IWETHEY v. 0.3.0 | TODO
1,095 registered users | 0 active users | 0 LpH | Statistics
Login | Create New User
IWETHEY Banner

Welcome to IWETHEY!

IWETHEY Home / IWETHEY Board / Databases Forum / Oracle params:
Post #49,060
by broomberg
8/12/02 11:10:44 AM
Reply
New Oracle params:

Note: I do not use mutithreaded server.

set shmsys:shminfo_shmmax=0xffffffff
set shmsys:shminfo_shmmin=1
set shmsys:shminfo_shmmni=100
set shmsys:shminfo_shmseg=10

set semsys:seminfo_semmni=100
set semsys:seminfo_semmsl=500
set semsys:seminfo_semmns=1000
set semsys:seminfo_semopm=100
set semsys:seminfo_semvmx=32767
Post #49,076
by boxley
8/12/02 12:58:04 PM
Reply
New hmm they look ok
I would kick up "set semsys:seminfo_semmns=1000" to 2000 and here is the reference manual on those settings.
[link|http://docs.sun.com/?p=/doc/806-7009/6jftnqsj7&a=view|link]
thanx,
bill
."Once, in the wilds of Afghanistan, I had to subsist on food and water for several weeks." W.C. Fields
Post #49,114
by broomberg
8/12/02 7:50:19 PM
Reply
New Why?
* SEMMNS Set to the sum of the PROCESSES parameter for each Oracle
* database, adding the largest one twice, then add an additional 10 for each
* database. See "Configure the UNIX Kernel for Oracle8i" for an example of this
* formula.
*
Single Oracle instance, 150 max
processes, more likely I'm using 40
of them, with only 7 of them being
the shadow processes for my code.
Post #49,118
by a6l6e6x
Picture of a6l6e6x
8/12/02 8:08:40 PM
Reply
New Don't guesstimate, get actual count.
Assumptions are sometimes wrong. :)
Alex
"Television: chewing gum for the eyes." -- Frank Lloyd Wright
Post #49,124
by broomberg
8/12/02 8:45:26 PM
Reply
New Good point
And now you have me thinking.

I bet I max out on parallel query
slaves. I'll look into it.
Post #49,128
by broomberg
8/12/02 9:09:39 PM
Reply
New Nah, not even close
This process doesn't paralleize since
I restrict each to a specific paritition.

Here are ALL the oracle processes running
(other than shells) when I have 6 going
at the same time:

oracle 18883 1 0 19:21:19 ? 0:01 ora_smon_aig4
oracle 18875 1 0 19:21:18 ? 0:00 ora_pmon_aig4
oracle 18899 1 0 19:21:33 ? 0:00 /u01/app/oracle/product/9.2.0/bin/tnslsnr LISTENER -inherit
oracle 19120 1 8 20:54:58 ? 4:01 oracleaig4 (LOCAL=NO)
oracle 18889 1 0 19:21:19 ? 0:00 ora_d000_aig4
oracle 19153 1 9 20:56:54 ? 2:56 oracleaig4 (LOCAL=NO)
oracle 18893 1 0 19:21:19 ? 0:00 ora_d002_aig4
broom 19243 19241 0 21:04:15 pts/29 0:00 grep ora
oracle 19167 1 8 21:00:09 ? 1:44 oracleaig4 (LOCAL=NO)
oracle 18879 1 0 19:21:19 ? 0:00 ora_lgwr_aig4
oracle 19232 1 3 21:04:05 ? 0:03 oracleaig4 (LOCAL=NO)
oracle 18891 1 0 19:21:19 ? 0:00 ora_d001_aig4
oracle 19206 1 8 21:02:13 ? 0:43 oracleaig4 (LOCAL=NO)
oracle 19219 1 7 21:03:17 ? 0:18 oracleaig4 (LOCAL=NO)
oracle 18881 1 0 19:21:19 ? 0:00 ora_ckpt_aig4
oracle 18887 1 0 19:21:19 ? 0:00 ora_s000_aig4
oracle 18877 1 0 19:21:19 ? 0:00 ora_dbw0_aig4
oracle 18885 1 0 19:21:19 ? 0:00 ora_reco_aig4
     Tuning crazies - (broomberg) - (25) - Aug. 8, 2002, 12:05:29 AM EDT
         Tuning... a few suggestions - (folkert) - (4) - Aug. 8, 2002, 10:53:50 PM EDT
             Thanks - (broomberg) - Aug. 9, 2002, 08:24:16 PM EDT
             Reread - (broomberg) - (2) - Aug. 9, 2002, 08:46:55 PM EDT
                 Just did the re-read... - (folkert) - (1) - Aug. 10, 2002, 11:12:28 AM EDT
                     Gut feel is Oracle - (broomberg) - Aug. 10, 2002, 11:02:13 PM EDT
         what is your /etc/system file look like on the sun - (boxley) - (6) - Aug. 12, 2002, 10:10:51 AM EDT
             Oracle params: - (broomberg) - (5) - Aug. 12, 2002, 11:10:44 AM EDT
                 hmm they look ok - (boxley) - (4) - Aug. 12, 2002, 12:58:04 PM EDT
                     Why? - (broomberg) - (3) - Aug. 12, 2002, 07:50:19 PM EDT
                         Don't guesstimate, get actual count. - (a6l6e6x) - (2) - Aug. 12, 2002, 08:08:40 PM EDT
                             Good point - (broomberg) - (1) - Aug. 12, 2002, 08:45:26 PM EDT
                                 Nah, not even close - (broomberg) - Aug. 12, 2002, 09:09:39 PM EDT
         I think I got it - (broomberg) - (12) - Sept. 8, 2002, 04:35:48 PM EDT
             ooh ooh ooh I know!!!!! - (boxley) - (11) - Sept. 8, 2002, 10:16:11 PM EDT
                 No locking - (broomberg) - (10) - Sept. 9, 2002, 01:27:20 PM EDT
                     yer locking via nfs - (boxley) - (9) - Sept. 9, 2002, 06:48:07 PM EDT
                         I'm confused - (broomberg) - (8) - Sept. 9, 2002, 07:22:48 PM EDT
                             yer right but lockd isnt running on linux - (boxley) - (7) - Sept. 9, 2002, 09:15:23 PM EDT
                                 Yup, my issue - (broomberg) - (6) - Sept. 9, 2002, 11:22:10 PM EDT
                                     Piece o'crap Dell Dual Xeon - (broomberg) - (5) - Sept. 10, 2002, 12:00:39 PM EDT
                                         Re: Piece o'crap Dell Dual Xeon - (folkert) - (4) - Sept. 10, 2002, 03:57:33 PM EDT
                                             I knew I was throwing you a bone - (broomberg) - (3) - Sept. 10, 2002, 09:07:00 PM EDT
                                                 Mmmmmmmmm... Tasty! -NT - (folkert) - (2) - Sept. 10, 2002, 09:28:23 PM EDT
                                                     Update - (broomberg) - (1) - Sept. 11, 2002, 07:12:45 PM EDT
                                                         Good to know... - (folkert) - Sept. 11, 2002, 08:25:23 PM EDT

Many of the dead have finished being dead without the assistance of medical staff or fancy equipment. It would be good if people could help with the swift opening of bags that are wiggling and/or noisy.
102 ms