7.00.02 - PUT Troubleshooting Reference Topics - Aster Execution Engine

Aster Instance Installation Guide for Aster-on-Hadoop Only

prodname
Aster Execution Engine
vrm_release
7.00.02
created_date
July 2017
category
Installation
featnum
B700-5022-700K

This information describes troubleshooting reference topics for issues that you may encounter when using PUT to install an Aster instance.

For information on other installation issues not related to PUT, see Installations on Teradata Hadoop Appliances Troubleshooting Reference Topics or Installation Issues.

Symptom and Solution
ERROR: Failed dependencies: libsgutils2.so.2()(64bit)
  To Correct: RHEL: yum install sg3_utils-libs
    SUSE: You should not encounter this issue.
  Comments: RHEL: sg3_utils-libs is correct even though the error indicates libsgutils2.
    SUSE: If you encounter this issue, contact your Teradata Customer Support Representative.
ERROR: Failed dependencies: libxslt.so.1()(64bit)
  To Correct: Ensure that you are on RHEL 6.6, 6.7 or 6.8. If you're on 6.6, 6.7 or 6.8 and still see this error, issue this command: yum install libxslt
PUT returns this warning: sh: /bin/fuser: No such file or directory
  Comment: You can ignore this warning.
PUT stops processing on the Simultaneous Operation screen. You might also see PUT log entries similar to ERROR: Could not get addressinfo for host ...
  Cause: There is no reverse address translation from the hostname to the host IP address.
  To Correct: Add lines similar to the following to the /etc/ directory on all nodes of the cluster. The queen must be able to ping each node using its respective hostname and each host must also be able to ping the queen by the queen's hostname.
IP-Address   Full-Qualfied-Hostname   Short-Hostname
10.100.2.100       ilo2-1                 ilo2-1
While installing PUT, Login failed messages are returned.
  Cause: There is no reverse address translation from the host name to the host IP address.
  To Correct: Add lines similar to the following to the /etc/ directory on all nodes of the cluster. The queen must be able to ping each node using its respective hostname and each host must also be able to ping the queen using the queen's hostname.
IP-Address Full-Qualfied-Hostname Short-Hostname
10.100.2.100 ilo2-1 ilo2-1
PUT returns Node time out of sync for Temporal Teradata after performing the Run System Check step.
  Cause: The clocks on the nodes of a cluster may have drifted.
  To Correct: Reset the clocks.
PUT displays this error: (ID-159) teradata-jre7 is not installed or properly configured.
  To Correct: You can ignore this error.
During the PUT procedure, the tool indicates TPA nodes have not been designated for this update.
  To Correct: Select Yes keep them as they are.
  Comment: This notice occurs after you've grouped your nodes and assigned roles to them.
During the PUT procedure, the tool indicates that the dependency on libasound is not satisfied.
  To Correct: Contact your Teradata Customer Support Representative.
For Teradata personnel:
During the PUT procedure, the tool indicates that the dependency on SNMP is not satisfied.
  Cause: net-snmp is not installed.
  To Correct: Contact your Teradata Customer Support Representative.

These corrections are actionable by only Teradata personnel:

RHEL:
  1. Issue this command: cd /etc/yum.repos.d.
  2. Configure aster-mirror.repo for the desired RHEL version.
  3. Rename all other .repo files so that they have a different extension. For example, mv foo.repo foo.repo.fixme.
  4. Run yum install net-snmp.
  5. Restore the original names of the files that were renamed.
SLES:
  1. Download the rpms from http://mirror.asterdata.com/sles/11.sp2/x86_64/suse/x86_64/rpm%20-Uvh%20perl-SNMP-5.4.2.1-8.12.6.1.x86_64.rpm%20net-snmp-5.4.2.1-8.12.6.1.x86_64.rpm.

    The two libraries must be installed together because they have a dependency on each other.

  2. Install the two SLES rpms into a directory on the cluster and point TDPUT to the location of these packages. Doing so will resolve the dependencies and install them along with the Aster packages.
PUT returns AccessControlException or Access denied message shortly after starting PUT.
  To Correct: Ensure that you have a supported version of Java running on the computer where your browser is running. Use Java 1.7 or 1.8.
During an Appliance installation, PUT indicates that package dependencies are not met on libstdc++.so.
  Cause: Aster relies on a newer version of libstdc++ than SLES 11 SP1 provides.
  To Correct: Contact your Teradata Customer Support Representative.

This correction is actionable by only Teradata personnel:

As a temporary workaround, download libstdc++44-4.4.2_20100116-28.1.x86_64.rpm from /asterfs2/mkohli/RPMS/common.
Do not copy all of the files in the common directory; copy only the specific files indicated in the message (for example, libstdc++).
PUT stops processing during the Node Details step.
  To Correct: Remove any large mounted drives and retry the PUT Node Details step.
The aster-db-common package fails to install.

For example:

Install> Could not create SQL/MR user (status 6): extensibility:extensibility

Install> error: %pre(aster-db-common-07.00.01.01-42728-ax7.0.x86_64) scriptlet failed, exit status 2

  Cause: The extensibility group is not defined on the nodes.
  To Correct: As a temporary workaround, add the groups on each of your nodes using the IP addresses for your cluster. This example command shows how to add the extensibility group to a node:

for ip in 10.80.169.{50..54}; do echo $ip; ssh $ip groupadd extensibility; done

  Comment: This example command shows how to determine if extensibility groups are defined:

for ip in 10.80.169.{50..54}; do echo $ip; ssh $ip getent group | grep extensibility; done

Internet Explorer won't connect to the AMC.
  Cause: This issue can result from an upgrade to AMC security.
  To Correct: To enable TLS v1.2 on Internet Explorer (IE), perform these actions in IE:
  1. From the Menu, select Tools Internet Options Advanced.
  2. Scroll to the Security section.
  3. Select TLS 1.2 Protocol, which enables the protocol.
When performing an upgrade using PUT, the installation stalls during installation of the aster-db-toolchain package.
  Cause: This issue may occur because the IOInterceptor is mounted. For example:

IOInterceptor on /data/ncluster/iointerceptor/mountdir type fuse.IOInterceptor (rw,nosuid,nodev,allow_other,user=beehive)

  To Correct: Unmount the partition on all nodes. To perform this action, issue the umount IOInterceptor command on all nodes in the cluster. You may need to run the command more than once on the nodes in the cluster.
Overall Package Install Results indicates that the aster-db-worker package failed to install.
  To Correct: On Overall Package Install Results, click Retry.
  Comment: The number of nodes where the aster-db-worker package was not installed is shown under the Fail column. To view log messages for these nodes: click aster-db-worker, and then click on a node to view the log messages.
A cluster is not found or grouped properly. For example, nodes that were previously grouped together are now in individual groups (one node per group).
  Cause: The issue may arise because asterenv.sh was not sourced, or was expected in the wrong directory.
  To Correct: Copy asterenv.sh to /etc/profile.d/.

If you are part way though the PUT procedure and encounter this problem after copying asterenv.sh to /etc/profile.d/, return to Find Systems step in the PUT procedure and continue the installation process from that step.

  Comment: For additional information, refer to the following Grouping problems symptom.
Grouping problems. Nodes that should all be in one group are not. For example, workers are in different groups.
  Cause: In some cases, this issue is caused by nodes in the group having different packages.
  To Correct:
  1. Run this command on each node:

    rpm -qa | sort > MyRPMS

  2. If there are differences in the output on each node, install the RPM files necessary for all nodes to have the same output from this command:

    rpm -qa | sort.

    Ensure that you check ALL packages. Nodes can be grouped differently because some have packages such as atop and some do not.

  3. If there are no differences in the output on each node, and if ncli is in the path (asterenv.sh was run), contact your Teradata Customer Support Representative.
  Comment: For additional information, refer to the previous symptom “A cluster is not found or grouped properly.”