : "Hadoop:service=NameNode,name=NameNodeStatus", "modelerType" : "org.apache.hadoop.hdfs.server.namenode.NameNode", "State" : "active", "NNRole" : "NameNode", "HostAndPort" : "host1:8020

PranayV · ‎09-27-2017

@Mohammad Shazreen Bin Haini If you are using ranger to manage permissions, there should 2 default policies. 1) HDFS policy - that gives hive full permission to read/write to /apps/hive/warehouse directory. 2) Hive policy - that gives hive full permission "hive" user to create, drop databases, tables.

PranayV · ‎09-27-2017

@Zack Riesland With smartsense, you can install activity analyzer and activity explorer that can parse all the job runs within the cluster. This does not provide the metrics at table lever rather it can provide you information at job level. You will have to make few modifications to extract table information from actual query and join it with smartsense metrics to capture your metrics.

PranayV · ‎09-27-2017

If you are looking for extracting text from PDF, I had done this via Apache Tika. Its simpler to use.

PranayV · ‎08-17-2017

if SLA is not the constraint, you can save RDD as temporary file and read it again via databricks. If you are running it via zeppelin dashboard, you can invoke shell interpreter and use sed to do an infile replace for xmlns: prior to reading it on your dataframe.

PranayV · ‎06-13-2017

Check application log for application_1497349747602_0006 in RM. Paste the stdout/stderr from job history

PranayV · ‎02-08-2017

I have seen this error for my customer. Issue was with memory footprint on the node hosting Zeppelin/Livy. Free memory was 1GB. This was since livy had many dead session which were not releasing memory. Deleting livy sessions helped increasing free memory. YOu can use livy rest api to view sessions and delete dead sessions.

PranayV · ‎01-25-2017

I think pre-emption is within leaf queues under same parent queue. That is why this behavior is observed.

PranayV · ‎01-25-2017

Is your cluster kerberized? Log says "Caused by: org.apache.hadoop.security.authentication.client.AuthenticationException: Authentication failed, status: 403, message: Forbidden" Can you check if you are running the job with user having valid kerberos ticket?

Online	Offline
Last Visited	‎05-15-2019 07:01 PM

Date Registered	‎01-04-2019 05:04 PM
Last Visited	‎05-15-2019 07:01 PM
Total Messages Posted	77
Total Kudos Received	27

Re: Hive migration to a new cluster & Hive metasto...

Re: Is it possibile to install services in a speci...

Re: Need help cleaning up after failed move of Hiv...

Re: Best way to re-add a server to HDP cluster?

Re: how to set number of map and reduce tasks

Re: Error: schema registry fails to start if user ...

Re: can we run more no of containers than the avai...

Re: Failed yarn job logs are missing when aggregat...

Re: How do I run simultaneous beeline batch jobs?

Re: can we run more no of containers than the avai...

Re: Hive migration to a new cluster & Hive metasto...

Re: RAM has been upgraded from 64 GB to 256 GB in...

Re: Auto Replication in HDFS after re-connecting m...

Re: Solr UI not working

Re: Is it possibile to install services in a speci...

Re: Hadoop jobs failing

Re: select count () and select : different behav...

Re: how to confirm if the job is in stuck state

Re: Hadoop unavailable after adding a new Host

Re: when i go to hive settings and try to set hive...

Re: How does Cloudbreak deploy HDP2.6.2?

Re: Hive Metastore won't start

Re: How to see all oozie jobs/workflows which fail...

Re: How to check if my Secondary Name Node is sync...

Re: How to see all oozie jobs/workflows which fail...

Re: Export from HDFS to mysql using Sqoop

Re: Name Node services fail over

Re: Can't execute query - Permission denied?,Can't...

Re: Is there a tool to track usage of each table

Re: how to develop map reduce pdf processing java ...

Re: Process an RDD through the databricks xml jar.

Re: Error while HDP-2.6.0.3 set up

Re: LivyServer exception

Re: Yarn queue limits don't apply

Re: Hive query exception from Java api