Simpler Navigation for Servers and Operating Systems - Please Update Your Bookmarks
Completed: a much simpler Servers and Operating Systems section of the Community. We combined many of the older boards, so you won't have to click through so many levels to get at the information you need. Check the consolidated boards here as many sub-forums are now single boards.
If you have bookmarked forums or discussion boards in Servers and Operating Systems, we suggest you check and update them as needed.
Operating System - Tru64 Unix
cancel
Showing results for 
Search instead for 
Did you mean: 

TRU64-V5.1B + oracle10G

edi_4
Advisor

TRU64-V5.1B + oracle10G

Hello! We are already running Oracle9 on our alpha ds25. Next week we are going to upgrade oracle to 10g.
Is there any system parameter to consider/tune?
Alpha DS25 has 3GB memory.

The current parameters are:
ubc_minpercent = 10
ubc_maxpercent = 100
ubc_borrowpercent = 20
vm_max_wrpgio_kluster = 32768
vm_max_rdpgio_kluster = 16384
vm_cowfaults = 4
vm_segmentation = 1
vm_ubcpagesteal = 24
vm_ubcfilemaxdirtypages = 4294967295
vm_ubcdirtypercent = 40
ubc_maxdirtywrites = 5
ubc_maxdirtymetadata_pcnt = 70
ubc_kluster_cnt = 32
vm_ubcseqstartpercent = 50
vm_ubcseqpercent = 10
vm_csubmapsize = 1048576
vm_ubcbuffers = 256
vm_syncswapbuffers = 128
vm_asyncswapbuffers = 4
vm_clustermap = 1048576
vm_clustersize = 65536
vm_syswiredpercent = 80
vm_troll_percent = 4
vm_inswappedmin = 1
vm_page_free_target = 768
vm_page_free_swap = 399
vm_page_free_hardswap = 12288
vm_page_free_min = 30
vm_page_free_reserved = 20
vm_page_free_optimal = 399
vm_swap_eager = 1
swapdevice = /dev/disk/dsk25b
vm_page_prewrite_target = 1536
vm_ffl = 1
ubc_ffl = 1
vm_rss_maxpercent = 100

proc:
max_proc_per_user = 1024
max_threads_per_user = 4096
per_proc_stack_size = 33554432
max_per_proc_stack_size = 33554432
per_proc_data_size = 268435456
max_per_proc_data_size = 4294967296
max_per_proc_address_space = 10737418240
per_proc_address_space = 4294967296
executable_stack = 0
autonice = 0
autonice_time = 600
autonice_penalty = 4
open_max_soft = 4096
open_max_hard = 4096
ncallout_alloc_size = 8192
round_robin_switch_rate = 0
sched_min_idle = 0
give_boost = 1
maxusers = 4096
num_wait_queues = 4096
num_timeout_hash_queues = 4096
enhanced_core_name = 0
enhanced_core_max_versions = 16
exec_disable_arg_limit = 0

ipc:
msg_max = 8192
msg_mnb = 16384
msg_mni = 64
msg_tql = 40
shm_max = 4278190080
shm_min = 1024
shm_mni = 1024
shm_seg = 256
sem_mni = 16
sem_msl = 25
sem_opm = 10
sem_ume = 10
sem_vmx = 32767
sem_aem = 16384
max_kernel_ports = 314912
ssm_threshold = 0
ssm_enable_core_dump = 1
shm_allocate_striped = 1
shm_enable_core_dump = 1

Thank you!


8 REPLIES
Michael Schulte zur Sur
Honored Contributor

Re: TRU64-V5.1B + oracle10G

Hi,

Oracle provides the required/suggested parameters with every version.
ubc_maxpercent = 100
for example is not a good idea, too high.

greetings,

Michael

https://metalink.oracle.com/metalink/plsql/f?p=130:14:3370285354129681264::::p14_database_id,p14_docid,p14_show_header,p14_show_help,p14_black_frame,p14_font:NOT,283751.1,1,1,1,helvetica#9
Hein van den Heuvel
Honored Contributor

Re: TRU64-V5.1B + oracle10G

There seem to be several values to tweak.
Please start by following the Oracle recommendations as suggested.

>> exec_disable_arg_limit = 0

I woudl set that to 1, for practical reasons, not for Oracle. It allows for deleting lots of files with a single wildcard and so on.

>> ssm_threshold = 0

You want SSM shared pages (or GH), not the basic flavor.
This needs to be set (8M is a typical value)

Hein
Michael Schulte zur Sur
Honored Contributor

Re: TRU64-V5.1B + oracle10G

Hi again,

am I assuming correctly that you are not running 5.1B?

Michael
edi_4
Advisor

Re: TRU64-V5.1B + oracle10G

I'll upgrade to 5.1B next week
Michael Schulte zur Sur
Honored Contributor

Re: TRU64-V5.1B + oracle10G

Hein,

I am a bit puzzled. According to docs Oracle does not support big pages for 10g. Is that correct and is there a reason on the side of HP for that?

greetings,

Michael
Hein van den Heuvel
Honored Contributor

Re: TRU64-V5.1B + oracle10G

Michael,

I deliberatly avoided using the 'big pages' thing in my reply. You had to bring that up huh?

Personally I like them a lot. It's simply the 'right way' to deal with the massive amounts of (vritual) memory typical needed to Oracle both for Code and for Data (SGA).

Bigpages have great performance potential, and they are so much more flexible than GH.

However, there have been enough reported problems with the initial implementation to force Oracle to be watchful of it being supported.

The support status has changed in the past and may or might not change in the future. It appears that Han P is 'most in the know', and I would blindly follow his advice if that is available. Certainly do not listen to me, as I have no official knowledge or point of view in this matter.

If you did listen to me anyway, then I would say to only use bigpages in relatively stable environments after extensive testing. A stable environment for me is for example one where Oracle instances are brought up and just stays up for good chunks of time, well before the users connect, with ample physical memory to go around. Where the number of instsances does not change much, and where the SGA size is stable. All of this is gutfeel, nothing you would want to rely on for your production box, but it worked in my benchmarks a few years back.


grins,
Hein.


Mark Poeschl_2
Honored Contributor

Re: TRU64-V5.1B + oracle10G

Another parameter commonly tweaked in Oracle installations is:

vfs: fifo_do_adaptive = 0

This is probably covered in some of the Oracle articles mentioned above as well.
Michael Schulte zur Sur
Honored Contributor

Re: TRU64-V5.1B + oracle10G

Hein,

how can I know what you do not say? ;-)
My last impression was that big pages were the choice to take. Another thing is new_wire_method which is recommended from 5.1B pk4. They do not explicitely say about pk3.

Michael