CSM fault tolerance takeover time

Hi,
I am currently testing the fault tolerance feature on the CSM. During the testing I discovered that the takeover from active to standby is immediate when you reboot the active CSM. However when the active CSM comes back online it does not take over immediately. What I am seeing is that it takes between 3-4 minutes before the active module does a preempt.
Is there a hidden timer similar to preempt delay in HSRP at work here and is it possible to tweak it somehow?
Many thanks,
Murtaza

there is no hidden parameteres.
The delay is to make sure the flows can be copied from active to standby before it takes over.
Gilles.

Similar Messages

CSM Fault Tolerance and IGMP Snooping

For "connection redundancy" the redundancy guide says to turn off IGMP snooping.
Is there any way around this?
I need to have multicasting everywhere and I don't want to multicast all streams to every port on this switch.

Most of the show statements are at the end of the attached file in an earlier post. The vservers details are at the end of this post.
I have an ARP entry for the VIP - 0001.64f9.1a64, but it does not respond to pings. I tried both the alias and the server vlan IP as the default gateway of the servers.
I took a trace and found that the VIP sends a TCP reset immediately after a request. I have tried versions 4.2.1 and 4.1.4 with the same result. I wonder if this could be a problem with the Sup720 with 12.2.17d IOS. I also tried the CSM in slots 2 and 3.
720Test2#sh mod csm 3 vserver detail
SOFTRICITY, type = SLB, state = OPERATIONAL, v_index = 10
virtual = 10.10.249.6/32:0 bidir, any, service = NONE, advertise = FALSE
idle = 3600, replicate csrp = none, vlan = ALL, pending = 30, layer 4
max parse len = 2000, persist rebalance = TRUE
ssl sticky offset = 0, length = 32
conns = 0, total conns = 1
Default policy:
server farm = SOFT1, backup =
sticky: timer = 0, subnet = 0.0.0.0, group id = 0
Policy Tot matches Client pkts Server pkts
(default) 1 1 0

Deterministic Fault Tolerant Load Balancing

The USA has an unfortunate penchant for granting patents that arguably do not merit patent protection. Some of these are things that are blindingly obvious. Others are just not sufficiently inventive.
Anyway, since I have no funds for patent searches, nor patent applications, and there are some other complications, I've decided to post this to establish prior-art for an algorithm. I don't claim that the algorithm is clever, nor novel, nor even that it violates no existing patents. This posting is simply to ensure that to the extent that someone might be granted a patent on it, they can't, because it has already been published.
The Java connection is that I've done a fair amount of the work required to turn this into a real system in Java.
Suppose you have set of processors, p0 thru pn-1, and each piece of work to be performed by a processor has some number k associated with it. The problem is to allocate the work roughly equally across the subset of processors that are actually functioning. Further, over a period of time, a series of related pieces of work may arrive with the same k. To the maximum possible extent you want each of the related pieces of work to be handled by the same processor. If a processor fails, you want its work to be distributed across the remaining processors, but still maintaining the property that pieces of work with a given value for k are handled by the same processor. In general we assume that the k values are randomly spread through a large number space.
The motivation for these requirements is that for a given k the processor may be caching information that improves performance. Or it may be enforcing some invariant, such as in a lock manager where each request for a given lock must go to the same processor, or it clearly won't function.
To achieve this, construct a list of integers of size n. Element i contains i if processor i is functional, and -1 otherwise.
Calculate k mod n, and use the result as an index into the list. If the value contained there is non-negative, then it is the number of the processor to use. If it is -1, remove the element from the list, decrement the value of n and repeat. Continue until a processor number is found.
This scheme is fault tolerant to a degree, in that the resulting system has a high level of availability.
It also has the property that the failure of a processor only impacts on the allocation of pieces of work that would have been allocated to the failed processor. It does not result in a complete rearrangement of the work allocations. This makes things a lot simpler when dealing with things like distributed lock managers.
The fault tolerance can be improved by an extension of the algorithm that allows a distributed master/slave arrangement, where the master number for a given k is determined as above, and a slave number is obtained by treating the master as if it were not functioning. Each processor is a master for some subset of the k values, and is a slave for another subset. For any given master, each of the other processors is a slave for a roughly equal portion of the given master's subset of the k values.
There are some boring details that I've not discussed, such as how an entity wanting work to be done determines which processors are functioning, and the stuff related to the exact sequence of steps that must be performed when a processor breaks, or is repaired. I don't believe anyone could patent them because once you start thinking about it, the steps are pretty obvious.

I wouldn't be so sure that a simple post to the java forums
is all you need to prove this 'prior-art' is it ?
Don't you need to actually use it? Or have you seen a laywer
and this was their advice. Even if you have no money for it Im sure
there are free legal services; even universities, you could contact.
I don't believe anyone could patent them because once you
start thinking about it, the steps are pretty obvious.The steps of anything a generally simple, it's the putting-them-together
that you can patent :)

WL 7.0 Fault-tolerant stateful beans. (Newbie)

Hello,
I was wondering if in WL 7.0 there is a way to make a stateful session
bean automatically persist its state without having to be passivated
as a whole.
The bean that I need to build would persist its state to a DB each
time a remote method call is executed in it. The purpose is to make
this bean fault tolerant without having to serialize the whole object
graph by using passivation.
I heard that WL 7.0 has a new feature that can make this happen but I
can't seem to find any info on this. Could anyone point me into the
right search path?
Thanks in advance.
Mateo.

I'm not aware of such feature in WLS 7.0 EJB container.
Where did you heard about it? If at all there's a feature you should
be able to locate at our online docs.
Here's the link
http://e-docs.bea.com/wls/docs70/ejb/index.html
Kumar
Mateo wrote:
Hello,
I was wondering if in WL 7.0 there is a way to make a stateful session
bean automatically persist its state without having to be passivated
as a whole.
The bean that I need to build would persist its state to a DB each
time a remote method call is executed in it. The purpose is to make
this bean fault tolerant without having to serialize the whole object
graph by using passivation.
I heard that WL 7.0 has a new feature that can make this happen but I
can't seem to find any info on this. Could anyone point me into the
right search path?
Thanks in advance.
Mateo.

UOO sequencing along with WLS high availability cluster and fault tolerance

Hi WebLogic gurus.
My customer is currently using the following Oracle products to integrate Siebel Order Mgmt to Oracle BRM:
* WebLogic Server 10.3.1
* Oracle OSB 11g
They use path service feature of a WebLogic clustered environment.
They have configured EAI to use the UOO(Unit Of Order) Weblogic 10.3.1 feature to preserve the natural order of subsequent modifications on the same entity.
They are going to apply UOO to a distributed queue for high availability.
They have the following questions:
1) When during the processing of messages having the same UOO, the end point becomes unavailable, and another node is available in order to migrate, there is a chance the UOO messages exist in the failed endpoint.
2) During the migration of the initial endpoint, are these messages persisted?
By persisted we mean that when other messages arrive with the same UOO in the migrated endpoint this migrated resource contains also the messages that existed before the migration?
3) During the migration of endpoints is the client receiving error messages or not?
I've found an entry on the WLS cluster documentation regarding fault tolerance of such solution.
Special Considerations For Targeting a Path Service
When the path service for a cluster is targeted to a migratable target, as a best practice, the path
service and its custom store should be the only users of that migratable target.
When a path service is targeted to a migratable target its provides enhanced storage of message
unit-of-order (UOO) information for JMS distributed destinations, since the UOO information
will be based on the entire migratable target instead of being based only on the server instance
hosting the distributed destinations member.
Do you have any feedback to that?
My customer is worry about loosing UOO sequencing during migration of endpoints !!
best regards & thanks,
Marco

First, if using a distributed queue the Forward Delay attribute controls the number of seconds WebLogic JMS will wait before trying to forward the messages. By default, the value is set to −1, which means that forwarding is disabled. Setting a Forward Delay is incompatible with strictly ordered message processing, including the Unit-of-Order feature.
When using unit-of-order with distributed destinations, you should always send the messages to the distributed destination rather than to one of its members. If you are not careful, sending messages directly to a member destination may result in messages for the same unit-of-order going to more than one member destination and cause you to lose your message ordering.
When unit-of-order messages are processed, they will be processed in strict order. While the current unit-of-order message is being processed by a message consumer, the next message in the unit-of-order will not be delivered unless it is to the same transaction or session. If no message associated with a particular unit-of-order is processing, then a message associated with that unit-of-order may go to any session that’s consuming from the message’s destination. This guarantees that all messages will be processed one at a time and in order, and any rollback or recover will not prevent ordered processing of the messages.
The path service uses a persistent store to save the state of which member destination a particular unit-of-order is currently using. When a Path Service receives the first message for a particular unit-of-order bound for a distributed destination, it uses the normal JMS load balancing heuristics to select which member destination will handle the unit and writes that information into its persistent store. The Path Service ensures that a new UOO, or an old UOO that has no messages currently on any destination, can be enqueued anywhere in the cluster. Adding and removing member destinations will not disrupt any existing unit-of-order because the routing decision is made dynamically and those decisions are persistent.
If the Path Service is unavailable, any requests to create new units-of-order will throw the JMSOrderException until the Path Service is available. Information about existing units-of-order are cached in the connection factory and destination servers so the Path Service availability typically will not prevent existing unit-of-order messages from being sent or processed.
Hope this helps.

RAID not fault tolerant on K7N2 Delta 2 Platinum?

Hi all,
I'd really appreciate your help with this puzzle.....
Got a new mobo & 3k Barton chip.... Installed 'em into rig described in my sig below.
Minimum hardware was installed for this operation:
new mobo & chip
1 gig mem card
CDRW
Floppy drive
SATA drives x2 each on separate SATA port
video card
Entered Bios on 1st boot:
Set SATA to enabled in IDE RAID window
Boot priority to CD Rom for install
Hard disk boot priority to SATA Mirror
F10'ed into NVRAID BIOS:
Set up the RAID as a mirror (RAID 1) with optimal stripe size
cleared & reformatted both discs
Mirror is recognised as healthy on reboot.
Booted into WIN XP SP2 slipstreamed install (this disc has been used successfully before on my previous mobo so I know it works)
Pressed F6 & loaded NV RAID class driver & NV nForce Storage Controller successfully
Did the Windows install routine & this is where it starts to deviate from what is written in the manual....
the request to install the RAID driver during the GUI part of the install never appeared (but this is described as a "might be prompted" in the manual so I wasn't too bothered by its non-appearance) and the driver floppy did spin up at one point so I assumed it had installed each drive automatically
Went into WinXP and finished the install...
Surprisingly, both RAID drives are now visible individually in the systray as removable drives under the SAFELY REMOVE HARDWARE icon.
Went into Disk Management expecting the Initialize & Convert Disk Wizard to appear but no sign of it: RAID drives are visible as ONE drive (as you would expect) but are NOT fault tolerant.
Things I've done differently each time I installed:
Tried to convert the mirror to a dynamic disc but this also failed every time.
Installed nForce2 system drivers before & after going into Disk Management on different installs
Installed NVRAIDMAN.exe and this reports the mirror as functional & healthy.
Left the floppy in the drive during the GUI part of WIN XP installing itself and removed it on other install
I've read the nvRAID FAQ & this isn't covered by it.
Can't find any reference to the Initialize & Convert Disk Wizard in Microsoft Help & Support, except in a Windows Server 2003 KB article, which I can't find again....
I've repeated the whole process several times to ensure I haven't missed anything, but the end result is always the same.
So I now have a weird mirror array that is not fault tolerant in Windoze Disk Management - is it doing what it's supposed to?
Any ideas??

As it seems you are doing extensive tests, i would suggest a simple one:
install as you have (where the RAID manager tells you it is setup correctly and redundant and windows does not see it as redundant), power off and take one drive off. Reboot and see the result. (it should still work and the controller should tell you that redundancy is not valid anymore)
After that, power off, replug the drive and see if it will reconstruct the array.
Try for the 2nd drive.
Then, you will be sure that redundancy is present...
It seems that with SATA, you can even hotswap the drives (hence the appearance of the drives in the SAFELY remove box).

How make DNS for real fault tolerance?

Hi,
Is there any change to make DNS-server fault tolerance for clients (member servers and applications)
Scenario:
Primary and Secodary DNS (not AD-integrated)
If we like to move or rebuild crashed DNS-server from old to new host, there will came service brake for one DNS-server (new server is up and running some time before DNS have been configured (you can ping it and icmp response)),
DNS-client is happy for that thus there is no DNS-service or zones and not use secondary DNS-server cause server answer with ICMP and criticals softwares stop working. Is there any change for DNS real fault tolerance?
I have understood this feature is same in Windows world and non-Windows world

Hi,
According to your description, my understanding is that there will came service brake for one DNS server when moving or rebuilding crashed DNS-server from old to new host.
There are 2 DNS servers: primary and secondary DNS server, both are not AD-Integrated.
DNS design specifications recommend that at least 2 DNS servers be used to host each zone. For standard primary-type zones, a secondary server is required to add and configure the zone to appear to other DNS servers in the network. This design may provide
a basic level of fault tolerance for resolving names. Once primary DNS server crashes, secondary DNS server is ready only and can’t process update requests, so we need to manually change the secondary server to primary, and then try to repair the crashed one
or add another DNS server.
Comparatively, AD-Integrated primary zones, secondary servers are supported but not required for this purpose. For example, two DNS servers running on domain controllers can be redundant primary servers for a zone, providing the same benefits of adding a
secondary server while including additional advantages.
Depending on your need, and choose a better one.
Best Regards,
Eve Wang
Please remember to mark the replies as answers if they help and unmark them if they provide no help. If you have feedback for TechNet Support, contact [email protected]

Fault tolerance for network printing

What options are available to provide fault tolerance for network printing? Currently, all our shared printers are on ServerA. Is it possible to have similar shared printers on ServerB and have the clients automatically switch to Server B if
Server A became unavailable? We don't care about load balancing as much as fault tolerance. Thanks to all who post!

A failover cluster node ? (not in nlb;
Servers in a Network Load Balancing (NLB) failover cluster cannot be used as print servers in Windows Server 2008 )
For environments that require high availability, you can use a failover cluster as a print server. If a node in the cluster fails, all print functionality will fail over to the next node in the cluster. To improve failover times, we recommend
that the administrator for the cluster force failover to each node when new print drivers are installed on the server. During a failover, the driver installation is forced to occur on the active node. The installation of the driver on each node
can require several minutes. Forcing this installation process during maintenance will make sure that any unplanned failovers during usual operation will be very quick, because the drivers will already be installed on each node.
Regards, Philippe
Don't forget to mark as answer or vote as helpful to help identify good information. ( linkedin endorsement never hurt too :o) )
Answer an interesting question ? Create a
wiki article about it!

Fault tolerance capabilities of PI

Hi all,
Can anybody explain the fault tolerance capabilities of PI or please refer to any document which explains the same.
Thanks and regards,
Rajesh

As stephen replied what kind of tolerence you are looking from the PI.
One capability I can locate is, for the Asynchronous communication, PI provides the guaranteed delivery service. If communication get failed then PI tries three times to deliver the message to the receiver. If all three times it failed to deliver the message then it set the status to failed.

More fault tolerance

Ok. Here goes. I am trying to use BPEL to generate a workflow based on information it gets from XML and use the workflow to call different web services (in a specific order) that provide some incremental functionality to my job. There may be more than one web service for a given functionality. Also a web service may fail at doing the task because it did not have essential resources. In the event of that happening, can BPEL go back one (or a few steps-depending on dependencies) and then replan ( generate a different script) to get the job done, using values that it might have stored in the oracle database?
This automatic workflow generation and fault tolerance would certainly help....
Thanks,
Faraz

Hi!
This is an interesting use case. We do not offer those features out of the box at this time. You might want to take a look at the replay fault capability. Combine with a programmic use of bpelc to generate and compile on the fly a bpel process, it might provide a step in the direction you are looking for.
Edwin

Fault-Tolerant Networks

Hi,
For testing i want implementation one technique from fault-tolerant network (Simple) in virtual environment.
Can you help me? Thanks

Hi Arash_89,
General practices to be considered when configuring networks in Failover Clusters about the network contains many aspects, such Identifying single points of failure and configuring
redundancy at every point in the network is very critical to maintain high availability, you can refer the following article to know more detail:
Configuring Windows Failover Cluster Networks
http://blogs.technet.com/b/askcore/archive/2014/02/20/configuring-windows-failover-cluster-networks.aspx
I’m glad to be of help to you!
We
are trying to better understand customer views on social support experience, so your participation in this
interview project would be greatly appreciated if you have time.
Thanks for helping make community forums a great place.

Fault tolerance client /server

hi
i use a very basic client/server socket program
very similar to sun:
http://java.sun.com/docs/books/tutorial/networking/sockets/clientServer.html
many client can connect to the server...
i would like to add a fault tolerance...
a client try to connect a few time... that don't work, the client try to the second server...
does a need to do something similar to
public class KKMultiServer {
    public static void main(String[] args) throws IOException {
        ServerSocket serverSocket1 = null;
        ServerSocket serverSocket1 = null;
        boolean listening = true;
        try {
            serverSocket = new ServerSocket(4444);
            serverSocket = new ServerSocket(4445);
        } catch (IOException e) {
            System.err.println("Could not listen on port.");
            System.exit(-1);
        while (listening){
         new KKMultiServerThread(serverSocket1.accept()).start();
            new KKMultiServerThread(serverSocket2.accept()).start();
        serverSocket1.close();
        serverSocket2.close();
}thanks

i need your help to finis my code
what i want to do is to create to start 2 tread of server in my class main server....
and each class server start a thread for each connection...
each time a client do a request, i count it
theses method is in server class
import java.net.*;
import java.io.*;
public class ServerMain {
    public static void main(String[] args) throws IOException {
         //will call 2 time server here
         Server server1 = new Server();
         Server server2 = new Server();
         server1(4444).start();
         server2(4445).start();
import java.net.*;
import java.io.*;
public class Server extends Thread{
    private static int nbRequete=0;
    public synchronized int ajouterRequete(){
        return nbRequete++;
    public synchronized int getRequete(){
        return nbRequete;
    public int port=0;
    public Server(int port) {
        super("Server");
        this.port = port;
    public void run() {
        ServerSocket serverSocket = null;
        Socket s=null;
        boolean listening = true;
       // Server server = new Server();
       // int port;
        try {
             //will put a port variable
            serverSocket = new ServerSocket(port);
            s=serverSocket.accept();
        } catch (IOException e) {
            System.err.println("Could not listen on port.");
            System.exit(-1);
        System.out.println("server run, wait connecting....");
        while (listening)
            new ServerThread(s).start();
        //serverSocket.close();
import java.net.*;
import java.io.*;
public class ServerThread extends Thread {
    private Socket socket = null;
    public ServerThread(Socket socket) {
        super("ServerThread");
        this.socket = socket;
    public void run() {
        try {
            System.out.println("connexion successfull");
            System.out.println("Attente de l'entree.....");
            PrintWriter out = new PrintWriter(socket.getOutputStream(), true);
            BufferedReader in = new BufferedReader(
                        new InputStreamReader(
                            socket.getInputStream()));
            String inputLine, outputLine;
            while ((inputLine = in.readLine()) != null) {
                System.out.println("Server: " + inputLine);
                inputLine = inputLine.toUpperCase();
                out.println(inputLine);
                if (inputLine.equals("Bye."))
                    break;
            out.close();
            in.close();
            socket.close();
        } catch (IOException e) {
            e.printStackTrace();
}the server need to have different port
i'm a little bit confuse
i don't know how to finish this project
:confused:
i have some error i don't understand
ServerMain.java:11: cannot find symbol
symbol : constructor Server()
location: class Server
        Server server1 = new Server();
                         ^
ServerMain.java:12: cannot find symbol
symbol : constructor Server()
location: class Server
        Server server2 = new Server();
                         ^
ServerMain.java:13: cannot find symbol
symbol : method server1(int)
location: class ServerMain
        server1(4444).start();
        ^
ServerMain.java:14: cannot find symbol
symbol : method server2(int)
location: class ServerMain
        server2(4445).start();
        ^
4 errors

How do I create a fault-tolerant DLL call from LV?

I've created an .exe from a VI that calls a DLL, when that DLL crashes,
it crashes the whole application. This DLL is too complicated to
completely debug and it, in turn calls DLLs I have no control
over. It crashes approximately once every 100,000 iterations.
How do I make a DLL call fault-tolerent?
Attachments:
crash.gif ‏19 KB

I might not have been very clear in my answer - this isn't a feature that CAN be added to LabVIEW...the DLL threw an unhandled exception across a DLL interface - which pretty much means a system exception (code shouldn't be throwing language exceptions across DLL boundaries as they are language specific). This means that something went very wrong with the code and there is a good chance that the process space has been corrupted. And from my own debugging experience, if it's only crashing 1 time out of 100,000, that doesn't mean it isn't corrupting memory or otherwise causing problems...it was just the time it was caught.
Considering this, the Call Library Node already is fault tolerant - we catch the exception and report it to the user. You have a valid position that we should simply return this as an error. It's also a valid position that it would be a road paved with good intentions. All you need to do is remember what life was like in Windows 3.1 - where one process could silently corrupt another - to realize the amount of untraceable errors that could occur. Yes, the error out value should be handled by the user, but when an application of 100's or 1000's of VIs crashes seemingly at random because someone didn't, you can see the problem.
I will say I like the idea of a separate memory space. It is unfortunate that Windows only offers that through creating a separate process, but it is a neat idea. I worked in the VXI/VME world for a while, which does have multiple address spaces.
BTW - using signal handlers to catch exceptions is unfortunately not a 100% solution...signals are a partial port of a Unix concept in Windows. You would be best to use SEH around the call (http://msdn.microsoft.com/library/default.asp?url=/library/en-us/vccelng/htm/key_s-z_4.asp). If the DLL is already a C++ DLL, then you can use the global catch of C++ instead
int err = 0;
try {
} catch(...) {
err = -1;
Again, you still are at risk of a corrupted process, but this allows you to avoid our handling of it.
Having spent years writing and (god help me) debugging device drivers, I love this topic. If you want to talk more about it and various options, feel free to shoot me an email at [email protected].
Brian Tyler
http://detritus.blogs.com/lycangeek

Load Balancing & Fault tolerance in Oracle Apps

Hello
I am pretty new to Oracle Apps and Oracle 9iAS
We are trying to asked to find out the ways of deplying load balancing and fault tolerance in Oracle Apps
I have gone thru the following articles of metalink and some more from google
Configuring Web Cache as a Load Balancer for Application Servers
Create new middle tier node in existing Apps 11i environment using cloning and then load balance
Integrating and using Web Cache with Forms 9i for Load Balancing
Running Multiple OC4J Instances From a Single Install of 9iAS 1.0.2.2
LOAD BALANCING ORACLE APPLICATIONS ON UNIX
Load Balancing in 11i
OC4J Clustering Setup
OC4J Load Balancing for Forms 9i
Setting up 11i E-Business suite using a hardware load balancer
Sharing an APPL_TOP in Oracle Applications 11i
In Oracle Apps Concepts pdf, found that
Load balancing occurs when there are multiple installations of web server, forms server, reports server, concurrent manager server etc
Lets consider forms server component of middle tier
Can there be multiple INSTANCES of forms server within SINGLE INSTALLATION ???
If it has to be MULTIPLE INSTALLATIONS, then it will require multiple physical machines
What is software load balancing then ???
Also read that
Oracle 9i AS instance is combition of Oracle HTTP Server (OHS) and one or more instances of Oracle9iAS Container for J2EE (OC4J)
Thus software load balancing can be implemented for Web server component of middle tier using multiple instances of OC4J ???
Forms metrix server configuration is hardware load balancing ??
What about Oracle9iAS Web Cache clustering ??
Please clarify the doubts and suggest me the way
Thanks a lot

dear all can any one help me on the following;
We want to install Oracle Apps 11.5.9 on 4 IBM AIX 5L boxes, the first node
will hold the database , concurrent and Admin Tiers. as for the other 3 nodes
it will hold the Forms/Web Tiers and we need to use forms metrics for load
balancing taking into consideration that we can't use Shared APPL_TOP. we tried
to find the way how to install and configure this solution but all the
documents are talking aboout the concept only and this concept is applicable.
We need the exact steps on how do the installation either using rapidwiz or any
thing else
fadi

Fault Tolerance - MCS 7825 H3 & MCS 7835 H2 Servers

Hi,
I have enabled network fault tolerance on my CUC and CCM Servers, i thought after enabling the network fault tolerance the server would change the mac address to Virtual mac address which indeed it may affect the licenses.
But after enabling fault tolerance and restarting the server also i didn't see any change in the MAC ID. Its the same MAC ID.
Pls advice.
Regards
Jagadish G

Hi Jagadish,
do u get the option of 1000 MB/s when u configure set network nic eth0 speed ? can u check?
secondly, Bond 0 would be showing Auto disabled which cannot be changed.
Can u share snapshot of show netowrk failover?
regds,
aman

CSM fault tolerance takeover time

Similar Messages

Maybe you are looking for