Oracle 10gR2 upgrade

[Danesh Qureshi]

My colleague is a DBA and he is trying to upgrade the databases on Alpha server ES47 to 10gR2 and the following warning messages have been reported. See below for errors.

The alpha ES47 server is running the following O/S
# sizer -v
Compaq Tru64 UNIX V5.1B (Rev. 2650); Tue May 31 15:30:20 BST 2005

How do I check if these patches/patch kits have been applied on this server and where can I get these patch kits from?

How do I change kernel parameters on my Tru64 unix server required for the upgrade of Oracle 10g R2?

Thanks

INFO: --------------------------------------------------------------------------------
INFO: Checking Kernel
This is a prerequisite condition to test whether the minimum required kernel parameters are configured. Checking for exec_disable_arg_limit=1; found exec_disable_arg_limit=0. Failed <<<<
Checking for msg_size=32768; found no entry. Failed <<<<
Checking for max_objs=5120; found no entry. Failed <<<<
Checking for max_async_req=256; found no entry. Failed <<<< Checking for rdg_max_auto_msg_wires=0; found no entry. Failed <<<<
Checking for rdg_auto_msg_wires=0; found no entry. Failed <<<<
Checking for aio_task_max_num=8193; found no entry. Failed <<<<
Checking for fifo_do_adaptive=0; found no entry. Failed <<<<
Check complete. The overall result of this check is: Failed <<<<

INFO: --------------------------------------------------------------------------------
INFO: Checking Patches
This is a prerequisite condition to test whether the patches recommended for installing the product are available on the system.
INFO: Checking for T64KIT0020879-V51BB22-E-20031125OSF540; found Not found. Failed <<<<
Checking for T64KIT0019662-V51BB22-E-20030818OSF540; found Not found. Failed <<<<
Checking for T64KIT0021665-V51BB22-E-20040220OSF540; found Not found. Failed <<<<
Checking for T64KIT0021681-V51BB22-E-20040223OSF540; found Not found. Failed <<<<
Checking for T64V51BB22AS0002-20030415OSF540; found Not found. Failed <<<< Check complete. The overall result of this check is: Failed <<<<

INFO: --------------------------------------------------------------------------------

[Kapil Jha · Accepted solution]

hi
I think patches asked are from PK5.
I suppose you have to contact HP to find out how to get those patches......
you can check if they are installed by
setld -i|grep
regarding kernel parameter

change or add parameter by sysconfigdb command
or by direct edit /etc/sysconfigtab then rebuild the kernel....(doconfig)
and reboot..machine...........new kernel parameters are there ;)

[Hein van den Heuvel]

The variables and values listed appear to be the normal recomemnded 'sysconfig' settings for Oracle installations on Tru64.
Is this a memory-channel cluster using RDG for the internode Oracle stack?
Most missing params seem to be from the 'sysconfig -q rdg'.

Hein.

[Danesh Qureshi]

Is this a memory-channel cluster using RDG for the internode Oracle stack?

Sorry what is the memory-channel cluster using RDG ?

[Hein van den Heuvel]

Sotty, but it seems you are out of your league. You need to get other/better support involved. Either within your own company/support structure or through your support contract with HP.
Do not wait for help here.

Good luck,
Hein.

[Danesh Qureshi]

This is a steep learning curve for me on the Oracle side of things.

Looking at the man pages it is not a good idea to edit the sysconfigtab file directly but instead use the sysconfigdb.Can you just advise me on the syntax to add the kernel parameters in the /etc/sysconfigtab file.
Thank you.

[Hein van den Heuvel]

You are supposed to use the configuration db.
But personally, working mostly on benchmark/test systems, I only ever just vi /etc/sysconfigtab
Go to the bottom,
See how stuff is defined,
Clone to add yours,
reboot and doublecheck the new params have taken using sysconfig -q

is one of (vm, ipc, proc, rdg,...)

hth,
Hein.

[Danesh Qureshi]

we are not using memory-channel cluster using RDG for the internode Oracle stack.

Any reason why the Oracle 10gR2 installation is reporting failure:-

Checking for fifo_do_adaptive=0; found no entry. Failed <<<<
Check complete. The overall result of this check is: Failed <<<<

The attribute is already set to 0 see below:-

# sysconfig -q vfs | grep fif
fifo_do_adaptive = 0

Any help would be appreciated.

[Hein van den Heuvel]

>> we are not using memory-channel cluster using RDG for the internode Oracle stack.

Ah! You are learning the Tru64 game. Excellent.
;-)
>> Any reason why the Oracle 10gR2 installation is reporting failure:-
>Checking for fifo_do_adaptive=0; found no entry. Failed <<<<

Thruthfully... I wondered about that from the first post, but I was hoping it would magically clear up.
It reports 'no entry', where it should perhaps have indicated 'wrong value'.

You almost would have to decode the scripts to see how oracle is looking for this stuff. Is is parsing text output from sysconfig? reading the database? (hmmm... maybe you need to create one just for oracle!)? Is is doing 'table' calls from a program?

If I had a system to play with, then I would probably do a 'trace -f -o xxx' to capture all it is doing and search the syscall log for things like fifo_ or 'Checking' in a write buffer. Unfortunately I do not have a system, and do not really have the time either.


>>> sysconfig -q vfs | grep fif
fifo_do_adaptive = 0

Excellent test. Looks like you have it, and have it right.
It may be time for a call to Oracle support with the evidence now gathered.

Be sure to triple check OS and Oracle versions / patch levels for certification status first of course.

Regards,
Hein.