Issue of cluster Configuration
HelloI have a 3-node cluster and 1 of the present has more cpu rather the other 2.
Is there a document that suggests that all nodes should be equal?
Hello
This question you can get you anwser using this note:
RAC: Frequently asked Questions [220970.1 ID]
Can I have different servers in my Oracle RAC? Can be from different vendors? Can be different sizes?
...
Oracle RAC supports a cluster nodes with different hardware configurations. An example is a cluster to another node with 6 CPUs and 3 knots with 4 processors. This can easily happen when you add a new node after the cluster has been in production for some time. For this type of configuration, clients must take into account some additional features for optimal cluster performance. The servers in the cluster can be from different vendors; It is fully supported as long as they run the same binary files. Given that many clients implement Oracle RAC for high availability, you must make sure that your hardware vendor will support the configuration. If you have a failure, you will be supported for the hardware configuration?
...
Kind regards
Levi Pereira
Tags: Database
Similar Questions
-
Help on consistency + Weblogic Cluster configuration
Hello
I am new to consistency. I explore the consistency to fill one of my condition.
My needs: I have an application deployed on a clustered server logical Web environment. I need to cache data specific to the application (java card) and that should be available to all nodes in the cluster WLS. Application that are cached, data can be updated from any of the node in the cluster WLS and updated the application cache data should be available for all the nodes in the cluster WLS. The nodes in the cluster of WLS resides in different machines from different places.
Can I use coherence as a services of caching for my requirement?
If Yes, do I need to start a server cache consistency on each node of the cluster WLS?
If so, do I need to specially configure these servers cache, so that these cache servers would communicate each other in case of updates of data in the cache?
It will be very helpful if you could also provide sample configurations to achieve the above scenarios.
Thanks in advance!
Kind regards
Srinivas MYou can start a cache server in the network by using for example
# home directory BEA_HOME="/home/oracle/bea" export BEA_HOME # java vendor (for example Oracle or Sun) JAVA_VENDOR="Oracle" export JAVA_VENDOR # cache server klasse CACHE_SERVER_CLASS="com.tangosol.net.DefaultCacheServer" export CACHE_SERVER_CLASS # coherence options COHERENCE_OPTIONS="-Dtangosol.coherence.management=all" export COHERENCE_OPTIONS COHERENCE_OPTIONS="${COHERENCE_OPTIONS} -Dtangosol.coherence.management.remote=true" #COHERENCE_OPTIONS="${COHERENCE_OPTIONS} -Dtangosol.coherence.cacheconfig=WEB-INF/classes/session-cache-config.xml" #COHERENCE_OPTIONS="${COHERENCE_OPTIONS} -Dtangosol.coherence.session.localstorage=true" WL_HOME="${BEA_HOME}/wlserver_10.3" export WL_HOME BEA_JAVA_HOME="${BEA_HOME}/jrockit_160_05_R27.6.2-20" export BEA_JAVA_HOME SUN_JAVA_HOME="${BEA_HOME}/jdk160_11" export SUN_JAVA_HOME if [ "${JAVA_VENDOR}" = "Oracle" ]; then JAVA_HOME="${BEA_JAVA_HOME}" export JAVA_HOME MEM_ARGS="-jrockit -Xms512m -Xmx512m -Xss128k -Xgcprio:throughput" export MEM_ARGS fi if [ "${JAVA_VENDOR}" = "Sun" ]; then JAVA_HOME="${SUN_JAVA_HOME}" export JAVA_HOME MEM_ARGS="-server -Xmx512m -Xms512m -Xmn256m -Xss128k -XX:PermSize=128m -XX:MaxPermSize=128m -XX:+UseParallelGC -XX:ParallelGCThreads=2 -XX:+UseParallelOldGC -XX:+AggressiveOpts -XX:+UseBiasedLocking" export MEM_ARGS fi # classpath for the cache server CLASSPATH="${WL_HOME}/coherence/coherence.jar" export CLASSPATH # start the cache server ${JAVA_HOME}/bin/java ${MEM_ARGS} ${COHERENCE_OPTIONS} ${CACHE_SERVER_CLASS}
-Dtangosol.coherence.management = all option gives made you some useful information about what the cache by using for example the browser mbean provided by jconsole or jrockit mission control.
You can create an application in which you create a cache, below is an example of a servlet
import com.tangosol.net.CacheFactory; import com.tangosol.net.NamedCache; import javax.servlet.ServletException; import javax.servlet.http.HttpServlet; import javax.servlet.http.HttpServletRequest; import javax.servlet.http.HttpServletResponse; import java.io.IOException; import java.util.Random; public class TestServlet extends HttpServlet { private NamedCache movies; public void init() throws ServletException { movies = CacheFactory.getCache("repl-movies"); movies.put(10, new Movie(10, "Rear Window", "Alfred Hitchcock")); movies.put(20, new Movie(20, "Vertigo", "Alfred Hitchcock")); movies.put(30, new Movie(30, "Double Indemnity", "Billy Wilder")); movies.put(40, new Movie(40, "Touch of Evil", "Orson Welles")); } protected void doGet(HttpServletRequest request, HttpServletResponse response) throws ServletException, IOException { Integer[] integers = {10, 20, 30, 40}; Random random = new Random(); while (true) { movies.get(integers[random.nextInt(4)]); } } }
Package the servlet into such a WAR file and deploy it in a WebLogic cluster configured.
By accessing the servlet on the different nodes in the cluster a new cache is created and added to the
already running DefaultServer.The most difficult part is to configure your cache. As you can see in servlet initialization we EU something like CacheFactory.getCache ("repl-movies");
Lack of consistency with some configurations of cache example that are contained in the coherence.jar (coherence-cache-config. (XML).
The following entry is usedrepl-* example-replicated example-replicated ReplicatedCache unlimited-backing-map true A replicated cache is probably what you need as well. You like all the data available on all nodes. If you want to use
your own cache configuration, you can add the option - Dtangosol.coherence.cacheconfig (see the for the default server startup script)A good introduction on coherence can be found here:
http://www.packtpub.com/article/installing-coherence-3.5-and-accessing-the-data-grid-1
and
http://www.packtpub.com/article/installing-coherence-3.5-and-accessing-the-data-grid-2?utm_source=rk_coherence_abr1_0310&utm_medium=content&utm_campaign=ramsai -
Is ASM cluster configuration, possible to detail all cluster hosts
In the case of configuration in cluster of ASM (where its possible to create one ASM DiskGroup is shared between all hosts in the cluster) is it possible to obtain information on all hosts (cluster node) (which are part of this cluster configuration) one machine.
For example. If we have three nodes Node1, Node2, and Node3 configured using ASM cluster (i.e. default ASM instance will be ASM2 ASM1, and + ASM3 on these machine) and then I create multiple starts (some shared between all three nodes and bit limited to single node or two) using a set of SAN disk visible to all.
In such a scenario, can I get the details of Node2 and Node3 comes to log in Node1 (i.e. Since one machine one single node part of ASM cluster configuration, I need information of all other machines/nodes).
Is this possible? If so, some tips on the way which will be a great help.
~ Stéphane.Hi stephane,
Maybe I'm missing something here, or I'm too detailed, but depending on how General should be your script, there are some steps, you must check to avoid the false assumptions - e.g., as here: "+ to get the CRS_HOME, I can grep for crsd.bin. If I don't get this running process, this means that configurations is a StandAlone OracleDB and turns to the ASM, ASM is too stand-alone instance. + »
If you can't find the crsd.bin running, it could also mean that it will not turn now - for some reason any. If you find a DSO running instance, but not crsd.bin, Yes, you can generally assume that you would have a configuration of ASM nonclustered (autonomous), unless you run into a BUG / failure.
Therefore, it might be a good idea to check the existence / first configuration, then to the current process. For example:
IF you have installed Oracle Clusterware, you will find an init.cssd under /etc/init.d (for Linux, similar directories for other OS). Therefore, check first for the operating system, if you have the OS specific scripts, then just check for the init.cssd init directory.
The advantages of checking the init.cssd are: you will get the Oracle Clusterware House - it is set as a variable:
= ORA_CRS_HOME / ora/clusterware AND you usually get the Oracle user: ORACLE_USER = oracleThen you can check if Clusterware is running. You will have more than one way to do it now. You can use ps - ef or you can use Clusterware orders as crsctl check CRS or things of that kind.
Of some people could argue that there is a DSMB for ASM in standalone (non-cluster) ASM facilities running and there is therefore an init.cssd. YES, but this init.cssd shouldn't be in init in this case - see the comment in the header of the script init.cssd: "In a local, nonclustered, installation without CARS, it must be placed in a directory rcX.d".
However, just to be safe, you can always check whether or not you will find init.crsd, init.evmd, which are very strong indicators that Clusterware has been installed on the system, unless it was not uninstalled properly...
Once you've found the Clusterware installed, running AND you know the path to the House - it's easy:
As said previously, you can use olsnodes to get the nodes in the cluster. You can still use crs_stat. It will tell you the ASM / RAC, the instance number and this node the instance number is running on:
Name = ora.myrac1.ASM1.asm
TYPE = application
TARGET = ONLINE
STATUS = online on pmrac1As for CARS DB instances:
Name = ora.myrac.myrac1.inst
TYPE = application
TARGET = ONLINE
STATUS = online on pmrac1To determine which is the node that you are running your script on: host name must do.
Of course, all this will work only IF you have installed Clusterware. For environments not clustered, you must work with the ps - ef command, sqlplus and oratab file as you mentioned below, which makes it a bit more complicated. We could predict that's SO so OTHERWISE in your script - IF cluster, THEN do foo, otherwise do BLAH. Just an idea.
For this question:
Belongs to ASM cluster - (yes/no)
(If crsd.bin is running, ASM is clustered. Can this method be invoked always?)IF you have a cluster installed, ASM should be part - is the default and recommendation. However, you can run in a broken configuration. You want to guess in your script and secure?
Just a few random ideas. Thank you
Markus -
NEWBIE: vSwitch in a matter of the host Cluster Configuration
Hello
I'm pretty new to vmware and will be starting my first implementation in the coming days. I had training and have a very good understanding of what it is I'm doing it but have a few issues of networking.
If we do not have Enterprise Plus and must use Standard vSwitches on each host in the cluster (using 3 host cluster), that mean each virtual NETWORK adapter configurations on each vswitch hosts will be exactly the same? So, if I put the vMotion group of ports on host A to use the physical network (vmnic0 and vmnic1) interface cards whose 192.168.10.1 IP address et.2. It means that I then configure the same IPS for vMotion Port Group (vmnic0 and vmnic1) on host B and C of the host in the cluster?
If this is the case then wouldn't NIC physical on the three ESX hosts in the cluster need to have same IP configs, in which case wouldn't that cause conflicts of intellectual property, no? Please advise if I am not clear enough. Thankx.
Rick,
You would need separate IP addresses for each VMkernel port that you are creating. Thus, for example, if you create 2 VMkernel ports per host, 1 host could avoir.1 et.2, host 2 may avoir.3 et.4 and host 3 can be affectee.5 et.6.
But remember, you are not assigning IP addresses for uplink adapters (vmnic), you will then assign to the VMkernel Ports. Your ESXi host will have a VMkernel default port for management (created during installation) which will have the IP address of your hosts assigned management.
If you create a separate network for vMotion (recommended), this would require a VMkernel port separated with the checkbox "Use this port for vMotion" checked, and who would live on, most likely, a separate vSwitch with its own uplinks. The VMkernel port would have its own unique IP address. You will need to repeat this configuration (with specific to host IPs) on each host.
-
HFM Application Cluster Configuration - failed
Hello
I use the EPM 11.1.1.3 System. The Oracle database version is 10g. When I tried to reconfigure HFM Application cluster via the EMP Configurator, I get error below. I just added the name of the HFM server to the cluster and click Next. Please could you help on this issue.
Error Msg in the log file:
The server name duplicated SERVERNAME line: 1062 file:.\CHFMConfig.cpp error code: 0 x 0 error: the operation completed successfully.
The user has been properly set up DCOM limitations
Try to configure the cluster server.
Cannot set the application servers and Clusters. Line: 515 file:.\CHFMConfigUtil.cpp Error code: 0x8004082a error: #1578 IDispatch error
General error. Line: 827 file:.\CHFMConfigUtil.cpp Error code: 0 x 80040815 error: error IDispatch #1557
Thank you
Michel KI removed the name of server in doubles in the HFM_Config.xml file and solved the problem.
Thank you
-
I'll put up a cluster between VCS - 02c in version 8.x and one of them will be the captain. I will put all endpoints in the VCS - C Master points, what would happen if the master of VCS - C problems? My end points would be automatically transferred to another VCS - C (backup)?
Hello
Setup is a cluster of VCS VCS a set up a Master and others as a slave.
Your points of termination/Codec must register to the VCS cluster (you need to update your DNS with VCS cluster name i.e. clustervcs.domain.com with the ip addresses of two VCS. You can use SRV records to do a primary and the other back). If your CODEC does not support the SRV record then you can put the IP of master VCS and the registration, the VCS Master will provide the idec with its ip address and ip address of the slave VCS.
A good link:
http://www.netcraftsmen.NET/blogs/entry/Cisco-VCs-clustering-configuration.html
I hope this helps.
Kind regards
Ahmed
-
Transmission of events when Cluster configured
When clustering is configured all nodes in the cluster is the transmission of events or is - it just address ILB?
All nodes to do.
-
Move cluster configured with replication SRM and Vsphere
All,
We go 2 groups in the near future to achieve a geographical distance between the 2 sites.
A cluster is equipped with VMware 5.0 replication SRM and vSphere and the other cluster running VMware 5.5 and SRM in combination with replication of vSphere.
What is the best way to move these clusters, I must stop the replication or interrupt the replication process and are these action to perform on SRM or we need to perform any action on the virtual machine itself?
I read that a stop is setting the State to not configured including destroy all files in the target directory.
Does anyone have an experience by physically moving a cluster managed by SRM (and replication of vSphere)
Any help is welcome
Thank you
Gerbert Coolen
Just pause replication before stop you the VR device on the site target/DR. stop vs break with vSphere replication - VMware vSphere Blog - Articles from VMware
-
Hello
Is it possible to have 4 ESXi clusters, each with their own set of SAN data warehouses, all joined to a Cluster of unique data store? I have not found exactly this configuration described anywhere. I understand there are limitations such as 32 data etc. stores and paths to data warehouses will be available from certain hosts.
Has anyone tried this before? I'm trying to simplify investment VM for Cloud management platform.
Thank you
Mike
This is a supported configuration, and I have a few clients that are running this configuration without problem. Have a look here for details: linking DRS multiple clusters in a single Storage DRS datastore cluster. - frankdenneman.nl
-
Issue of cluster master / slave
Hello
I read the document here scale for the Information Explosion: Cluster of master / slave with vFabric Postgres 9.2 on vSphere | VMware vFabric Blog... and the document that details how to configure the workspace to use an external postgres DB but missing some info configuration when you use a cluster of master / slave
When I put in place the cluster needs the url jdbc point to the master host? If the cluster rocking how Workspace knows to point to the secondary host? He ddon'tappear that there is a dedicated cluster IP/hostname can I report?
Or should I create a CNAME DNS that points to the master and change it manually in a failover event?
I posted this on the forum of the workspace, but no one was able to answer here.
Since 9.2.4, vPostgres device includes a couple of scripts dedicated to this task, make your life much easier. If you do not need to do these steps manually on the server. More information here:
Then to manage a transition to a slave in the case of a master failure, JDBC has a base implementation in the code which was introduced last year with 39b085d of validation and you can specify multiple URLS like this:
JDBC:PostgreSQL://host1:PORT1, host2:port2 / test
Connection is selected alternately until success. There is no documentation on this subject, but... I'll ping the devs of jdbc to correct this. If you are using an older version of JDBC, you switch to an alternative, such as a layer of tomcat.
-
Netspp old storage and issues of Cluster VMWARE
Hello
I saw a former NETAPP FAS 2020 connected storage and confiigured with the cluster of 4.1
Customer is polanning to install new vcenter and ESXi 5 new hosts and want to migrate virtual machines to new Cluster and get rid of the old because of some problems of old hosts
I heard some mouth sharing old storage Netapp Storage can cause some strange problems and can cause some virtual machines get corrupted
is that correct? strange problems known?
Sharing storage should be the correct way to migrate virtual machines safely to the news of ESXi
It's exactly the way I use to migrate virtual machines to a cluster. I've done a couple of times (also with 3.5/4.x/5.0 storage Netapp and ESXi) and never had corrupted VMs. what VMFS is designed for.
Of course, I would not share a LUN to ESXi hosts between different vCenters as final configuration but for migration, it is reasonable.
Just know that:
-Do not upgreade to VMFS5 until all ESXi4 hosts are disconnected.
-Storage i/o Control does not work across multiple vCenters.
-migrate does not mean "vMotion" but: power off > remove inventory > add to the inventory > voltage (of course...)
-
Consistency of unicast environment Cluster configuration
Hi all
I have a scenario where I need to set up a cluster of consistency so that it can be used in a cluster SOA or OSB. A BPEL application will be using coherence of cache to store temporary data, and retrieve that data later in the different process. Could you please explain to me what all the steps I have to follow in order to configure the cluster of coherence.
Thank you
Ankush
-
Second cell in vCD cluster configuration/license?
I install the first server and provided the response file to the 2nd cell and finished the installation/configuration. Now I can see my 2nd cell under 'Cloudy cells' on the web interface of the cell from my 1st. However, I can't access the web interface of the 2nd cell (layout not found but all services are currently running) and I see the 'vCenter' is checked only for cell1 under "cloudy cells. Is this expected behavior, or I guess to see the web interface on cell2 and configure the license, vcenter and vshield information?
Renou - kamel
There is a single VC proxy for each instance of vCenter server that you have registered with vCD and the proxy works on a cell in the cluster, so it is expected.
You should look at the logs on the disk of the cell 2nd to see if errors have been reported. I start with cell.log in $VCLOUD_HOME/logs and if that doesn't show any errors, then check vcloud-container - debug.info.
-
Consistency Cluster configuration
Given the configuration,
If there are two packages of ejb - jar A and B, and each has a configuration file from the cache checked in these packages of jar and they have both the same name of the cluster and cluster-port specified.
Should I wait:
There will be two instances of consistency from one for every A and B and then these two instances would form a consistency group?
Thank youHi Pierre,.
Yes I think it's okay. Two EJB applications need to be isolated from each other within the app for each server is a member of the cluster.
JK
-
When adding the second node to the cluster I m getting the errro message "the following nodes cannot verify that they can host the quorum... resource." (hr = 00000000, {B8C4066E-0246-4358-9DE5-25603EDD0CA0}, {A976DC09-0108-410A-AF57-68C05F9A42F7}, 1, 1, 1), (null)
I have attached the log files.
And as soon as I turn on the second node of the cluster first machine configuration file value "scsi0:1.present ="TRUE"turns to"False"
Clustering is not supported on VMware - it cannot share disks - so that's why you receive the warning of quorum. However, under VMware 5.5 and before it works but has need of a bit of configuration I think (google it).
Maybe you are looking for
-
How much RAM in a 2650XDVD satellite?
I have 64 MB in this laptop! They write that I can put only 128 MB this 64 MB!I put 512 MB in there with Windows XP? Or I get into trouble with the Bios or so?I have to do a Bios Update? or 192 MB ram is really the highest level for this laptop?
-
Questions/problems new today Lenovo Y70 network and touchpad
Ok Although it has a dual band wireless card, I can't get 72Mbps on 2.4 ghz and 5 GHz when my old laptop becomes a complete 144Mbps. Can I fix? Also, how do enable you scrolling on the touchpad? I can't find that
-
Have a lot of jobs in the queue and would like to erase everything and start a new.
-
HEPL disable me the system dv 4000
HEPL me. disable the System 4000 HP dv 00736
-
Hello I can not type an apostrophe in my emails. It won't let me. I LOVE my apostrophe and struggling to express things when I can't use it. Someone at - it a response?