From: Frye, Matthew (Matthew.Frye@REXHEALTH.COM)
Date: Tue Nov 05 2002 - 16:16:09 EST
Right. I'd rather not create extra vgs or filesystems. That's more work
than this is worth since everything is on SAN. I followed Paul's advice
below instead with good results, except that I chose to reboot the secondary
instead of loading the kernel extensions manually. I couldn't have easily
tested srvmrgl easily anyhow since those filesystems are part of HA.
> ----------
> From: Gene Sais[SMTP:Gsais@CO.PALM-BEACH.FL.US]
> Reply To: IBM AIX Discussion List
> Sent: Tuesday, November 05, 2002 10:26 AM
> To: aix-l@Princeton.EDU
> Subject: Re: HACMP and the post-Oracle 8.1.7-install error message
>
> Its recommended to install Oracle on the failover node which is usually
> not a problem since I always use the failover node as an Oracle test
> server. Some use the failover Oracle Home and others use the primary
> Oracle Home, your choice. I prefer to use the primary, export them and
> import them on the failed node. For this to work, your filesystems/vgs
> must be named differently in the cluster.
>
> hth,
> Gene
>
> >>> Paul@BHAILE.NET 11/02/02 04:58AM >>>
>
> You are right. Rootpre.sh adds inittab entries for post-wait kernel
> extensions during the Oracle installation. So you need to copy these two
> files -
> /etc/pw-syscall
> /etc/ora_kstat
>
> onto your secondary node and then correct inittab. The primary will have
> some entries like this ....
> orapw:2:wait:/etc/loadext -l /etc/pw-syscall
> orakstat:2:wait:/etc/loadext -l /etc/ora_kstat
>
> Just run the extensions manually to test with svrmgrl.
> Should be fine.
> Paul.
> ----- Original Message -----
> From: "Frye, Matthew" <Matthew.Frye@REXHEALTH.COM>
> Newsgroups: bit.listserv.aix-l
> To: <aix-l@Princeton.EDU>
> Sent: Saturday, November 02, 2002 1:43 PM
> Subject: HACMP and the post-Oracle 8.1.7-install error message
>
>
> > A story and a questions for AIX SysAdmins working with Oracle 8.1.7:
> >
> > We installed Oracle 8.1.7 last week and had no problems until this
> morning
> > when the HACMP failed over thanks to dead-man-switch. When my DBA tried
> to
> > start srvmgr on the secondary node, it gave the following error:
> >
> > 0509-136 Symbol aix_ora_pw_version3_required (number 275) is not
> exported
> > from
> > dependent module /unix.
> >
> > Due to an environmental variable problem Oracle was not up at the time
> this
> > error occurred. However, I have read that this error is related to the
> > "rootpre.sh" script not having been run and certain kernel extensions
> having
> > been installed. Is it correct to presume that since this script was run
> on
> > my primary node, and not on my secondary that this cause the error once
> the
> > HA complex failed over? And if so, is it correct to presume that
> running
> > this pre-install script (or at least the relevant commands contained
> > therein) will implement the kernel extensions as required by Oracle
> 8.1.7
> > and it's gang of executables? Anyone had experience with this scenario?
> >
> >
> > Matt Frye
> > Sr. Systems Pgmr, RS/6000 Group
> > Rex Healthcare
> > (919) 784-3791 Office
> > (919) 310-2011 Pager
> >
>
This archive was generated by hypermail 2.1.7 : Wed Apr 09 2008 - 22:16:18 EDT