HPE Ezmeral Data Fabric (MapR) Client Installation 7.1.0

[SunnyC]

We have installed a 3-node cluster of HPE Ezmeral Data Fabric (MapR) on RHEL 8.6. We are using Core 7.1.0 with mapr-patch-7.1.0.12 and EEP 9.1.2.

We are preparing the installation of MapR Client on our BI Application Server for data connection, below are the package installed on the BI Server:

mapr-client-7.1.0.0.20221012092141.GA-1.x86_64.rpm
mapr-hadoop-util-3.3.5.0.202307032352-1.x86_64.rpm
mapr-librdkafka-0.11.3.202201311452-1.x86_64.rpm

After the installation, I notice that hadoop version 3.3.4 is embedded in the binary files of the mapr client as below:

[root@<hostname> mapr]# grep '''hadoop-3.3.4''' */*
Binary file lib/libMapRClient_c.so matches
Binary file lib/libMapRClient_c.so.1 matches
Binary file lib/libMapRClient.so matches
Binary file lib/libMapRClient.so.1 matches

This causes error for the cmd 

#hadoop fs -ls /

some time after the maprticket is generated.

2024-06-06 15:19:48,3572 ERROR Client fc/client.cc:604 Thread: 51509 GetCoreSiteXmlPath: could not open /opt/mapr/hadoop/hadoop-3.3.4/etc/hadoop/core-site.xml errno 2
2024-06-06 15:19:48,3722 ERROR LibHdfsFCClusterConf fc/fcclusterconf.cc:888 Thread: 51509 QueryDns: res_query failed: No address associated with name

The cmd works again for short time after maprlogin is rerun.

Given the mapr server is using hadoop 3.3.5, is there any patch for the mapr-client?

Or I should upgrade the mapr client or downgrade the hadoop for the client server?

[Jitendra_Mohan]

-This issue looks like a bug and you need to open a support case to look further.

[Dave Olker · Accepted solution]

As to your question about whether there are any mapr-patches for the EDF 7.1.0 client, they can be found by going to https://sftp.mapr.com and logging in with the uesr name "maprpatches" and no password.  Once logged in, navigate to the v7.1.0 directory and then either rpm or deb depending on the flavor of Linux you're running on the client.  There you will see the following files:

-rw-r--r-- 1 root root 581 Apr 12 18:09 20240401014219-md5sum.txt
-rw-r--r-- 1 root root 581 May 25 11:22 20240522091320-md5sum.txt
-rw-r--r-- 1 ec2-user ec2-user 12372 May 25 11:26 CHANGELOG-7.1.0EBF.txt
-rw-r--r-- 1 ec2-user ec2-user 855372829 Apr 12 18:01 mapr-patch-7.1.0.12.20240401014219.GA-1.x86_64.rpm
-rw-r--r-- 1 ec2-user ec2-user 855504208 May 25 11:17 mapr-patch-7.1.0.13.20240522091320.GA-1.x86_64.rpm
-rw-r--r-- 1 ec2-user ec2-user 280693602 Apr 12 18:03 mapr-patch-client-7.1.0.12.20240401014219.GA-1.x86_64.rpm
-rw-r--r-- 1 ec2-user ec2-user 280694853 May 25 11:17 mapr-patch-client-7.1.0.13.20240522091320.GA-1.x86_64.rpm
-rw-r--r-- 1 ec2-user ec2-user 40727592 Apr 12 18:03 mapr-patch-loopbacknfs-7.1.0.12.20240401014219.GA-1.x86_64.rpm
-rw-r--r-- 1 ec2-user ec2-user 40727608 May 25 11:17 mapr-patch-loopbacknfs-7.1.0.13.20240522091320.GA-1.x86_64.rpm
-rw-r--r-- 1 ec2-user ec2-user 397848 Apr 12 18:03 mapr-patch-nfs4server-7.1.0.12.20240401014219.GA-1.x86_64.rpm
-rw-r--r-- 1 ec2-user ec2-user 397856 May 25 11:17 mapr-patch-nfs4server-7.1.0.13.20240522091320.GA-1.x86_64.rpm
-rw-r--r-- 1 ec2-user ec2-user 358932 Apr 12 18:03 mapr-patch-posix-client-basic-7.1.0.12.20240401014219.GA-1.x86_64.rpm
-rw-r--r-- 1 ec2-user ec2-user 358936 May 25 11:17 mapr-patch-posix-client-basic-7.1.0.13.20240522091320.GA-1.x86_64.rpm
-rw-r--r-- 1 ec2-user ec2-user 359032 Apr 12 18:03 mapr-patch-posix-client-platinum-7.1.0.12.20240401014219.GA-1.x86_64.rpm
-rw-r--r-- 1 ec2-user ec2-user 359032 May 25 11:17 mapr-patch-posix-client-platinum-7.1.0.13.20240522091320.GA-1.x86_64.rpm

Since you are using mapr-patch 7.1.0.12 on the server, you can install mapr-patch-client-7.1.0.12.20240401014219.GA-1.x86_64.rpm on the client system so that they are using the same code base.  I'd suggest trying that to see if it helps.

Also, looking at one of your errors:

fc/fcclusterconf.cc:888 Thread: 51509 QueryDns: res_query failed: No address associated with name

That sounds like an issue resolving the entries in the mapr-clusters.conf file, so I'd investigate that as well.  Make sure any cluster nodes can be resolved by DNS.

[SunnyC]

Thank you for the update.

I was looking for the patch file under v7.1.0/client and overlooked the rpm folder. 

Both error are fixed after patch is applied.

[Sunitha_Mod]

Hello @SunnyC,

That's Excellent! 

We are extremely glad to know the issue has been resolved.