OFA for 11g.. Are multiple mount points (ie 2 or more) still recommended

Folks
Client has a Clariion SAN (the latest generation) They are pushing strongly to provide 2 mount points for all Oracle databases as a space saving initiative. Is performance affected with this configuration given the improvements in spindle speeds and controllers.
Is going to 4 mount points 2 for Oracle and 2 for volume data and indexing that much of difference.
Peter Johnson
Oracle ACS

Similar Messages

Split datafiles in multiple mount points

Hi,
I would like to split my Oracle Applications datafiles in multiple mount points to improve the performance, but I didn�t find any document(best pratices) talking about wich tablespaces should be put on the same or different mount point.
For instance:
I have mount points data1, data2 and data3. How to split the "data" datafiles into this 3 mount points.
I already split data, index, logs and system in differents mount points.
Anyone had already done something like that or have any document/information about it.
Thanks,
Marcelo.

Hi,
Its an simple concept.
You got to move the datafiles to respective mount points.
Recreate the control file based on the datafiles that locates.
Thats it.
Thanks and Regards
A.Riyas

BPC for Excel: Are multiple key ranges really possible?

Hi Experts,
in a training document I found to possibility of multiple key ranges, to make an input schedule sheet more attractive or insert rows with excel formulas etc.
For example: ColKeyRange: EVRNG(H13:J13;M13:O13)
RowKeyRange: EVRNG(G15:G20;L15:L20)
In my case I want the balance sheet separated. On the left side asset positions, on the right side liability positions (in columns for bot: Previous Year | Actual | Plan
If I define my input schedule like that it gets messed up. No figures, weird formatting etc. Have you implemented an input schedule with multiple key ranges and did it work ok?
Thank you,
Cora

Hi Cora,
Yes, multiple key ranges do work, but it is a bit complicated as Dzmitry suggests. Your example row expansion will not work, as the row expansion needs to remain in the same columns for each part of the expansion. (In other words, EVRNG(G15:G20;L15:L20) is wrong but EVRNG(G15:G20;G25:G30) is ok.) To achieve your result, Dzmitry is correct that you will probably need to use two EVDREs.
That said, multiple expansions work as follows:
If you are not using an expansion, or you use NOEXPAND in your columns or rows, you can use a multiple key range for formatting purposes and simply hardcode in the members that you wish to see in each key range.
If you want to see different expansions in each key range, then you need to use a special expansion syntax. To take your example, say you have a single dimension in your column with column range defined EVRNG(H13:J13;M13:O13). You could have your column expansion on the category dimension be "PRIORYR,ACTUAL,PLAN|PLAN,ACTUAL,PRIORYR", giving you the same set of categories reversed in each column.
Ethan

How to mount a device/filesystem at multiple mount points

Hi, my device /dev/disk1s1 is auto mounted at /Volumes/UNTITLED.
i want to mount the device again on different path like /Volumes/temp.
so that i can have same device mounted at two different place.
When i type " mount /dev/disk1s1 /Volumes/temp" in Terminal , i get the following message : /dev/disk1s1 on /Volumes/temp: Resource busy
Can anyone help me in this?

I know you can do it with linux
Searching the Darwin mailing list revealed this thread;
http://lists.apple.com/archives/darwin-development/2001/Jan/msg00605.html
Seems it's not possible, but I didn't dig deep, sorry.

R3load cannot export more than 100 mount points for Oracle?

We have a DB with more than 390 sapdata### mount points (HPUX-PARISC). They are truly mount points, NOT directories under some other mount points.
After export using R3load (i.e NO DB-specific ), the keydb.xml generated for import has only from sapdata1 to sapdata100.
Is there any limit here for R3load?
Thanks!

R3load doesn't copy the filesystem structure but unloads the content of the database after having checked the size of it and then distributes it across the files.
Why do you have so many different mountpoints? Is there a technical reason behind? Just curious...
Markus

Mounting multiple directories with same name on different severs to a single mount point on another server

We have a requirement where in we have multiple solaris servers and each solaris server has a directory with the same name.
The files in these directories will be different.
These same name directories on multiple severs has to be mounted to a single directory on another sever.
We are planning to use NFS, but it seems we can not mount multiple directories with same name on different severs to a single mount point using NFS, and we need to create multiple mount points.
Is there any way we can achieve this so that all the directories can be mounted to a single mount point?

You can try to mount all these mount points via NFS in one additional server and then export this new tree again via NFS to all your servers.
No sure if this works. If this works, then you will have in this case just an additional level in the tree.

How to create Mount Point in Linux

Hi,
I m new to Linux OS, want to install Oracle EBS R12 on OEL 5.6 . I have downloaded s/w and installed all required packages for OEL.
Now i want to create Staging directory at location /u01/StageR12.
Can anyone guide me for how to create Mount point /u01 and how to assign space of 53 GB to it.... ?
Reply awaited...

You're on the right path.
Keep in mind though what you need partition wise just for o/s install. You should have at least 3 partitions:
- a ?? GB partition for the / (root) mount (the size depends on what you will install and how much space you will need for running that s/w)
- a 128MB or larger partition for the /boot mount
- a ?? GB partition as swap space mount (typically rule of thumb is 2x RAM for 4GB and lower, else around 25% of total RAM)
I would not create separate partitions for the old-style +/u01+ mount points for Oracle. That part of the OFA (Oracle Flexible Architecture) standards are old and pre-ASM and pre- automated Oracle managed data files.
You need space for installing an Oracle Home. I do not see the need for a dedicated mount point for that. I use the root file system.
You need space for the actual database. If you use Oracle ASM, then I would create raw partitions (across disks) to be used for database storage. These will not be mounted or formatted. Instead these will be assigned to an ASM diskgroup (and striped by default) and used for database storage.
If you do not use ASM, then you need a cooked file system for storage of database file. In that case separate mount point(s) make sense. Also, it makes sense not to manually apply the old OFA standards, but instead use Oracle managed database files. In that case you need a single mount point (e.g./oracledb) - and set that as the base directory for Oracle to use for the database. Oracle will create an OFA compliant database directory tree and files under that base directory/mount point.
Keep in mind that the more mount points you have for the database, the more difficult your job becomes to manage storage. The easiest is a single mount point for the database and using Oracle to manage the OFA compliant side for you on that mount point.
Also, you cannot and should not attempt some kind of manual striping layout of the database across multiple mountpoints. This may have made some sense back 10+ years ago. It no longer does.

Logical Disk (Mount Point) Performance Counters

We're currently running SCOM 2012 SP1, Management Servers and Agents are all at UR8
The Windows Server Operating system management pack is at 6.0.7292.0
We're successfully collecting performance metrics which are not mount points but the performance metrics for mount points are missing. We've enabled the discovery for mount points.
The error that I'm seeing on the agent machines is:
In PerfDataSource, could not resolve counter instance LogicalDisk, Avg. Disk Write Queue Length, G:\Archive\. Module will not be unloaded.
One or more workflows were affected by this.
Workflow name: Microsoft.Windows.Server.2008.LogicalDisk.AverageDiskWriteQueueLength.Collection
Instance name: G:\Archive\
Instance ID: {EE651C3D-A5B8-E5C2-459E-8B6C27F8B813}
We used to be able to collect this performance data but now it's failing. Does anyone know how to resolve this?

Hi!
There was a new MP version fixing Mount Point monitoring online earlier this year but it has been pulled since there where some bugs.
Expect the new version available within the next couple of weeks. I assume your issue will be fixed there.
Cheers,
Patrick
Please remember to click “Mark as Answer” on the post that helped you.
Patrick Seidl (System Center and Private Cloud)
Website: http://www.syliance.com
Blog: http://www.systemcenterrocks.com

Exlucde mount point from host metric collection

Customer is using 10.2.0.3 GC, Can a mount point on a host be excluded from metric collection? If so how? Is it possible to specify multiple mount point to exclude from metric collection.

I have seen documentation that says to set the warning and critical thresholds to null. You can give that a try but it isn't working for me. Only difference is I am trying to exclude an UNDO TS.

Mount point containing the redologs

We were told Oracle typically writes 512b redolog records to the redologs. If the mount point is created with a blocksize > 512b the remaining space is wasted.
For example, if the mount point is created with a 4K blocksize, most likely 3.5K of the block is wriiten with spaces or null characters. This would than cause ~75% of the redolog to contain spaces or non - redolog information.
should the mountpoint containing the redologs be created with the 1k block size?
Regards

user13484479 wrote:
We were told Oracle typically writes 512b redolog records to the redologs. If the mount point is created with a blocksize > 512b the remaining space is wasted.
For example, if the mount point is created with a 4K blocksize, most likely 3.5K of the block is wriiten with spaces or null characters. This would than cause ~75% of the redolog to contain spaces or non - redolog information.
should the mountpoint containing the redologs be created with the 1k block size?
Oracle uses the physical sector size as the block size of the redo log files and uses the same unit for doing the IO as well. If you want to change the physical sector size, that is only possible if the physical disk of yours would be supporting the size more than 512b which is still not happening by the storage vendors AFAIK. If your vendor confirms that its possible to have a physical block size more larger than 512b and of 4kb, than only you would be able to do so. And if your storage allows you to go more than 512 bytes, you have the option from 11.2 (only) to change teh block size of the redo log file to 4kb .
Besides the links given by Girish, I wrote about it some time back on my blog too. Have a read,
http://blog.aristadba.com/?p=124
HTH
Aman....

Need charting help - multiple data points between major ticks

Hi,
I am trying to create an AreaChart in which the data shows a
consumption rate over the course of a week. The major points along
the x-axis will be Monday, Tuesday, ... , Sunday (the days of the
week). However, there are multiple data points between each day. I
only want the days of the week to show on the axis but I want to
plot all of the data points. How would I go about accomplishing
this task?
Thanks in advance,
Brent Schooley

Thanks, it is getting closer now. My only problem now is that
the data points do not line up correctly with the tick marks. I've
look through the documentation and I can't figure out what to do to
ensure this happens. Here are the relevant portions of my code:
[Bindable]
public var energyDataAC:ArrayCollection = new
ArrayCollection( [
{date: "2007, 7, 27, 0", saved: 0},
{date: "2007, 7, 27, 12", saved: 50},
{date: "2007, 7, 28, 0", saved: 35},
{date: "2007, 7, 28, 12", saved: 0},
{date: "2007, 7, 29, 0", saved: 0},
{date: "2007, 7, 29, 12", saved: 46},
{date: "2007, 7, 30, 0", saved: 39},
{date: "2007, 7, 30, 12", saved: 0},
{date: "2007, 7, 31, 0", saved: 0},
{date: "2007, 7, 31, 12", saved: 44},
{date: "2007, 8, 1, 0", saved: 45},
{date: "2007, 8, 1, 12", saved: 0} ]);
public function myParseFunction(s:String):Date {
// Get an array of Strings from the comma-separated String
passed in.
var a:Array = s.split(",");
// Create the new Date object.
var newDate:Date = new Date(a[0],a[1],a[2],a[3]);
return newDate;
<mx:AreaChart id="Areachart" height="100%" width="100%"
paddingLeft="5" paddingRight="5"
showDataTips="true" dataProvider="{energyDataAC}" >
<mx:horizontalAxisRenderer>
<mx:AxisRenderer styleName="myAxisStyle"/>
</mx:horizontalAxisRenderer>
<mx:horizontalAxis>
<mx:DateTimeAxis alignLabelsToUnits="true"
dataUnits="hours" labelUnits="days" dataInterval="12" interval="1"
parseFunction="myParseFunction"/>
</mx:horizontalAxis>
<mx:series>
<mx:AreaSeries yField="saved" xField="date"
form="segment" displayName="Energy Saved"/>
</mx:series>
</mx:AreaChart>
<mx:Legend dataProvider="{Areachart}"/>
Any ideas?
Thanks,
Brent

Why does Arch suggest /boot for $esp mount point?

Hi all,
Something has been bugging me, and I can't figure out online why this is the case. Arch always suggests to mount the $esp on /boot, and by reading the Beginner's Guide/Installation wiki pages (and forums), it looks like the reasons are:
- gummiboot upstream expects it there
- pacman can only update kernels in /boot
Other distributions suggest to mount the $esp on /boot/efi, and this has the good effect, that all files of distributions are located in subfolders ($esp/EFI/opensuse/, for example), and are not littering the root $esp folder. If every distribution did what Arch does, then on a multi-distro the root folder of the $esp would have every single kernel/fallback img.
I noticed the FHS doesn't have a standard mount point/folder for the $esp, either.
Is there another reason that Arch suggests this?
Thanks

Gummiboot's method only works if distros keep the /boot folder free of files, and use directories to cleanly categorise kernels, etc. I have yet to find a distribution that does this.
"Works" is relative here. Gummiboot doesn't do things the way you want; that's fine, because a) you don't use it; and b) you can always organize your EFI images as you see fit. Gummiboot will still accomplish its intended task: booting a working Linux kernel from an EFI image. The Gummiboot Arch wiki page states that the ESP must be mounted at /boot for it to work, but that's it. The UEFI wiki page doesn't mention mounting at boot at all---only to /boot/efi. The EFISTUB page mentions that if you don't mount to /boot, you need to mount the ESP somewhere and manually copy the images to the ESP. The GRUB wiki page doesn't make statements on where the ESP should be mounted, but does make it clear that GRUB needs to be told where the ESP is mounted during installation. The Beginner's Guide does recommend mounting the ESP to /boot, and then links to another page with more information.
Reading the Beginner's Guide without following the links may give the impression that mounting to /boot is something of a requirement, but it depends on the use case.You have four options when booting from UEFI: rEFInd, Gummiboot, GRUB and EFISTUB. The first requires the ESP to be mount at /boot/efi, like it or not. The second requires it to be mounted at /boot, like it or not. The latter two have no strict requirements of where the ESP be mounted, but mounting anywhere other than /boot requires extra work on the part of the user. Yes, if you mount the ESP to /boot and run multiple distributions then the kernel images are still going to clutter the /boot directory, which is exactly what happens if you share a /boot partition between distributions no matter what (and as I alluded to earlier, there are cases in which this is necessary). UEFI has nothing to do with that, and it's one reason that each distribution has its own naming conventions for its kernel image (a Linux kernel built from source will always be called "bzImage" regardless of the distribution its built on). Arguably, none of that matters unless you're manually moving and deleting kernel images and get confused about which one is which---a case that almost never arises nowadays, and should never arise for the typical desktop user.

Under Options , Applications There are multiple entries for the same item (Acrobat Document). how can I delete the duplicates?

Under Options, Applications there are multiple entries for the same item. (Adobe Acrobat) . Three show Adobe Acrobat Reader 9.3 and the other shows (ask) or (save file) with no ability to select Acrobat Reader . Is there a way to edit this list to remove the duplicates and
the incorrect entry? Adobe installed a very quick update today.

They are all different. Hover your mouse pointer over each of the "Content Type" descriptions and you should see a "tooltip" to see that each has a different description.
You need to update your plugins. It is important to keep them updated due to continuing security fixes and improvements in those plug-ins:
* Adobe Shockwave for Director Netscape plug-in, version 11.5 (you '''may''' need to update)
* Shockwave Flash 10.1 r53
* Next Generation Java Plug-in 1.6.0_20 for Mozilla browsers
#Check your plugin versions: http://www.mozilla.com/en-US/plugincheck/
#*'''Note: plugin check page does not have information on all plugin versions'''
#'''Update Shockwave for Director'''
#*NOTE: this is not the same as Shockwave Flash; this installs the Shockwave Player.
#*Use Firefox to download and SAVE the installer to your hard drive from the link in the article below (Desktop is a good place so you can find it).
#*When the download is complete, exit Firefox (File > Exit)
#*locate and double-click in the installer you just downloaded, let the install complete.
#*Restart Firefox and check your plugins again.
#*'''Download link and more information''': http://support.mozilla.com/en-US/kb/Using+the+Shockwave+plugin+with+Firefox
#'''Update Shockwave Flash'''
#*Use Firefox to Download and SAVE to your hard drive from the link in article below
#*SAVE to your Desktop so you can find it
#*After download completes, close Firefox
#*Click on the file you just downloaded and install
#**Note: Vista and Win7 users may need to right-click the installer downloaded and choose "Run as Administrator"
#**Note: Most browsers other than IE will also get updated with this one download
#**Note: To update IE, same procedure '''but use IE''' to go: http://fpdownload.macromedia.com/get/flashplayer/current/install_flash_player_ax.exe
#*After installation, restart Firefox and check your version again.
#*'''Download link and other information''': https://support.mozilla.com/en-US/kb/Managing+the+Flash+plugin#Updating_Flash
#* Also see: http://support.mozilla.com/en-US/kb/Installing+the+Flash+plugin
#* Also see (if needed): http://kb2.adobe.com/cps/191/tn_19166.html#main_Uninstall
#'''Update Java:'''
#* Download and update instructions: https://support.mozilla.com/en-US/kb/Using+the+Java+plugin+with+Firefox

How to determine the mount point for directory /tmp ?

Folks,
Hello. I am installing Oracle 11gR2 RAC using 2 Virtual Machines (rac1 and rac2 whose OS are Oracle Linux 5.6) in VMPlayer and according to the tutorial
http://appsdbaworkshop.blogspot.com/2011/10/11gr2-rac-on-linux-56-using-vmware.html
I am installing Grid infrastructure. I am on step 7 of 10 (verify Grid installation enviroment) and get this error:
"Free Space: Rac2: /tmp"
Cause: Could not determine mount point for location specified.
Action: Ensure location specified is available.
Expected value: n/a
Actual value: n/a
I have checked the free space using the command:
[root@Rac2 /]# df -k /tmp
Output:
Filesystem 1k-blocks used Available Use% Mounted on
/dev/sda1 30470144 7826952 21070432 28% /
As you see above, the free space is enough, but could not determine mount point for /tmp.
Do any folk understand how to determine the mount point for directory /tmp ?
Thanks.

I have just checked "/home/oracle/.bash_profile". But in my computer, there is no "oracle" under /home directory.Is this your first time Linux and Oracle installation? I had a brief look at your referenced link. The reason why you do not find a "oracle" user is because the instructions use "ora11g" instead, which, btw, is not standard. The directories of your installation and your installation source can be somewhat different from known standards and you will have to adjust it to your system.
My best guess is that you have either missed something in the instructions or you need to ask the author of the blog what is wrong. The chance to find someone here who has experience with these custom instructions is probably unlikely.
I suggest you try to locate the cluster verification tool, which should be in the bin directory of your grid installation. Alternatively you might want to check the RAC, ASM & Clusterware Installation forum: RAC, ASM & Clusterware Installation

Messaging Server and Calendar Server Mount points for SAN

Hi! Jay,
We are planning to configure "JES 05Q4" Messaging and Calendar Servers on 2 v490 Servers running Solaris 9.0, Sun Cluster, Sun Volume Manager and UFS. The Servers will be connected to the SAN (EMC Symmetrix) for storage.
I have the following questions:
1. What are the SAN mount points to be setup for Messaging Server?
I was planning to have the following on SAN:
- /opt/SUNWmsgsr
- /var/opt/SUNWmsgsr
- Sun Cluster (Global Devices)
Are there any other mount points that needs to be on SAN for Messaging to be configured on Sun Cluster?
2. What are the SAN mount points to be setup for Calendar Server?
I was planning to have the following on SAN:
- /opt/SUNWics5
- /var/opt/SUNWics5
- /etc/opt/SUNWics5
3. What are the SAN mount points to be setup for Web Server (v 6.0) for Delegated Admin 1.2?
- /opt/ES60 (Planned location for Web Server)
Delegated Admin will be installed under /opt/ES60/ida12
Directory server will be on its own cluster. Are there any other storage needs to be considered?
Also, Is there a good document that walks through step-by-step on how to install Messaging, Calendar and Web Server on 2 node Sun Cluster.
The installation document doesn't do a good job or atleast I am seeing a lot of gaps.
Thanks

Hi,
There are basically two choices..
a) Have local binaries in cluster nodes (e.g 2 nodes) ... which means there will be two sets of binaries, one on each node in your case.
Then when you configure the software ..you will have to point the data directory to a cluster filesystem which may not be neccasarily global. But this filsystem should be mountable on both nodes.
The advantage of this method is that ... during patching and similar system maintenance activities....the downtime is minimum...
The disadvantage is that you have to maintain two sets of binaries ..i.e patch twice.
The suggested filesystems can be e.g.
/opt for local binaries
/SJE/SUNWmsgr for data (used during configure option)
This will mean installing the binaries twice...
b) Having a single copy of binaries on clustered filesystem....
This was the norm in iMS5.2 era ...and Sun would recommend this...though I have seen type a) also for iMs 5.2
This means there should no configuration files in local fs. Everything related to iPlanet on clustered filesystem.
I have not come accross type b) post SUNONE i.e 6.x .....it seems 6.x has to keep some files on the local filesystem anyway..so b) is either not possible or needs some special configuration
so may be you should try a) ...
The Sequence would be ..
After the cluster framework is ready:
1) Insall the binaries on both side
2 ) Install agent on one side
3) switch the filesytem resource on any node
4) Configure the software with the clustered FS
5) Switch the filesystem resource on the other node and useconfig of first node.
Cheers--

OFA for 11g.. Are multiple mount points (ie 2 or more) still recommended

Similar Messages

Maybe you are looking for