Questions of RAC setup

Questions are:
1. How many voting disks do we need per cluster (considering we have a redundant Fibre-channel connections)?
2. How many OCR disks do we need and what is the recommended access (OCFS2 or RAW)?
3. How do we calculate the bandwidth needed for the cluster interconnect? We currently have configured 2 x 1Gbit Ethernet ports for this.
The installation is on "Oracle Enterprise Linux 5.2 x86_64" and Oracle RAC 10GR2 on IBM x3850 M2 Servers and IBM N-Series storage connected via redundant Fibre-channel switches. Using ASM for disk handling.

I believe there are two issues here.
The first is that raw devices will not be supported in future Linux kernels. Raw device is a logical (character device) layer for the actual physical (block) device. In Linux for example, this will be +/dev/raw/raw1+ and it will be the logical raw device for physical device +/dev/sdf+.
With 10.2 CRS, you no longer need to use a raw device - you can run CRS's OCR and voting disks directly on the physical device (as the kernel now support the O_DIRECT flag for I/O on physical devices too and a raw device interface is not needed). (see Metalink note 401132.1)
Thus raw devices (as in the logical device required to be created on a physical device) are already on the way out.
The second issue is that we need a means of allocating shared storage for the OCR and voting devices at CRS startup time. Which means we have two basic choices. We either allocate a device directly (via raw for example), or we install some 3rd party (non-CRS) software that creates a sharable device for us - such as gfs, ocfs, and so on.
The problem with additional software - it increases the number of moving parts for CRS and is another software layer that can fail. Which leaves something like using raw devices a better option (less complex).
But this choice is forced because there is no "+native method+" we can use. This has changed with 11.2 and later, as ASM now can provide that functionality for us.
However, if you are still on pre 11.2 CRS and RAC/ASM, what are your choices then? Using a cluster file system or raw is pretty much the only choices available. And in that respect, raw is the one that will cause less problems and issues and is the preferred option (from personal experience).

Similar Messages

Ethernet Cards(NICs) in RAC setup

DB version: 11g
I came across this nice link about RAC setup
http://www.akadia.com/services/ora_rac.html
Question 1.
In the diagram, I can see eth0 and eth1 in both nodes. They are physical ethernet cards (NICs). Right?
Question 2.
Each node having 2 ethernet cards is industry's standard practice. Right?
Question 3.
For Private InterConnect ===>One ethernet card from each node, connected via Switch1
For Public InterConnect plus Connection to Shared storage ====> One ethernet card from each node, connected via Switch2Is this the standard practice in Production DBs?

resistanceIsFruitful wrote:
In the diagram, I can see eth0 and eth1 in both nodes. They are physical ethernet cards (NICs). Right? Yes. (or virtual ones in a virtualised environment).
Question 2.
Each node having 2 ethernet cards is industry's standard practice. Right?Yes and No. Yes, typically 2 ethernet ports. Servers usually will have 4 or more ports, if one also includes the management ethernet ports of the server. (but these cannot be used by o/s or RAC/Grid software).
If one wants to have failover for either the private or public interfaces, then at least 2 physical interfaces are needed for each. E.g. eth0 and eth1 will be bonded as logical interface bond0 for the public network, and eth2 and eth3 bonded as bond1 for the private network.
Question 3.
For Private InterConnect ===>One ethernet card from each node, connected via Switch1
For Public InterConnect plus Connection to Shared storage ====> One ethernet card from each node, connected via Switch2
Is this the standard practice in Production DBs?Yes - fundamentally this is correct. You want your Interconnect to be a fully private network with a dedicated switch. That said, high availability does not mean all your connectivity eggs in a single basket. If you have a single private switch, that is a single point of failure for the entire cluster. If you have a single private interface, that is a single point of failure for that node to the Interconnect (cable failures are not uncommon).
Thus you should be looking at redudancy at both the switch and the NIC/interface levels.

Oracle 11gR2 RAC setup with ASM option running on two nodes . Seeking help

Hi ,
Recently we have migrated to 11gR2 extended RAC for one of our application using ASM . Its a two node RAC setup . We are using ASM Normal redundancy feature maintaining two failure group .
I need few clarification regarding our new DR set up .
We are planning to have EMC SRDF replication from any of node storage ( Either 1 or 2) to the DR disks .
Its stright to use when we use the normal file systems . In case of ASM , I am not sure how this replication will work ? and on which level the replication should happen ?
Do i need an ASM instance up and running on DR server , Which will recognize the disks as it is once DR disks are separated from prod replication and assigned to the server ?
What should be prerequisite on DR server before we can start the srdf replication (EMC) and make it work .
Thanks for All your Valuble suggestion .

Before you decide to use EMC replication give considerable thought to the fact that you are introducing a single point of failure.
Hardware snapshot replication is very fast. That is all it is. If you database crashes because of corruption on disk ... your replica will be equally corrupt.
What is fast can not do what DataGuard does ... verify quality. If the point is DR. Go with DataGuard unless EMC is willing to, in writing (not verbally) give you a guarantee (in hard cash) to cover all costs if they replicate bad blocks and it causes an outage.
And just so no one thinks I'm picking o EMC ... the same weakness exists with all of the other storage vendor's products.

How to identify Which ports are using in RAC setup?

Hi,
how to identify Which ports are using in RAC setup?
Thnaks,

[Viewing Port Numbers and Access URLS|http://download.oracle.com/docs/cd/B19306_01/install.102/b14205/ports.htm#sthref1208]

Document on 10g RAC setup on solaris using vmware

Hi All,
I am planning to setup "Oracle 10g RAC setup on Solaris using vmware", but I am strucked up at installation of Soaris 10 in VMWare.
Can any body please help/provide me the document on Solaris 10 installation for RAC setup .
The main problems I am having during the SOlaris 10 setup is
1) Setting up the static public and private addresses required for RAC
2) Partitioning the disk space.
Thanks in advance,
Mahipal Reddy

Refer these,
http://www.scribd.com/doc/15650880/Install-Rac-on-Solaris-Vmware
http://www.disperu.com/using-vmware-server-install-10g-rac-on-solaris/
http://nayyares.blogspot.com/2008/11/step-by-step-rac-10g-r2-solaris-10.html
Thanks
Edited by: Cj on Dec 13, 2010 2:38 AM

Linux flavors for oracle 10gR2 RAC setup

Hi,
Oracle10gR2
Which are the alternate Linux flavors (most similar to RHEL 5.5) that can be used for Oracle10gR2 RAC setup ??
Regards

Hi,
while there other Linux versions are also vertified, best is to use Oracle Linux http://www.oracle.com/us/technologies/linux/index.html
This is based on RHEL (you also have a 5.5. version).
Regards
Sebastian

RAC setup on VMware Workstation is possible?

Hi,
Is it possible to create a RAC setup on VMware workstation(VMware-workstation-6.5.1-126130) or VMware server is mandatory for such a setup.
My planned RAC setup will as show below.
OS:Enterprise-R5-U2-Server-i386
Oracle clusterware:10201_clusterware_linux32
Oracle Database:10201_database_linux32
I am planning to store OCR and Voting disks on shared RAW devices and database files on ASM.
Thank You and Regards...

Yes its possible to install it over workstation. There is as such o difference in both the products except that server is free but workstation is paid. Otherwise, both are virual machine softwares.
HTH
Aman....

ORA-12504 in an Oracle 9.2.0.1 RAC setup

Hi I keep getting the following error when connecting to either nodes in a RAC setup:
ORA-12504: TNS:listener was not given the SID in CONNECT_DATA
I tried different permutations for the tnsnames.ora but keep getting this error. Does anybody know how to fix this problem?
Here's my tnsnames.ora entries:
LISTENERS_dotbk =
(DESCRIPTION =
(ADDRESS_LIST =
(ADDRESS = (PROTOCOL= TCP)(Host= srvrwin.corp.wassup.com)(Port= 1521))
(ADDRESS = (PROTOCOL= TCP)(Host= srvrtwo.corp.wassup.com)(Port= 1526))
LISTENER1_dotbk1 =
(DESCRIPTION =
(ADDRESS = (PROTOCOL= TCP)(Host= srvrwin.corp.wassup.com)(Port= 1521))
LISTENER2_dotbk2 =
(DESCRIPTION =
(ADDRESS = (PROTOCOL= TCP)(Host= srvrtwo.corp.wassup.com)(Port= 1526))
dotbk1 =
(DESCRIPTION =
(ADDRESS_LIST =
(ADDRESS = (PROTOCOL= TCP)(Host= srvrwin.corp.wassup.com)(Port= 1521))
(CONNECT_DATA = (SERVICE_NAME = dotbk1)(INSTANCE_NAME = dotbk1))
dotbk2 =
(DESCRIPTION =
(ADDRESS_LIST =
(ADDRESS = (PROTOCOL= TCP)(Host= srvrtwo.corp.wassup.com)(Port= 1526))
(CONNECT_DATA = (SERVICE_NAME = dotbk2)(INSTANCE_NAME = dotbk2))

Hi Carsten
Thanks for the response. My versions of OWB are the same it's just the database versions that are different.
I ended up opening an SR but since the version is so old there wasn't much help except to try and apply the 10.1.0.5 OWB patch and cross my fingers.
I think I'll just end up putting the new repository in a different Oracle 10g database.
Thanks again for your help.
-gary

RAC setup for R12

Hi,
I read docs about EBS R12.1 RAC setup that said:
The traditional and still widely-used configuration where the Concurrent (Parallel) Processing servers
are located on the database tier. Unless there is a network constraint, the recommendation now is
to place them on the application tier for greater flexibility and scalability.Right now we have a multi node setup (not yet on RAC) , the AppsTier is on node1 (including the Concurrent Managers which is the default setup), while the Db-Tier is on node2. Does this mean I do not need to transfer anymore the Concurrent Manager to the Db-Tier on node2?
Are Concurrent Managers also referred to as the Concurrent Processing servers?
If my Concurrent Managers are located on the AppsTier, How can I convert it to Concurrent Paralllel Processing servers? Can you please give me notes on how to setup this type of configuration?
Thanks a lot,
Msk

Hi,
Right now we have a multi node setup (not yet on RAC) , the AppsTier is on node1 (including the Concurrent Managers which is the default setup), while the Db-Tier is on node2. Does this mean I do not need to transfer anymore the Concurrent Manager to the Db-Tier on node2?You have the option of moving the CM to the database node, but it is not required here (recommended to keep it on the application tier node where you have it running now).
Are Concurrent Managers also referred to as the Concurrent Processing servers?Yes.
If my Concurrent Managers are located on the AppsTier, How can I convert it to Concurrent Paralllel Processing servers? Can you please give me notes on how to setup this type of configuration?See the docs referenced in this thread.
PCP Configuration
PCP Configuration
Regards,
Hussein

Questions about RAC

I was reading the article about installing the RAC.
My conclussion , I can have till 127 instaces of a Oracle 10G database, working simultane. That is good really good and powerfull.
My questions
1º). ¿Can I have , by example, 3 , or any other number, of computer working with a HTMLDB instance or Oracle Aplication Server everyone serving pages to Internet and all then conect to one of those nodes? I think that yes, but prefer confirmation
2º) In case of failure of any of these nodes, broken computer ,by example, and I need to replace it, or I want add new nodes to a RAC system. ¿How I can do this? or ¿Where I can find documentation about this?
thanks for your help.

Is posting Question about RAC some kind of thought experiment on clustering? Like, if you post the same question several tyimes do you get faster answers?
Cheers, APC

RAC setup- Configure hostname in ODI agent

I have my Master & Work repositories hosted on a Oracle RAC environment. I have a standalone ODI agent which might installed in the same physical server or in a different server.
In a RAC setup, there are multiple hostnames for the same database. How do I make the ODI agent aware that it should look into a 2nd server (when the 1st host is down) when trying to connect to the repository ? I want to know where this should be configured in the agent setup.
Does this get automatically taken care of once I include both the hostnames in the JDBC URL of the repositories ?

All depends on the JDBC you're using...
JDBC Thin work with the Oracle listener,
and JDBC OCI (which need an Oracle Client on the machine) works with a TNS description (here you can define several HOST).
I had have several issues with the Thin on our RAC so we've changed for an OCI one...
regards,
Brice

Help needed on SRVCTL importance in RAC Setup.

Hi gurus,
We are using 11GR2 database with RAC on AIX machines also we are using oracle clusterware only. It was implemented by some body else and we are supporting this environment. We observed that we are not able to start the Database/listerners using srvctl. it is throughing the error. We just like to know if this is acceptable. i am trying to say that SRVCTL is critical in RAC setup and we should be in a position to start the DB using srvctl else we need to start DB manually when any problem happens due to cluster level problem.
My requirement is
1) Is my point is valid. i.e SRVCTL is compulsary for cluster or not?
2) if SRVCTL is compulsary then i need to have metalink note saying SRVCTL is compulsary with out that cluster will not start the DB automatically.
REgards
DBA.

Yes ... you are right .
You can use SRVCTL to start and stop the database and instances, manage configuration information, and to move or remove instances and services. You can also use SRVCTL to add services. SRVCTL also manages configuration information.
Some SRVCTL operations store configuration information in the Oracle Cluster Registry (OCR). SRVCTL performs other operations, such as starting and stopping instances, by sending requests to the Oracle Clusterware process (CRSD), which then starts or stops the Oracle Clusterware resources.
You should be able to perform operations using srvctl.
Check out the trace files for any errors ?
Also srvctl is kind of mandatory thing and without this you wont be able to make change to configurations.
So find the root cause and you can also raise SR to oracle support.
Regards
Rajesh

Hardware requirements for RAC Setup on Linux

Hi,
Can any body help me...We are soon going for the RAC setup on linux operating system.So,i need to provide the customer,the hardware needed for this setup.
Please send me the list of all the things right from small things....
regards,
Mohammed

Hi Mohammed,
In addition you may find good information in below doc:
http://download.oracle.com/docs/cd/B19306_01/rac.102/b28759/toc.htm
Regards,
Marcello M.

ISCSI Recommendations for RAC Setup

Hi all,
My company is interested in setting up Oracle RAC for back-of-house processing (all traffic would be internal...no client facing apps or anything) and are looking to do so by using an iSCSI shared storage device. We are currently using the Dell MSD3000i in an Oracle fail-safe solution, but I'm not 100% confident that it will be able to handle a RAC setup. Has anyone out there had any experience using this device with RAC? Any recommendations out there from someone who has had good experiences with a similar device?
The vision for the current setup is having a two-node RAC with two quad-core servers running Windows 2008 64-bit with 32 GB of RAM each. We would be running Oracle Standard Edition.
Any feedback given would be much appreciated.
Thanks,
Joe

Hi
1) Oracle does not certified hardware, but Oracle RAC certifies storage technologies as Fiber channel, SCSI and iSCSI. Please take a look to: http://www.oracle.com/technology/products/database/clustering/certify/tech_generic_window_new.html
2) With your storage device i think this link could help you:
http://www.dell.com/downloads/global/solutions/FINAL_10gR2_SE_RAC_on_MD3000i_RefConf_v1%203.pdf?c=ae&l=en&s=bsd
But it is on linux, maybe you can switch to Red Hat or Oracle Enterprise Linux.
Regards
Miguel.

Question about RAC and ASM

Hi,
We are thinking about build RAC using ASM for OEM database. It'll have two nodes, oracle 10g and Hitachi san with solaris ( or linux ). I've few questions about RAC and ASM.
1) Do I need to have ASM instance running on each node ? ( most likely yes... but want to make sure )
2) can I share oracle_home between ASM instance and database instance ? what is the best choice ?
3) I'm planning to use shared disks for all files, all databases.... what are the pros/cons ?
4) what should be the installation procedure ? Meaning, first create ASM instance on each node, then install cluster software, build RAC databases...... can someone explain..
5) I believe RMAN is the only option for backup since I'm using ASM, correct ?
I'm a newbie to RAC and ASM..
Thanks,

user4866039 wrote:
1) Do I need to have ASM instance running on each node ? ( most likely yes... but want to make sure )yes.
2) can I share oracle_home between ASM instance and database instance ? what is the best choice ?in 10g you can, in 11gr2 you cannot. and it might be better to seperate them because it will give you more flexibility with patching.
3) I'm planning to use shared disks for all files, all databases.... what are the pros/cons ?if you share your oracle_home you won't be able to do rolling updates. so i recommend to keep oracle_homes local.
4) what should be the installation procedure ? Meaning, first create ASM instance on each node, then install cluster software, build RAC databases...... can someone explain..follow the install guide for your respective version. for 10g you'd install clusterware first, then asm and database is last.
5) I believe RMAN is the only option for backup since I'm using ASM, correct ?pretty much. you could stop your database and dump the raw devices or use asmcmd/asmftp but rman is definitely the best choice
Bjoern

Questions of RAC setup

Similar Messages

Maybe you are looking for