RH Linux vm.swappiness kernel parameter

The linux kernel parameter vm.swappiness affects the tendency of the OS to swap. It has a range from 0 to 100, with 0 causing swapping to occur only when needed and 100 to swap more aggressively. The default setting is 60.
There is little information available regarding an appropriate setting for this for Oracle RAC nodes. Do any of you have experience with swappiness? I have seen a few articles which recommend setting it to zero (0).
Best regards,
Bill

Our systems administrator have inquired about this parameter. I am not fixated on this and do not suffer from compulsive tuning disorder.
You, on the other hand, are obviously compulsive about too many of the wrong things. I'm sure everyone loves working with you...

Similar Messages

Could not update the kernel parameter when oracle 10g Installation.

Hi,
I tried to update ip_local_port_range in sysctl.conf file.
vi /etc/sysctl.conf
net.ipv4.ip_local_port_range=1024 65000
After changing the value i executed following command,
sysctl -p
Error : Unknown error 22 setting key 'net.ipv4.ip_local_port_range'
Please help

You didn't specify if you are on a linux based OS, it looks like you are, but it would be so kind on your side if you clearly specify the OS version and which DB version you are planning to install.
When I perform installations of DB10gR2 on OEL4/OEL5 or RHEL4/RHEL5 i issue this commands:
echo net.ipv4.ip_local_port_range = 1024 65000 >> /etc/sysctl.conf
Once the kernel parameter have been set, the kernel must be re-read with the /sbin/sysctl -p command.
~ Madrid
http://hrivera99.blogspot.com

Confirm that SHMMAX kernel parameter is set datastore size

Hi,
I'm getting below exception when i'm trying to create datastore. The error is complaining about failure to create shared memory segment.
I did go through the T-10 documentation and various post on this forum, and have made sure that the OS and the DSN settings are as per the recommended values. The values meets(or rather exceeds) the minimum values.
The values are based on below formulas:
kernel.shmmax (in bytes) = Should be >= PermSize + TempSize + LogBufMB +~50MB overhead
kernel.shmall (in pages) = Should be >= ceil ( SHMMAX / PAGE_SIZE)
SwapTotal (in KB) = Swap space should generally be at least 1.5x total physical memory size. It should be sufficient to allow allocation of a large enough shared memory segment
TempSize (in MB) = Should be >= 40 MB + ceiling( PermSize / 8 MB)
I have pasted below following things. Please, can any one help me resolve this issue.
- exception in ttmessg.log
- output of cat /etc/sysctl.conf
- sys.odbc.ini datastore configuration
- cat /proc/meminfo
- Swap space
- output of ttstatus
- output of ttversion
[oracle@ttnode1 info]$ ttdaemonadmin -stop
TimesTen Daemon stopped.
[oracle@ttnode1 info]$ ttdaemonadmin -start
TimesTen Daemon startup OK.
[oracle@ttnode1 info]$ ttisql "dsn=tt"
Copyright (c) 1996-2011, Oracle. All rights reserved.
Type ? or "help" for help, type "exit" to quit ttIsql.
connect "dsn=tt";
*836: Cannot create data store shared-memory segment, error 22*
The command failed.
Done
The exception in ttmesg.log
[oracle@ttnode1 info]$ tail -f ttmesg.log
2013-01-14 17:42:49.98 Warn: : 6533: TimesTen Daemon Release 11.2.2.4.0.tt started.
2013-01-14 17:42:59.45 Err : : 6533: TT14000: TimesTen daemon internal error: Error 22 creating shared segment, KEY 0x0200778b
2013-01-14 17:42:59.45 Err : : 6533: -- OS reports invalid shared segment size
2013-01-14 17:42:59.45 Err : : 6533: -- Confirm that SHMMAX kernel parameter is set > datastore size
2013-01-14 17:45:32.67 Warn: : 6533: request id check 'MKB9+MF7V7DybTjglCEC/10d5/N7FCEFtayvZF4a9xQ=' failed (-5)
2013-01-14 17:47:18.45 Warn: : 6533: request id check 'u6GF7bBtmRQCXKYMAWa7LhwCv1GtIG8HXdW0GLRLHEU=' failed (-5)
2013-01-14 17:47:41.62 Warn: : 6533: request id check 'nNAzUc/plNUStN85qP9htbsz/E5psODMEeRYLmE/n50=' failed (-5)
2013-01-14 23:15:34.58 Warn: : 8590: TimesTen Daemon Release 11.2.2.4.0.tt started.
2013-01-14 23:15:43.44 Err : : 8590: TT14000: TimesTen daemon internal error: Error 22 creating shared segment, KEY 0x02008294
2013-01-14 23:15:43.44 Err : : 8590: -- OS reports invalid shared segment size
2013-01-14 23:15:43.44 Err : : 8590: -- Confirm that SHMMAX kernel parameter is set > datastore size
2013-01-14 23:16:11.23 Warn: : 8590: request id check 'v4TJ+ZyP3KRvGxfkguLgxNyRUdtfw/MMFrvR2tWMUBk=' failed (-5)
2013-01-14 23:18:01.34 Warn: : 8590: request id check 'p4DVyvW75Fjlo3dQ5f8OFbk4qpolqj8gFyQhqfW4ekQ=' failed (-5)
2013-01-14 23:19:50.28 Warn: : 8590: request id check 'DPQ8gSYcljQtkuvx6ccI5a5OGcjXklItV4Mj3Cf0DmA=' failed (-5)
2013-01-14 23:41:01.97 Warn: : 8820: TimesTen Daemon Release 11.2.2.4.0.tt started.
*2013-01-14 23:41:10.10 Err : : 8820: TT14000: TimesTen daemon internal error: Error 22 creating shared segment, KEY 0x02008294*
*2013-01-14 23:41:10.10 Err : : 8820: -- OS reports invalid shared segment size*
*2013-01-14 23:41:10.10 Err : : 8820: -- Confirm that SHMMAX kernel parameter is set > datastore size*
2013-01-14 23:42:20.31 Warn: : 8820: request id check 'sHRtjK5AUKI/RYPN6Mf6ivOPeGVFysbXT4eU9dfBZPw=' failed (-5)
2013-01-14 23:43:24.14 Warn: : 8905: TimesTen Daemon Release 11.2.2.4.0.tt started.
2013-01-14 23:43:29.64 Err : : 8905: TT14000: TimesTen daemon internal error: Error 22 creating shared segment, KEY 0x02008294
2013-01-14 23:43:29.64 Err : : 8905: -- OS reports invalid shared segment size
2013-01-14 23:43:29.64 Err : : 8905: -- Confirm that SHMMAX kernel parameter is set > datastore size
Here is the OS setting.
*[oracle@ttnode1 ~]$ cat /etc/sysctl.conf*
# Kernel sysctl configuration file for Red Hat Linux
# For binary values, 0 is disabled, 1 is enabled. See sysctl(8) and
# sysctl.conf(5) for more details.
# Controls IP packet forwarding
net.ipv4.ip_forward = 0
# Controls source route verification
net.ipv4.conf.default.rp_filter = 1
# Do not accept source routing
net.ipv4.conf.default.accept_source_route = 0
# Controls the System Request debugging functionality of the kernel
kernel.sysrq = 0
# Controls whether core dumps will append the PID to the core filename
# Useful for debugging multi-threaded applications
kernel.core_uses_pid = 1
# Controls the use of TCP syncookies
net.ipv4.tcp_syncookies = 1
# Controls the maximum size of a message, in bytes
kernel.msgmnb = 65536
# Controls the default maxmimum size of a mesage queue
kernel.msgmax = 65536
# Controls the maximum shared segment size, in bytes
kernel.shmmax = 68719476736
# Controls the maximum number of shared memory segments, in pages
kernel.shmall = 4294967296
Here's the DSN setting
* Datastore configuration in sys.odbc.ini file*
*[oracle@ttnode1 info]$ grep -v ^# sys.odbc.ini | sed '/^ *$/d'
[ODBC Data Sources]
TT=TimesTen 11.2.2 Driver
[TT]
Driver=/home/oracle/TimesTen/tt/lib/libtten.so
DataStore=/home/oracle/TimesTen/tt/info/DStore/itt_11224
DatabaseCharacterSet=AL32UTF8
PermSize=128
TempSize=96
ReplicationApplyOrdering=0
ReplicationParallelism=4
LogBufParallelism=4
LogBufMB=32
LogFileSize=32
LogFlushMethod=2
*[ice@ttnode1 ~]$ cat /proc/meminfo*
MemTotal: 3865624 kB
MemFree: 2287936 kB
Buffers: 9744 kB
Cached: 189340 kB
SwapCached: 62588 kB
Active: 739244 kB
Inactive: 796816 kB
HighTotal: 0 kB
HighFree: 0 kB
LowTotal: 3865624 kB
LowFree: 2287936 kB
SwapTotal: 2031608 kB
SwapFree: 1755480 kB
Dirty: 316 kB
Writeback: 0 kB
AnonPages: 1310800 kB
Mapped: 24260 kB
Slab: 12952 kB
PageTables: 8712 kB
NFS_Unstable: 0 kB
Bounce: 0 kB
CommitLimit: 3964420 kB
Committed_AS: 2886040 kB
VmallocTotal: 34359738367 kB
VmallocUsed: 1616 kB
VmallocChunk: 34359736599 kB
HugePages_Total: 0
HugePages_Free: 0
HugePages_Rsvd: 0
Hugepagesize: 2048 kB
*[ice@ttnode1 ~]$ grep SwapTotal /proc/meminfo*
SwapTotal: 4128752 kB
Earlier the swap space was 2031608 KB, but snice it shuold be 1.5 times the physical memory, i bumped it up to its current value
*[oracle@ttnode1 info]$ ttstatus*
TimesTen status report as of Mon Jan 14 17:47:32 2013
Daemon pid 6533 port 53398 instance tt
TimesTen server pid 6550 started on port 53399
Accessible by group oinstall
End of report
*[oracle@ttnode1 info]$ ttversion*
TimesTen Release 11.2.2.4.0 (64 bit Linux/x86_64) (tt:53398) 2012-09-24T08:28:05Z
Instance admin: oracle
Instance home directory: /homankse/oracle/TimesTen/tt
Group owner: oinstall
Daemon home directory: /home/oracle/TimesTen/tt/info
Thanks,
Ved

Hi,
On continuing my debugging steps, if checked the limits for IPC facility.
ipcs -m -l
------ Shared Memory Limits --------
max number of segments = 4096
max seg size (kbytes) = 32768
max total shared memory (kbytes) = 8388608
min seg size (bytes) = 1
This says that "max number of segments = 4096", where as the kernel parameter is defined as "kernel.shmall = 4294967296".
Am i correct in understanding that "max number of segments" is same as "kernel.shmall" and the values should be same.
Also, it says that "max seg size (kbytes) = 32768" ie around 32MB, where as the kernel parameter is defined as "kernel.shmmax = 68719476736" i.e around 64BG.
and since, the DSN is requesting to create shared memory segment of 128MB, which is greater that 32MB(as per max seg size (kbytes) = 32768), Timesten is throwing the error.
Thanks everyone for helping me out.
_ved

Maxuproc kernel parameter

Hi,
I need to modify maxuproc kernel parameter in linux (Red hat).
Please assist.
Thanks,

This is a Oracle forum, not a linux forum. Please close that post and open a new one in the correct forum.

Kernel parameter : Share memory.

Hi All,
I have some question about setup linux and oracle 8i. How to
change kernel parameter such as SHMMNI ?.
Look forward to hearing from you soon.
Best Regards,
Pichet R.
null

Pichet Ratanayant (guest) wrote:
: Hi All,
: I have some question about setup linux and oracle 8i. How to
: change kernel parameter such as SHMMNI ?.
: Look forward to hearing from you soon.
: Best Regards,
: Pichet R.
You need to recompile the kernel adjusting the shared memory
parameters
the file you need is /usr/src/linux/include/asm/shmparam.h
Alan
The recompilation is simple but difficult to describe.
null

UEFI boot problems with Linux 3.7 kernels

According to this thread in the Arch Linux forums, some people are having problems booting some versions of the Linux 3.7 kernel. Here are the numbers:
3.7.3 works for some, but not others
3.7.4 works for some, but not others
3.7.5 works
3.7.6 does not
3.7.7 works
3.7.8 does not
The problem occurs with both the Gummiboot and rEFInd boot managers, and has so far only been reported on Lenovo hardware. I have a ThinkPad W530, and can confirm that on Gummiboot, 3.7.6 does not work, whereas 3.7.5 and 3.7.7 does work. It looks like a Lenovo-related issue to me. It has only been reported on Arch Linux as far as I know. Has anyone else encountered this problem?

Thanks for your reply. Yes, the cursor did blink in the top left corner.
I gather from your comments that, if I install a new HD, I will be able to re-install the Windows 7 32-bit O/S from the Product Recovery DVD GMR410014EN0, which was supplied by Toshiba with the laptop and it would also follow, if that is true, that I would be able to Format (Boot n' Nuke) the existing HD and re-install the O/S onto it before I get my friend to lay out cash for a new HD.
I would appreciate your confirmation that the above conclusions are correct as I was under the, hopefully incorrect, impression that the Recovery Disk could only be used to repair an existing system. I have the Microsoft licence number as it is printed on the bottom of my friends machine.
Thanks for your time and I look forward to hearing from you.

Kernel parameter tuning in Solaris 10

Hi Guru's
Can you any one help me to find the some good docs for Solaris kernel tuning. How to change the kernel parameter online?
How to do kernel tuning in online and what is the purposre of /etc/system and sysdef command.
How I can search for exisiting kernel value.
Thanks in advance.
Jimbe

There is interesting info at [Solaris Internals|http://www.solarisinternals.com/wiki/index.php/Solaris_Internals] .
have a good day,
Glen

Rlim_fd_max Kernel Parameter on Solaris 9

I am trying to install Oracle Application Server 10g Release 2 on Solaris 9. The installation errors out on the "Select Configuration Options" screen with the following error:
"Oracle Application Server 10g Web Cache requires the Kernel Parameter rlim_fd_max to be set to a value of 65536 or higher. Please set rlim_fd_max to 65536 or higher, reboot the machine and restart the installation".
I checked the value of the set rlim_fd_max Kernel Parameter prior to starting the install and it is set to 65536. Does anyone know why I'm getting this error? Thanks in advance!

I was able to get an answer by opening a SR with Metalink. The problem was that even though the rlim_fd_max default on Solaris 9 is 65536, it must be in the /etc/system file. Must be the installer looks at the file to determine that it is set correctly. So, adding set rlim_fd_max=65536 and set rlim_fd_cur=1024 to the /etc/system file fixed the problem.

Oracle 10g on solaris 10 : kernel parameter warning

I am installing oracle 10g on Solaris 10.
I have a warning on kernel parameter missing (shminfo_shmmax for example) during the system check.
I understood that with the 10 version of solaris these parameters are replaced by ressources control and I have checked them. They are correctly setted.
Need I do something before continue the install to release the warning or can I bypass this issue and continue the install as I have no warning ?
Thanks for your answers.
Sorry for my english.

You are right. Since Solaris 10 kernel parameters settings in /etc/system is obsolete and some parameters are deprecated.
The steps how to configure required parameters you can find in my paper (step #1.): http://ivan.kartik.sk/oracle/install_ora10gR2_solaris.html
I bypass this issue and continue the install as I have no warning You can safely bypass this issue during installation process but you should set these parameters after installation.

Determining OS Kernel parameter settings in HP ITANIUM

Hi
Can any one help me out to determine the depedant OS kernel parameters settings in HP IA-64 for Oracle database 10.2.0.4 after increasing the processes parameter value in the database level .
As after increasing the processes parameter to 10000 , we are getting n number of issues related to one or the other OS kernel parameter , so we want to take one single downtime and settle all the issues.
Regards,
Phani Kumar P

Hi
Acutally we already increased the processes parameter to 10000 by modifying nproc value to 30000 , now we are receiving issues respect to database as below and inturn there is problem in getting connected to the database giving ora-12502 error
Errors in file /ora/newdump/ACEPROD/udump/aceprod1_ora_4712.trc:
ORA-00603: ORACLE server session terminated by fatal error
ORA-27504: IPC error creating OSD context
ORA-27300: OS system dependent operation:bind failed with status: 227
ORA-27301: OS failure message: Can't assign requested address
ORA-27302: failure occurred at: sskgxpcre3
kkjcre1p: unable to spawn jobq slave process
Tue Jan 5 18:28:57 2010
Errors in file /ora/newdump/ACEPROD/bdump/aceprod1_cjq0_20387.trc:
So now we want to fix the issues by setting optimal values for OS Kernel parameter for NPROC value equals 30000.
Regards,
Phani Kumar p
+919948677317

Replace Linux with Hurd kernel

Hi
Can i Replace Linux with Hurd kernel?

Lone_Wolf wrote:
hayderctee,
there is an Archlinux Hurd project, http://www.archhurd.org/ but i have no idea how usable it is.
I would say that's dead, with no updates to any packages since March last year...

What is the resource control corresponding to each kernel parameter

Hi
I am trying to prepare (i.e. tune) a Solaris 10 instance for an Oracle 10g install using a Solaris 10 project.
I am new to using Solaris 10 project resource controls so I might get some of this totally wrong.
I know all of the "old" kernel parameter names: what each represents and its' required value and I know some of the corresponding resource controls but not all of them.
That is I do not know the resource control corresponding to each and all of the kernel parameters.
e.g. I know:
semsys:seminfo_semmsl == process.max-sem-nsems == The maximum number of sempahores that can be in one semaphore set
shmsys:shminfo_shmmax == project.max-shm-memory == The maximum size(in bytes) of a single shared memory segment
shmsys:shminfo_shmmni == project.max-shm-ids == The number of shared memory identifiers
But I do not know, for example, what resource control corresponds to:
semsys:seminfo_semmns == The number of semaphores in the system
Is there a list somewhere of the resource control corresponding to each of the kernel parameters ?
I know I can set some kernel parameters in /etc/system but that should only be used for system - wide kernel parameters, right ?
I know there are project, process, task and zone classes of resource controls.
Can a resource control of any and all of these types be set in a project ?
Are zone resource controls equivalent to system-wide resource controls ?
Are there corresponding resource controls to all kernel parameters e.g. such as noexec_user_stack ?
I have set noexec_user_stack in /etc/system.
And finally - just to be sure - using a project makes the resource control values permanent, right ?
Thanks
Brett.

Hi;
Please check below which could be helpful for your issue:
Kernel setup for Solaris 10 using project files [ID 429191.1]
Regard
Helios

Cannot install oracle vpn on linux getting error kernel-sourcecode is neede

cannot install oracle vpn on linux getting error kernel-sourcecode is needed by gds_cisco_vpn_client-2.0-llrhel4.i386.. i already installed kernel-devel still it get the same error. pls help.

1. kernel (.rpm)
2. kernel-devel (.rpm) - containing only Makefiles and some kernel header files
3. kernel source rpm (src.rpm) - source for rebuilding the kernel
4. kernel-sourcecode (.rpm) - source-tree for developing custom kernels
are four different packages.
If you are running OEL-4 (2.6.9 kernel) and you have an ULN account, you can download kernel-sourcecode rpm from ULN. There is no kernel-sourcecode for OEL-5 available on ULN site.
NJ

Kernel parameter for zones

Hi everybody,
I trying to install Oracle Application server in solaris native zone. I am getting the following error " Oracle application Server 10g Web Cache requires the kernel parameter rlim_fd_max to be set to a valuse of 65536 or higher. Please set rlim_fs_max to 65536 or higher , reboot the machine asn restart the installation.
but the value is already set to same value
$ ulimit -Hn
65536
Any idea how set this value for zone ?
Thanks
John

Sorry, doubled response
Edited by: brusell on Jun 10, 2010 4:24 AM

How to verify RTSJ use linux real time kernel?

Hi,
I'm complete newbie into this area and I'm trying out the RTSJ 2.1 beta Linux. I'm using this on Ubuntu with the linux real time kernel. Is there anyway to find out whether the real time kernel is being actually used or in other words is there a way to find out whether RTSJ works fine with the real time kernel on Ubuntu? I see that the programs get compiled and run irrespective of whether I use the real time kernel or the generic kernel.
Thanks,
Vidura

Hi,
I would assume, like other distributions, that you either boot the real-time kernel or you don't. uname should show you what you are running, but you'd have to ask the Ubuntu folk what you should see for the RT kernel.
To see if you are benefiting from real-time, you need to run a RTSJ app that tracks deadline misses or measures latency/jitter. Try it on the non-real-time kernel and the real-time one and see what you get. Some of the examples in the "Getting Started" guide should be usable for these purposes.
David Holmes

RH Linux vm.swappiness kernel parameter

Similar Messages

Maybe you are looking for