avr 14 2011

HPUX / RAC 11gR2 various tips

Tag: DBUggla @ 11 h 36 min

Here is a list of tips useful to install/troubleshoot a 11gR2 RAC cluster with HPUX 11.31.

  • mrouted configuration to allow multicast.
    Activate the mrouted daemon.
    export MROUTED=1
    
    into /etc/rc.config.d/netdaemons
  • Route to 169 from interlink.
    If like me you don’t have the interconnect link routed and known by DNS, ensure you have a route to 169 via your interconnect interface and IP.
    169.254.0.0           172.16.15.140      U     0    lan5002    1500
    
  • Do not forget patch hppac.
    This patch is not common and required by Oracle Clusterware.
    swinstall -p -x logdetail=true -s calisson.osdgre.external.hp.com:/var/depot/hp-ux/11.31/qpk PHSS_37042
    swinstall -x logdetail=true -s calisson.osdgre.external.hp.com:/var/depot/hp-ux/11.31/qpk PHSS_37042
    
  • Restart root.sh script for troubeshooting.
    This is an excellent tip to troubleshoot root.sh, this will allow to restart only root.sh and not the full install.
    Follow the excellent note : http://www.rachelp.nl/index_kb.php?menu=articles&actie=show&id=61
    dd if=/dev/zero of=/dev/oracle/asmdisk_ocr1 bs=1024k count=2048
    dd if=/dev/zero of=/dev/oracle/asmdisk_ocr2 bs=1024k count=2048
    dd if=/dev/zero of=/dev/oracle/asmdisk_ocr3 bs=1024k count=2048
    
    rm /var/opt/oracle/scls_scr/hx004105/grid/cssfatal
    rm /var/opt/oracle/ocr.loc
    /appl/grid/product/11.2.0/cluster/root.sh
    

    The Oracle Metalink articles ID 1050908.1 and ID 1053970.1 may greatly help to find issues.
  • Full cleanup crs to reinstall from scratch.
    I got issues not cleaning up /var/tmp/.oracle and /tmp/.oracle.
    Cleanup inittab, remove : "h1:3:respawn:/sbin/init.d/init.ohasd run >/dev/null 2>&1 </dev/null"
    init q
    rm -rf /appl/oraInventory
    rm -rf /appl/grid
    rm -rf /var/tmp/.oracle
    rm -rf /tmp/.oracle
    rm -rf /var/opt/oracle
    rm /sbin/init.d/init.ohasd
    rm /sbin/init.d/ohasd
    find /etc/rc*.d -name *ohasd -exec rm {} \;
    for i in 1 2 3; do
    dd if=/dev/zero of=/dev/oracle/asmdisk_ocr$i bs=1024k
    done
    rm /etc/oratab
    rm /usr/local/bin/dbhome /usr/local/bin/coraenv /usr/local/bin/oraenv
    

    Rebuild home.
    mkdir /appl/grid
    chown grid:oinstall /appl/grid
    chmod g+w /appl
    
  • cluvrfy command to check cluster status.
    Before adding a node, the following command should exit successfully.
    cluvfy stage -post hwos -n <existing and new nodes> -verbose
    cluvfy stage -pre nodeadd -n <new node> -fixup -verbose
    

    Especially take care of $ORACLE_HOME ownership and rights.
    A good article about cluvfy : http://satya-racdba.blogspot.com/2010/01/cluvfy-cluvfy-help-or-cluvfy-h-cluvfy.html
  • Add node command reminder.
    ./addNode.sh -silent CLUSTER_NEW_NODES={node_name} CLUSTER_NEW_VIRTUAL_HOSTNAMES={node_name-vip}
    
  • Root.sh to run from /appl/grid ($ORACLE_BASE) (strange).
    To be confirmed, but I got issues depending from the running location. So try to run from various location…
  • Copy 2nd DVD
    11gR2 is composed of 3 DVDs (2xDB + 1xGI), installation will be fine although second can not be extracted. But it will lack DBCA assistants etc… that will make next steps painful.
  • Delete engine to restart installation.
    To remove an engine after installation (because of missing DVD as an example. ;) ).

    • Edit oraInventory/ContentsXML/inventory.xml by removing engine entry.
    • Remove engine’s oracle home.