Connect MicroStrategy to HIVE with ODBC over KNOX failed

[#1]

Hi,
we have troubles to connect MicroStrategy(=MSTR) to HIVE on HDP3.1 over Knox, when trying to connect we receive:
 

Unknow failed: Unable to connect.
Error type: System call error. Error communicating with connection process.
Find target process dead while waiting for reply.

We installed ClouderaHiveODBC-2.6.4.1004-1.x86_64.rpm driver.
MSTR Inteligent Server is a RHEL-7.4 machine, installation has been realized according to Cloudera-ODBC-Driver-for-Apache-Hive-Install-Guide.pdf:

yum --nogpgcheck localinstall [RPMFileName]

As we are not sure that yum resolves all dependencies, we install the sasl libraries as well:

cyrus-sasl    x86_64    2.1.26-23.el7    rhel-7-server-rpms

odbc.ini looks like:

[hdp.hive.lxprod]
Driver=/opt/cloudera/hiveodbc/lib/64/libclouderahiveodbc64.so
Host=hdp.lxprod
PORT=443
Schema=default
ServiceDiscoveryMode=0
HiveServerType=2
AuthMech=3
ThriftTransport=2
HTTPPath=/gateway/DM/hive
SSL=1
AllowSelfSignedServerCert=1
CAIssuedCertNamesMismatch=1
TrustedCerts=/etc/pki/tls-ca-bundle.pem

We tested the ODBC-Connection to hdp.lxprod on a Windows machine, with the same driver (ClouderaHiveODBC64.msi, Cloudery ODBC Driver for Apache Hive 2.06.04.1004, Win 64-bit) and it works, so we assume that on HDP-side is everthing ok, and Knox and HIVE is configured well.

We suppose that the problem is located on the MSTR server.

Any ideas?

Thanks in advance

Michael.

[#2]

Hi Mike,

Try to add LogPath=/path/to/dir and LogLevel=6 in ODBC configuration to enable TRACE logging and see what might be shown in the ODBC trace logs.

Please share the output once collected.

Cheers
Eric

[#3]

Hi Erik,

the odbc section looks now like this:

[hdp-hive-lxrls]
Driver=/opt/cloudera/hiveodbc/lib/64/libclouderahiveodbc64.so
Host=hdprelease.dm-drogeriemarkt.com
PORT=443
Schema=default
ServiceDiscoveryMode=0
HiveServerType=2
AuthMech=3
ThriftTransport=2
HTTPPath=/gateway/DM/hive
SSL=1
AllowSelfSignedServerCert=1
CAIssuedCertNamesMismatch=1
TrustedCerts=/etc/pki/tls-ca-bundle.pem
LogPath=/srv/mstr/log/odbcLog/
LogLevel=6

We make sure, that /srv/mstr/log/odbcLog/ is writeable for the process, but no Logfile has been written.

But we found segfault messages in /var/log/messages:

Jul  8 10:03:36 mstr-intranet-01 kernel: M8MulPrc[86769]: segfault at 0 ip 00007feeeccdb016 sp 00007feeda211d58 error 4 in libc-2.17.so[7feeecb9d000+1c2000]
Jul  8 10:03:50 mstr-intranet-01 kernel: M8MulPrc[86808]: segfault at 0 ip 00007f7de2325016 sp 00007f7dcf85bd58 error 4 in libc-2.17.so[7f7de21e7000+1c2000]
Jul  8 10:05:41 mstr-intranet-01 kernel: M8MulPrc[87501]: segfault at 0 ip 00007f2772bb1016 sp 00007f27600e7d58 error 4 in libc-2.17.so[7f2772a73000+1c2000]
Jul  8 10:05:55 mstr-intranet-01 kernel: M8MulPrc[87535]: segfault at 0 ip 00007f1a5f2b0016 sp 00007f1a4c7e6d58 error 4 in libc-2.17.so[7f1a5f172000+1c2000]
Jul  8 10:09:41 mstr-intranet-01 kernel: M8MulPrc[88893]: segfault at 0 ip 00007ff50653b016 sp 00007ff4f3a71d58 error 4 in libc-2.17.so[7ff5063fd000+1c2000]
Jul  8 10:09:55 mstr-intranet-01 kernel: M8MulPrc[88926]: segfault at 0 ip 00007fb919338016 sp 00007fb90686ed58 error 4 in libc-2.17.so[7fb9191fa000+1c2000]

uname -r : 3.10.0-862.14.4.el7.x86_64

cat /etc/redhat-release : Red Hat Enterprise Linux Server release 7.6 (Maipo)

I will try to install an older Cloudera Driver 2.5.x, and see what happens.

Regards,

Michael.

[#4 · Accepted solution]

Hi Eric,

we installed the ClouderaHiveODBC-2.5.25.1020-1.el7.x86_64.rpm driver on the RHEL-Server, and it works!

Seems to be a problem with ClouderaHiveODBC-2.6.4.1004-1.x86_64.rpm driver.

Kind regards,

Michael.

[#5]

Hi Michael,

Thanks for the info. I am curious on whether 2.6.1 version will work or not? 2.6.4 is marked as C6 compatible, I am not sure if that is meant for C6 or not.

Cheers
Eric

[#6]

Hi Eric,

I think the "C6 Compatibility" can be excluded because the driver for Windows works in conjunction with the same HDP cluster. Probably a problem with the implementation of the driver for Linux or RedHat.

I will not further investigate and will mark this post as solved.

Regards,

Michael.