admincenter and test the connectivity Warnning

Unable to identify the problem

Matlab is working fine but these warnings are not resolved by the application. When i use admincenter and test the connectivity the following warning occurs

Kindly Guide Me How To Fix It

ID #  Start Time  Stop Time  Executing Node  Subject Node  Test Category  Test Description  Result  Message  
1  Sep 19, 2012 7:26:48 PM  Sep 19, 2012 7:26:50 PM  ac2-pc  ac2-pc  PORT_CONNECT  PingServerSocketHost Test  SUCCESS    
2  Sep 19, 2012 7:26:48 PM  Sep 19, 2012 7:26:50 PM  aclp2  ac2-pc  PORT_CONNECT  PingServerSocketHost Test  *WARNING  Host AC2-PC does not respond to java.net.InetAddress.isReachable().  *
3  Sep 19, 2012 7:26:49 PM  Sep 19, 2012 7:26:52 PM  ac2-pc  aclp2  PORT_CONNECT  PingServerSocketHost Test  *WARNING  Host ACLP2 does not respond to java.net.InetAddress.isReachable().  *
4  Sep 19, 2012 7:26:52 PM  Sep 19, 2012 7:26:52 PM  ac2-pc  aclp2  PORT_CONNECT  ConnectToServerSocket Test (remote port 27355)  SUCCESS    
5  Sep 19, 2012 7:26:53 PM  Sep 19, 2012 7:26:53 PM  ac2-pc  ac2-pc  PORT_CONNECT  ConnectToServerSocket Test (remote port 27355)  SUCCESS    
6  Sep 19, 2012 7:26:53 PM  Sep 19, 2012 7:26:54 PM  aclp2  ac2-pc

Answer 1

[Jason Ross]

You have a name resolution problem.

You need to check and possibly modify:

    Firewall settings (if firewalls are in use)
    DNS records (forward and reverse lookup)
    DNS search domains

For each of the problematic hosts.

I can't tell you any exact fixes because you don't indicate what operating systems are in use, and these settings may or may not be accessible to you -- you will likely need root/Administrator privileges to change them. These settings can also vary widely by site, as well -- so you might need to contact networking staff (if you have them) to get the proper settings to make it work.

Please open new questions. Don't add on to existing ones that have been resolved for some time.

Answer 2

[leo]

Sir there is no client server environment. Window 7 is Installed on both the clients. I have the administrative rights. Application is also added in the firewall. if we off the firewalls no effects appear on warnings. Application is working properly.I have allowed these ports for inbound or outbound traffic TCP 27350-27470 and 28350-27450.

What can i do now? Is there any particular service to restart? Is there any registry setting required over it? Does any Trojan virus leads to this? should we deploy matlab on client a server environment?

Kindly Help Me I Need Your Help To Fix It Out.

[Jason Ross]

If you have two machines, you have a client-server environment. One is running the job manager, that's the server. The other is the client. They need to be able to talk to one another, and that means that their network settings must be correct.

If you have eliminated the firewall as the source of the error, then you need to move onto name resolution. I don't know how your site is set up in terms of DNS infrastructure, but you should be able to do the following in the Command Prompt, substituting the machine names and IP addresses where appropriate:

from machineA:

• ping (machineA IP address)
• ping machineA
• ping machineA.full.machine.name
• ping (machineB.ip.address)
• ping machineB
• ping machineB.full.machine.name

from machineB:

• ping (machineB IP address)
• ping machineB
• ping machineB.full.machine.name
• ping (machineA IP address)
• ping machineA
• ping machineA.full.machine.name

As for the exact settings to change, I can't tell you. They will most likely be in the following locations:

Computer Properties (Start, Computer, right-click, properties) gives you the "Name" and "Full Computer Name" properties.

You can change them under "Advanced System Settings" -- but you may need to be a Domain Administrator to (or have the right) to change the machine's name on the domain. Please, if you don't know what you are doing here, find someone who does -- if you change this setting and do not have the correct permissions, you'll remove your computer from the domain entirely and you'll likely be worse off than you are now. Proceed with caution!

The second location will be in the IPv4 Properties panel. You can get there by looking at the properties of your networking connection, selecting the "Internet Protocol Version 4" entry, clicking the "Advanced" tab, and then looking at the "DNS" tab. Items of interest here are the DNS search domains and DNS suffix.

Again, if you do not know what you are doing here, find someone who knows the proper settings for your environment. If you set up things improperly, you can end up with a machine that is in a worse state than now. And it's highly possible that your site has a certain way of setting this up, as well -- if you don't know what that is, find out first.

And please, please, please stop commenting on unrelated posts.

[leo]

i am using hostfile for DNS Resolution. Both system are on workgroup, it means they don't have any domain.

I know what i am trying to do.If i would like to check that application ports are blocked or not, What command should i use for it

This condition is working perfectly, What should i have to look now?

from machineA:

    ping (machineA IP address)
    ping machineA
    ping machineA.full.machine.name
    ping (machineB.ip.address)
    ping machineB
    ping machineB.full.machine.name

from machineB:

    ping (machineB IP address)
    ping machineB
    ping machineB.full.machine.name
    ping (machineA IP address)
    ping machineA
    ping machineA.full.machine.name

[Jason Ross]

Leave the firewalls off until you get the DNS stuff figured out.

If you are using hostfiles, make sure that the above tests work with the short hostname and fully qualified one both ways.

Also, ensure that the NetBIOS setting is as you expect. It's a second level of name resolution beyond DNS, and it can add another place for the names to be inconsistent.

I would also make the assumption that both machines are on the same subnet as well as (likely) the same switch? Or are there other things on the network between the two hosts?

For port blocking, you can check with netstat -a and use "<matlabroot>\toolbox\distcomp\bin\nodestatus -infolevel 2" to see more information. But you must get the name resolution items down before you move on, as they are the foundation things build on.

Also, be on the lookout for multiple NICs where you might not expect them -- for example if you are on a laptop with a wireless connection and a wired connection -- that can cause confusion, especially if there is a duplicate hostname out on the network and it gets resolved first, since I think Windows favors a DNS response over the hosts file.

Answer 3

[leo]

Ping with ip and by hostname is successful but DNS Resolution Errors "Inbound And Outbound Name Resolutions" are still on my way

For port blocking, you can check with netstat -a and use "<matlabroot>\toolbox\distcomp\bin\nodestatus -infolevel 2" when i run nodestatus bat file a blank command prompt appears with no information.

Host File is place on each system put in a TCP/IP Connection 192.168.0.1 is used as a preferred dns

On One System While Configuring Network Adopter Multiple Default Gateway appears when we change static ip ACLP2

I don't know what is the reason with it.

I have turned off the firewalls but it has no effect on admin center connectivity test.

While adding host to admincenter Multicast added by default and it is not resolved by the Matlab.

How to check NetBios Setting?

I have completely describe the situation below.

rogram called Matlab installed under Program Files\Matlab\R2011a Relevant Toolbox of the program is under \toolbox\distcomp\bin\

The toolbox is used to create a cluster of computers (called Nodes)

A Node can be a Client, a Job Manager or a Worker

Many cased a host can be a client and a job manager and also one of the workers

For example PC1: Client, JM, W PC2: W

Host Name: PC1 (this machine) is AC2-PC (192.168.0.8) PC2 is ACLP2 (192.168.0.11)

Service is called MDCE and must be isntalled and running on the JM and each Worker.

Problem 1: unable to remote start service. Example start mdce service on PC2 remotely from PC1

Problem 2: unable to check (sometimes) status of service remotely

Problem 3: service status erratic

How to install mdce -run cmd as Administrator and go to the distcomp direcoty -run "mdce uninstall -clean" -run "mdce install" -run "addMatlabToWindowsFirewall" -run "mdce start" -to check status run "nodestatus -remotehost

How to create jobmanager and worker on PC1

-open admincenter as Administrator

Answer 4

[leo]

How to edit Hostfile For DNS Resolution?

PC1 (this machine) is for Host Name AC2-PC (192.168.0.8) PC2 is Host Name ACLP2 (192.168.0.11)

Just Write this code at the bottom will work for Window 7?

AC2-PC 192.168.0.8 ACLP2 192.168.0.11

What Should be the preferred DNS Server in TCP/IP Ver 4?

If 192.168.0.1 is default gateway?

Should the similar configuration file should be placed on each system and what should be the preferred DNS for each system ?

Kindly Help Me Out As Soon As Possible

[Jason Ross]

You will need to populate the hosts file with the ip addresses and hostnames in the ways they will be addressed. According to Microsoft, this is one entry per line. So you would need these entries in the file. Make sure to use the same file on both PCs.

192.168.0.8 AC2-PC
192.168.0.8 ac2-pc
192.168.0.8 AC2-PC.YOURDOMAIN.COM
192.168.0.8 ac2-pc.yourdomain.com
192.168.0.11 ACLP2
192.168.0.11 aclp2
192.168.0.11 ACLP2.YOURDOMAIN.COM
192.168.0.11 aclp2.yourdomain.com

Note that the hostnames above are case-sensitive!!!! Were I doing this, I would keep everything lower-case, as is Internet convention. You also need to be sure that in DNS, the "Use hosts file for name resolution" is checked, to ensure that these names are resolved. You also need to replace "yourdomain.com" with the DNS domain settings for your network.

192.168.0.1 is by convention the preferred gateway for a network -- but that's a convention. You need to confirm it with the people who run your network.

DNS is another setting that is per-network. You need to check with the people who run your network what the DNS Server(s) should be set to. I have no way of knowing.

The bottom line here is that you need to get the network working properly. Right now it's not working properly, and you really need to reach out to people who understand how the network you are running on is configured, as correct hostname resolution is an absolute requirement for any clustered system to work.

Hosts file: http://technet.microsoft.com/en-us/library/cc751132.aspx

Answer 5

[leo]

when firewall is off every thing is oka, but when firewall gets on the following Warning comes our way

ID # Start Time Stop Time Executing Node Subject Node Test Category Test Description Result Message 1 Sep 19, 2012 7:26:48 PM Sep 19, 2012 7:26:50 PM ac2-pc ac2-pc PORT_CONNECT PingServerSocketHost Test SUCCESS 2 Sep 19, 2012 7:26:48 PM Sep 19, 2012 7:26:50 PM aclp2 ac2-pc PORT_CONNECT PingServerSocketHost Test *WARNING Host AC2-PC does not respond to java.net.InetAddress.isReachable(). * 3 Sep 19, 2012 7:26:49 PM Sep 19, 2012 7:26:52 PM ac2-pc aclp2 PORT_CONNECT PingServerSocketHost Test *WARNING Host ACLP2 does not respond to java.net.InetAddress.isReachable(). * 4 Sep 19, 2012 7:26:52 PM Sep 19, 2012 7:26:52 PM ac2-pc aclp2 PORT_CONNECT ConnectToServerSocket Test (remote port 27355) SUCCESS 5 Sep 19, 2012 7:26:53 PM Sep 19, 2012 7:26:53 PM ac2-pc ac2-pc PORT_CONNECT ConnectToServerSocket Test (remote port 27355) SUCCESS 6 Sep 19, 2012 7:26:53 PM Sep 19, 2012 7:26:54 PM aclp2 ac2-pc

How to allow java.net.InetAddress.isReachable() that it can easily through from firewall

What are the recommended ports should be added into the firewall?

[Jason Ross]

Good to know that progress is being made!

This technote describes the ports that need to be open on the clients, workers and job managers -- be very careful as they are not all the same ports, and it's also version dependent.

http://www.mathworks.com/support/solutions/en/data/1-39LCCA/index.html?solution=1-39LCCA

[leo]

I am using r2011a kindly tell me the exact port to add on the clients, workers,job managers on firewall and configuration as well.So i can fix it quickly Window 7 Is the current OS of all the systems.

I have to set 4 workers and 1 manager. if i would like to exceed the worker what should i do? kindly send me the reference material to understand base port concept?

Thanks for your help.

[leo]

Kindly help me out to fix it

i have added ports in firewall in manager,client,worker

27351-27363 ,28351-28363

but still have same problem R2011a is my Matlab Version according to the above manual

Below is a MDCE_DEF FIle ::::::::::::: :: mdce_def :: This file contains the definitions and the defaults for the mdce service and :: the processes it manages. The file is read only when starting and stopping :: the mdce service, or when running the mdce service command with the :: arguments console or, on Windows only, install and uninstall. :: When workers and job managers are started and stopped, they contact the mdce :: service they are running under to obtain the values of the definitions and :: defaults stored in this file. Thus, this file is not read again when :: starting and stopping workers and job managers. :: We recommend using a single mdce_def file for all the job managers and :: workers in your cluster, in order to ensure that they are set up in a :: consistent manner. :::::::::::::

:: Copyright 2004-2010 The MathWorks, Inc.

:::::::: :: :: MDCE Process and Logging :: ::::::::

:: HOSTNAME: The name of the host reading this file. :: Only change this line if the hostname command is unavailable, if the :: hostname command returns an incorrect host name, or if it returns a host :: name that cannot be resolved by all the other computers that the MDCE :: processes will communicate with. :: All the MDCE processes will advertise themselves using this host name. :: The job manager must be able to resolve the host name that the MATLAB :: workers advertise, and all MATLAB workers and clients must be able to :: resolve the host name that the job manager advertises. for /F %%i in ('hostname') do set HOSTNAME=%%i

:: BASE_PORT: The base port of the mdce service. :: The mdce service will use as many ports as needed, starting with BASE_PORT. :: On a machine that runs a total of nJ job managers and nW workers, the mdce :: service reserves a total of 5+nJ+3*nW consecutive ports for its own use. :: All job managers and workers, even those on different hosts, that are going :: to work together must use the same base port, otherwise they will not be :: able to contact each other. In addition MPI communication will occur on ports :: starting at BASE_PORT+1000 and using up nW consecutive ports. :: Some operating systems are reluctant to immediately free TCP ports from the :: TIME_WAIT state for use by the same or other processes, so you should allow :: unfirewalled communication on 2*nW ports for MPI communications. set BASE_PORT=27350

:: MDCEUSER: The username that is used when starting the mdce service. :: If unset, the mdce service will run as the user that starts the service. :: All the processes that the mdce service manages use the same username as :: the mdce service itself. :: Usually, the mdce service is started by default as LocalSystem. :: This username should be specified as DOMAINNAME\USERNAME, or as :: .\USERNAME if the user is a local user. set MDCEUSER=

:: MDCEPASS: The password for the user MDCEUSER. :: In order to run the mdce service as MDCEUSER, you need to specify a password :: for that user. This can either be done in the variable below or, if left :: blank, you will be prompted for a password when the service is installed. set MDCEPASS=

:: LOGBASE: The full path to a folder in which all logfiles should be written. :: The user the mdce service runs as must have write access to this folder. :: %TEMP% here is the system temp environment variable. :: The folder name should not be enclosed in double quotes, and the :: name must not contain parentheses. set LOGBASE=%TEMP%\MDCE\Log

:: CHECKPOINTBASE: The full path to a folder in which all checkpoint :: folders should be written. :: The user the mdce service is run as must have write access to this :: folder. On the host that runs the job manager, the job manager database :: is written to this folder, and it might require substantial diskspace. :: On the hosts that run the workers, all the data files that are transferred :: with the tasks are written to this folder. :: %TEMP% here is the system temp environment variable. :: The folder name should not be enclosed in double quotes, and the :: name must not contain parentheses. set CHECKPOINTBASE=%TEMP%\MDCE\Checkpoint

:::::::: :: :: Job Manager Security :: ::::::::

:: SECURITY_LEVEL: Choose the level of security in the cluster. :: The following levels are available: :: Level 0: no security. (Similar to R2009b and earlier releases.) This is the :: default value. :: Level 1: users are warned when they try to access other users' jobs and :: tasks, but can still perform all actions. :: Level 2: users need to enter a password to access their jobs and :: tasks. Other users do not have any access unless specified by the job :: owners (job property AuthorizedUsers). :: Level 3: same as level 2, but in addition, the jobs and tasks are run on the :: workers as the user to which they belong. The password needs to be the :: system password used to log on to a worker machine ("cluster password"). :: NOTE: This level requires settting USE_SECURE_COMMUNICATION to true. set SECURITY_LEVEL=0

:: USE_SECURE_COMMUNICATION: Use secure communication between services. :: By default, job managers and workers communicate over non-secure channels. :: In most cases this is the preferred setting, as either there is no need to :: protect the data or the connection is already protected from unauthorized :: access (e.g., the cluster network is isolated and has no access to the :: Internet). :: Setting this property to true results in encrypted communication between :: job managers and workers. This also requires a shared secret file on each :: participating host (see SHARED_SECRET_FILE below), and results in a :: performance degradation as all data is encrypted. :: This must be set to true if SECURITY_LEVEL is set to 3. set USE_SECURE_COMMUNICATION=false

:: SHARED_SECRET_FILE: The shared secret file used for secure communication. :: To establish secure connections between job managers and workers, a shared :: secret file is required on all participating hosts. Each service expects to :: find a copy of this file here at startup. Use the createSharedSecret.bat :: script to create a shared secret file. :: NOTE: secret files contain sensitive data and should be protected against :: unauthorized access. Anyone gaining access to the secret might be able to :: eavesdrop on the connections between services. :: If unset, the shared secret is expected in %CHECKPOINTBASE%\security\secret. set SHARED_SECRET_FILE=

:: ALLOW_CLIENT_PASSWORD_CACHE: Remember user passwords for future sessions. :: If set to true this option allows users to let the MATLAB client remember :: their logons for future client sessions. Users can still choose to not store :: any information at the password prompt in MATLAB. set ALLOW_CLIENT_PASSWORD_CACHE=true

:: ALLOWED_USERS: A list of users allowed to log on to the job manager. :: The following variable defines a list of users that are allowed to access the :: job manager. Multiple usernames are separated by commas. :: To allow any user to access the job manager, use the keyword ALL instead of :: a list of usernames. set ALLOWED_USERS=ALL

:: WORKER_DOMAIN: The Windows domain used by workers when logging on at :: security level 3. To run tasks as the user to which they belong, Windows :: requires a domain in addition to the user name. If a task belongs to USER, :: it will be run as %WORKER_DOMAIN%\USER. :: In most circumstances the default value will be correct and should not be :: altered. :: NOTE: This is required only when running on Windows at security level 3. set WORKER_DOMAIN=%USERDOMAIN%

:::::::: :: :: Job Manager and Worker Settings :: ::::::::

:: DEFAULT_JOB_MANAGER_NAME: The default name of the job manager. :: When a new job manager is started, it needs to be identified by a name on :: the network, and when a new worker is started, it needs to know the name of :: the job manager it should register with. This is the default job manager :: name used in both of these cases. :: The default job manager name can be overriden by the -name argument with the :: startjobmanager command, and by the -jobmanager argument with the :: startworker command. set DEFAULT_JOB_MANAGER_NAME=default_jobmanager

:: JOB_MANAGER_HOST: The host on which the job manager lookup process can be :: found. :: If specified, the MATLAB worker processes and the job manager process will :: use unicast to contact the job manager lookup process. :: If JOB_MANAGER_HOST is unset, the job manager will use unicast to contact :: its own lookup process. You can then also allow the MATLAB workers to :: unicast to the job manager lookup process by specifying -jobmanagerhost :: with the startworker command. :: If you are certain that your network supports multicast, you can force the :: job manager and the workers to use multicast to locate the job manager :: lookup process by using the -multicast flag with the startjobmanager and :: startworker commands. set JOB_MANAGER_HOST=

:: JOB_MANAGER_MAXIMUM_MEMORY: The maximum heap size of the job manager java :: process. set JOB_MANAGER_MAXIMUM_MEMORY=512m

:: DEFAULT_WORKER_NAME: The default name of the worker. :: The default worker name can be overridden by the -name argument with the :: startworker command. :: Note that worker names must be unique on each host. Therefore, you must :: use the -name flag with the startworker command if you want to start more :: than one worker on a single host. set DEFAULT_WORKER_NAME=%HOSTNAME%_worker

:: WORKER_START_TIMEOUT: The time in seconds worker sessions allow for MATLAB :: to start before detecting a stall. This value should be greater than the :: time it takes for a MATLAB session to start. set WORKER_START_TIMEOUT=600

[Jason Ross]

From the technote:

In MATLAB 7.9 (R2009b) and later, the port range 27370 through 27470 should be open.

That's pretty straightforward. Open those on all client hosts. Onto the job manager and workers:

BASEPORT through BASEPORT + 5 + N TCP ports (where N is the number of worker nodes in your cluster). If you plan on running parallel jobs, or if you plan to use MATLAB pool or pmode, you will also need to open BASEPORT+1000 through BASEPORT+1000+5+ N TCP ports.

BASEPORT=27350

N=4

27350+5+4=27359.

So the first range is 27350-27359.

For the second range you add 1000 to each number to get 28350-28359.

Open these ports on the jobmanager and worker hosts.

[leo]

in window xp firewall we can't define range. we have to add ports individually. So it is very difficult to define all these ports in the firewall

Is there any way to define the range of ports in firewall in window xp? what is the last range where we have to stop adding these intervals if starting limit is 27350-27359 Why second range is required 28350-28359 Is there any third range as well?

We can define range of ports in Window 7 Easily.

Why do we need to add second range?

For the second range you add 1000 to each number to get 28350-28359.

[Jason Ross]

Yes, I've had to test this with winxp before and added the ports. You might have some luck with the "netsh" command. There are some differences between capabilities in XP and Vista/7, but it might be helpful. Google for exact syntax in regards to the firewall and it's settings.

The technote explains the second set of ports:

If you plan on running parallel jobs, or if you plan to use MATLAB pool or pmode, you will also need to open BASEPORT+1000 through BASEPORT+1000+5+ N TCP ports.

I'm expecting that you'll be using matlabpool and/or parallel jobs. That's why you need to add them.

I've given you all the ports you should need to open on the client machines, worker machines and jobmanager host. Please refer to the tech note I referenced above for a full explanation.

[leo]

Today i have added these port one by one into Window xp firewall but nothing works for me 27350-27359. 28350-28359. i have configure matlab for parallel jobs. This application make me sick.

[Jason Ross]

If you are connecting from a client, 27370-27470 should be open, as well.

[Jason Ross]

And you are already up and running ... it's just fiddling with the firewall settings that's tripping you up a bit. Once you get that figured out you are good to go.