- AD6.20.06 to AD6.20.08 upgrade sometimes hangs
Details: On occasion, upgrading from AD6.20.06 to AD6.20.08 using TDput on SLES12 SP2 or RHEL7.x results in a hung state.
To work around this problem, shut down the cluster services before performing an upgrade by issuing the ncli system softshutdown command.
- For kerberized clusters, open secure port 1019 as alternative to unsecure port 50010
Details: All Aster nodes should be able to access all Hadoop data nodes over port 50010. Port 50010 is for unsecure access. To securely access Hadoop Data nodes when a cluster is kerberized, open port 1019. See the Hadoop Ports listed in the Teradata Aster® Database User Guide.
- LDAP configuration in Aster fails to accept multiple CA certificates
Issue: BH-9633Details: Perform these tasks to enable multiple CA certificates:
- Copy the certificates into a directory.
- List chown and chmod commands to make the certificates readable by the proper users.
- Add /usr/bin/c_rehash /home/beehive/data/queenDb/<name of the directory>/ to hash the certificate files so OpenSSL can use them.
- Perform a soft restart on the cluster.
- Verify by logging into the act.If the common name or subject alternative name on the certificate provided by the LDAP server does not match the hostname to which the server is connecting, the parameter 'LDAPTLS_REQCERT’ must be set to allow in the profile, For example, 'LDAPTLS_REQCERT=allow'. By default, this parameter is set to demand.
- Use bynet0 on Aster Backcup clusters
Details: The Aster Database and Aster Backup clusters must be configured to use the same bynet. Either bynet0 or bynet1. By default the Aster Database cluster is configured to use bynet0. So the Aster Backup cluster should also be configured to use bynet0.
- Disk Quota Monitoring must be disabled
This feature must be disabled.Although the feature is documented in the version of the Teradata Aster® Database User Guide that supports the AD 6.20.01 release, support for the feature has been removed from the AD 6.20.01 release.
To disable this feature, run the following on the queen:ncli qos pqfeatureoff
If ACT does not respond with a prompt after a short amount of time, enter the following commands on the queen:
pkill queenExec pkill QosManagerExec ncli qos pqfeatureoff
- Failover to secondary queen node fails if packages are not identical
Details: Discrepancies in primary and secondary queen node packages prevent a successful failover.
Make sure that the packages installed on the secondary queen are identical to that of the primary queen node for a failover to occur successfully.
- Aster Clients and Teradata Aster R version support on AD6.20.01
Details: The Aster Client software bundle includes Aster Client packages and Aster R packages. Depending on your product needs, these versions of the Aster Client software are supported for use with Aster Database version 6.20.01:
- To install Aster clients and utilities, use the 6.20 release of the Aster Client software.
- To install Aster R, use the 6.21 release of the Aster Client software.
- Parent child query tree is affected by filters applied in Processes tab
Issue ID: BH-7743
Details: Application of a Process Filter in the Processes tab of the AMC is likely to cause incomplete information in the Parent-Child Process Relationships tree of the tab.
- AMC shows incorrect session user after set role
Issue ID: BH-7354
Details: Statements run by an impersonated user (role) no longer show the impersonated user. AMC shows the user who originally started the sessions.
changepartitioncountcheck may underestimate the space required for OPS
Issue ID: BH-7185
Details: When OPS is used, partitions compressed at the HIGH or MEDIUM level are converted to be compressed at the LOW level. The calculations done by changepartitioncountcheck to estimate disk space requirements do not take into account the increased space required by the conversion.
To work around this issue, use the ALTER TABLE or ALTER INDEX commands, specifying COMPRESSION LOW, before partition splitting so that HIGH or MEDIUM compression is not used.
- Unable to connect to AMC
Issue ID: BH-6591
Details: In rare cases, the Aster Management Console (AMC) does not start. If this happens, your cluster’s Java might not be compatible with Aster’s.
To work around this problem, run the following two commands to restart the AMC:
source /home/beehive/config/asterenv.sh unset JRE_HOME && JAVA_HOME=/home/beehive/toolchain/x86_64-unknown-linux-gnu/jdk1.8.0_45 CATALINA_BASE=/home/beehive/amc/webserverCATALINA_OPTS=-Xmx256m /home/beehive/toolchain/noarch/apache-tomcat-8.0.24/bin/catalina.sh run &
- After connecting to AMC, Firefox consumes CPU and will not exit
Issue ID: BH-7103
Details: After pointing Firefox to a cluster, FireFox starts to consume about 25% of CPU even when it is not doing anything. Even after exiting Firefox, it continues to run and consume CPU.
ncli system addnode
--clean option, error messages, and help text not handled properly
Issue ID: BH-5832/BH-5230
Details: The --clean option and MAC address are no longer supported for the ncli system addnode command. However, using the option does not generate an error.
- Partition splitting fails intermittently with duplicate key error
Issue ID: BH-4134
Details: A duplicate key error is intermittently returned when a CREATE TABLE statement is issued for a logically partitioned table.
To work around this issue, retry the partition splitting operation.
ncli sshkeys disable removes keys that are not intra-cluster keys
Issue ID: BH-4930
Details: After using ncli sshkeys disable, a worker or loader node can lose keys that are unrelated to the cluster.