As I just went through the installation of DB2 pureScale 9.8 on an AIX 6.1 machine I thought I’d share a few things that weren’t so obvious and made it a not so smooth experience.
There’s a real good article on Developer Works which describes the steps to install pureScale including some pre-req steps which are very important. In this particular installation we installed both a member and a CF on the same LPAR as it was for a test and not production environment.
The first thing that caused a problem was the uDAPL driver on AIX 6.1. You really need to make sure to have the correct driver matching the OS level. In addition you want to check that the correct driver and interface is configured in /etc/dat.conf.
Here’s some example:
root@node101:/> ifconfig ib0
ib0:flags=e3a0063<UP,BROADCAST,NOTRAILERS,RUNNING,ALLCAST,MULTICAST,GROUPRT> pmtu 2048 qkey 0x0 qpn 0x3f lid 0x34 hca iba1 port 1
inet 10.10.10.1 netmask 0xffffff00 broadcast 10.10.10.255
tcp_sendspace 262144 tcp_recvspace 262144 rfc1323 1
Notice in the second line the entry which defines the name of the interface iba1.
This is important to note as you want to make sure that in /etc/dat.conf the same name is used in order to address the right port.
root@wp7801a1:/> cat /etc/dat.conf
hca0 u1.2 nonthreadsafe default /usr/lib/libdapl/libdapl.a(shr_64.o IBM.1.1 “/dev/iba1 1 ib0″ ” “
#hca1 u2.0 nonthreadsafe default /usr/lib/libdapl/libdapl2.a(shr_64.o) IBM.1.1 “/dev/iba0 1 ib0″ ” ”
Now when looking at the /etc/dat.conf file, we first need to determine which version of DB2 pureScale we are installing. For version 9.8 we want version 1.2, for 10.1 version 2.0. So depending on your version, uncomment the respective line. Side note: after installing a driver update, check again. I learned this also the hard way :).
This solved my communication problems which prevented the members and CF to communicate and therefore start the cluster.