Netezza Performance Server 11.0.6.0 release notes

Deployment options: Netezza Performance Server for Cloud Pak for Data System

Version 11.0.6.0 provides several stability fixes and enhancements.

Enhancements and improvements

Netezza Performance Server is certified with IBM Spectrum Protect (formerly TSM) 8.1.10 and EMC Networker 19.1.

Resolved issues

Upgrade stability
  • Fixed the lowercase upgrade issue.
  • Improved the error message for container upgrade failure cases.
  • Preserved INZA license across upgrade.
Container stability
  • Fixed the issue where ssh into the Netezza Performance Server container was failing. This issue was caused by systemd-logind not clearing sessions and filling up /run.
  • Fixed ERROR: Too many open files.
  • Fixed nzstart: Error: insufficient tmpfs size. tmpfs of size 155586 mounted on /dev/shm is needed.
  • Fixed ERROR: Sysctl value for 'kernel.sem' is '128 1048576 32 128'; must be at least '128 1048576 32 65535'.
System Management
  • Added support for the nzds -reblance CLI for improving data slice ownership reassignment.
  • Fixed the issues with handling different vendor disks.
  • Fixed the system management policy issue with handling unresponsive SPUs.
  • Fixed the issue with nzevent -template not showing pre-defined events.
  • Fixed the issues in FPGA to prevent DISK_FPGA_ERROR.
  • Fixed the SFP link down issue.
SQL stability
  • Improved compilation time for large CASE WHEN nested queries.
  • Fixed the issue with timezone difference on the IPS host and SPUs.
  • Fixed the system crash issue which occurred while loading a table with the fixed format option.
  • Fixed an isue to prevent postmaster core segmentation fault.
  • Fixed the permission issue, which was causing backup to fail with ERROR: LOCK TABLE: permission denied when role was set.
  • Fixed the issue with setting view owner when the view was created inside the procedure.
  • Fixed ERROR: Unterminated quoted string, which was caused when the generate statistics keywords were present inside SQL.
  • Fixed the issue with non-admin user getting an error EXCEPTION thrown: invalid source table name even though the user had required permissions to run the procedure.
  • Fixed the Postgres nzstart crash which occurred during metadata transfer of versioned tables.
  • Fixed ERROR: Memory exhausted in AllocSetAlloc() which occurred while significant amount CREATE OR REPLACE View statements were fired as part of single transaction.
  • Fixed the issues with the nz_md5_qsum()utility.
  • Fixed the issue with enabling crypto on Netezza Performance Server.
  • Fixed the issue with cross database being written to replicated database from non-replicated databases in a replicated environment, which was causing inconsistencies on subsystem.

Known issues

1. Upgrade
  • Upgrading from 11.0.4.1 to 11.0.6.0
    The upgrade is successful with a non-fatal error:
    “WARNING: Could not drop nz_internal_temp_user user”
    Use this mandatory procedure, after the system comes online with 11.0.6.0.

    WORAKAROUND:

    As the nz user, run the commands inside the container:
    1. nzsql -c "update  _t_object set objowner = U.usesysid, objmodifier = U.usesysid from _t_user U where objid = 5342 and U.oid = 4900; "
    2. nzsql -c "drop user NZ_INTERNAL_TEMP_USER;"
  • Upgrading from 11.0.6.0 to 11.1.0.0
    nzstart returns a 
    Sysctl value for 'kernel.sem' is '128 1048576 32 32000'; must be at least '128 1048576 32 32768
    error after upgrading the container from 11.0.6.0 to 11.1.0.0. Note that in this case Netezza Performance Server is still on 11.0.6.0.
    WORKAROUND:
    1. As the root user, run the commands inside the container:
      sed -i '/^kernel.sem =/s/65535/32768/' /etc/sysctl.conf
sysctl -w kernel.sem="128 1048576 32 32768"
    2. As the nz user, run:
      nzstart
  • Downgrading from 11.0.6.0 to 11.0.4.1
    While downgrading from 11.0.6.0 to 11.0.4.1, nzstart fails with this error message:
    WARNING: SELinux is configured to be started on reboot (/etc/selinux/config)
Use of uninitialized value $ENV{"NZ_REMOTE_FILE_SERVER"} in concatenation (.) or string at /nz/kit/bin/adm/nzchkhost line 415.
Use of uninitialized value in concatenation (.) or string at /nz/kit/bin/adm/nzchkhost line 415.
Use of uninitialized value in concatenation (.) or string at /nz/kit/bin/adm/nzchkhost line 415.
Use of uninitialized value in concatenation (.) or string at /nz/kit/bin/adm/nzchkhost line 415.
mount.nfs: Failed to resolve server : Name or service not known
mount.nfs: Operation already in progress
Use of uninitialized value $ENV{"NZ_REMOTE_FILE_SERVER"} in concatenation (.) or string at /nz/kit/bin/adm/nzchkhost line 417.
Use of uninitialized value in concatenation (.) or string at /nz/kit/bin/adm/nzchkhost line 417.
Use of uninitialized value in concatenation (.) or string at /nz/kit/bin/adm/nzchkhost line 417.
Use of uninitialized value in concatenation (.) or string at /nz/kit/bin/adm/nzchkhost line 417.
ERROR:  Unable to mount gpfs export: :/ipshost on /mnt/gpfs_test/
    WORKAROUND:
    Before downgrading Netezza Performance Server to 11.0.4.1, copy nzchkhost from the source kit to the target kit (11.0.4.1):
    1. As the nz user, run the commands inside the container:
      mv /nz/kit.11.0.4.1/bin/adm/nzchkhost
/nz/kit.11.0.4.1/bin/adm/nzchkhost.original
cp /nz/kit.11.0.6.0/bin/adm/nzchkhost /nz/kit.11.0.4.1/bin/adm/
    2. As the nz user, run:
      nzstart
2. Netezza Performance Server authentication
If Netezza Performance Server authentication is other than LOCAL (for example, LDAP/ AD/Kerberos), the upgrade to 11.0.6.0 will fail with ERROR: Failed to bring IPS offline: Error: login error - LDAP authentication failed for user 'NZ_INTERNAL_TEMP_USER'. WORKARAOUND:
Note: Before running step 7 from Upgrading the software, complete the below steps:
  1. Check the current authentication on system:
    nzsql -c "show authentication;"
    The expected output is: 
    SYSTEM.ADMIN(ADMIN)=> show authentication;
      AUTH_OPTION      | AUTH_VALUE
-----------------------+------------
 AUTHENTICATION METHOD | LDAP
(1 row)OR SYSTEM.ADMIN(ADMIN)=> show authentication ;"
       AUTH_OPTION       |      AUTH_VALUE
-------------------------+-----------------------
 AUTHENTICATION METHOD   | KERBEROS
  2. Modify line number 34 of /nz/kit.11.0.6.0/share/upgrade/plugins/nz/Plugin/CheckLocalOption.pm Change: 
    my $query = "create user nz_internal_temp_user with password 'password' ";
    To:
    my $query = "create user nz_internal_temp_user with password 'password' auth local ";
  3. Continue with step 7 from Upgrading the software, for example:
    su -c "/nz/kit.11.0.6.0/sbin/nzupgrade  -f upgrade
3. System Management
After a successful SPU activation, the NzLocal partition is incorrectly reported by nzspupart [show] [-issues] as Degraded even though it is Healthy.

Limitations

System Management
  • Don't activate an SPU that is in the Booting state.
  • By default sysmgr.enableAutoRegen and sysmgr.enableAutoRebalance are ON. Never activate an SPU if user explicitly set sysmgr.enableAutoRegen, sysmgr.enableAutoRebalance, or both are turned OFF.
  • Always use nzds -rebalance. Don't use nzhw -rebalance as it has unpredictable effects.