Stanley,
Thanks for all the information. The telnet information you did verify
that tcp server was listening.
For discovery, gms relies on multicast udp. We do have a test program
to assist in
diagnosing if two machines are seeing each other via multicast. The test
program is
strictly Java MulticastSocket.
Here is how to run the test on your two machines:
Open a terminal window on machine A and run
the following command in shoal/gms workspace.
% ant test-mcastsniffer
Then on another terminal window on machine B, cd into shoal/gms
directory and run
$ ant test-mcastsender
If the two machines can see each other, you will see the following
printed on the terminal
window of the mcast-sniffer terminal window.
test-mcastsniffer:
[java] Here's some multicast data
[java]
[java] Here's some multicast data
[java]
[java] Here's some multicast data
[java]
[java] Here's some multicast data
[java]
[java] Here's some multicast data
[java]
If no data is printed, then we need to work some on getting them to see
each other.
There are many reasons that they may not be able to see each other.
When I run
VPN, two terminal windows on same machine do not see each others
broadcast UNLESS
I set the IP address to use to be 127.0.0.1. (that is third argument to
test program denoted below
To customize the Network Interface to use (the above just uses defaults),
modify the following in build.xml.
<target name="test-mcastsender" depends="build" description="Run
MulticastSender">
<java classname="net.MulticastSender" fork="true">
<classpath>
<path refid="compile.classpath"/>
<pathelement location="${dist.dir}/${shoal-gms.jar}"/>
</classpath>
<arg value="229.9.1.1"/> <!-- default multicast group address -->
<arg value="2299"/> <!-- default mulicast port -->
<arg value="XXX.YYY.ZZZ.AAA"/> <!-- ip address of network interface to
use for multicast broadcast -->
</java>
</target>
<target name="test-mcastsniffer" depends="build" description="Run
MulticastSniffer">
<java classname="net.MulticastSniffer" fork="true">
<classpath>
<path refid="compile.classpath"/>
<pathelement location="${dist.dir}/${shoal-gms.jar}"/>
</classpath>
<arg value="229.9.1.1"/> <!-- default multicast group address -->
<arg value="2299"/> <!-- default multicast port -->
<arg value="XXX.YYY.ZZZ.BBB"/> < !-- ip address of network interface to
use for multicast receive -->
</java>
</target>
-Joe
On 1/4/11 4:36 AM, Stanley Xu wrote:
> The following is the output of the ifconfig on those two machines:
>
> Machine1
> eth1 Link encap:Ethernet HWaddr 00:A0:D1:EE:B1:09
> inet addr:192.168.11.91 Bcast:192.168.11.255
> Mask:255.255.255.0
> inet6 addr: fe80::2a0:d1ff:feee:b109/64 Scope:Link
> UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
> RX packets:38251006946 errors:0 dropped:246862 overruns:0
> frame:0
> TX packets:20934615765 errors:0 dropped:0 overruns:0 carrier:0
> collisions:0 txqueuelen:1000
> RX bytes:49418086991604 (44.9 TiB) TX bytes:8526692127415
> (7.7 TiB)
> Memory:fac60000-fac80000
>
> lo Link encap:Local Loopback
> inet addr:127.0.0.1 Mask:255.0.0.0
> inet6 addr: ::1/128 Scope:Host
> UP LOOPBACK RUNNING MTU:16436 Metric:1
> RX packets:2892195678 errors:0 dropped:0 overruns:0 frame:0
> TX packets:2892195678 errors:0 dropped:0 overruns:0 carrier:0
> collisions:0 txqueuelen:0
> RX bytes:23821021433042 (21.6 TiB) TX bytes:23821021433042
> (21.6 TiB)
>
> Machine2
> eth0 Link encap:Ethernet HWaddr F0:4D:A2:01:7E:6E
> inet addr:114.80.69.183 Bcast:114.80.69.191
> Mask:255.255.255.240
> inet6 addr: fe80::f24d:a2ff:fe01:7e6e/64 Scope:Link
> UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
> RX packets:130743485 errors:0 dropped:0 overruns:0 frame:0
> TX packets:82548770 errors:0 dropped:0 overruns:0 carrier:0
> collisions:0 txqueuelen:1000
> RX bytes:163177496353 (151.9 GiB) TX bytes:6858432901 (6.3 GiB)
> Interrupt:90 Memory:d6000000-d6012800
>
> eth2 Link encap:Ethernet HWaddr F0:4D:A2:01:7E:72
> inet addr:192.168.11.124 Bcast:192.168.11.255
> Mask:255.255.255.0
> inet6 addr: fe80::f24d:a2ff:fe01:7e72/64 Scope:Link
> UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
> RX packets:3931753690 errors:0 dropped:0 overruns:0 frame:0
> TX packets:132645337 errors:0 dropped:0 overruns:0 carrier:0
> collisions:0 txqueuelen:1000
> RX bytes:679219389327 (632.5 GiB) TX bytes:46934167055
> (43.7 GiB)
> Interrupt:106 Memory:da000000-da012800
>
> lo Link encap:Local Loopback
> inet addr:127.0.0.1 Mask:255.0.0.0
> inet6 addr: ::1/128 Scope:Host
> UP LOOPBACK RUNNING MTU:16436 Metric:1
> RX packets:21507 errors:0 dropped:0 overruns:0 frame:0
> TX packets:21507 errors:0 dropped:0 overruns:0 carrier:0
> collisions:0 txqueuelen:0
> RX bytes:3282262 (3.1 MiB) TX bytes:3282262 (3.1 MiB)
>
>
> I hava add the following code for setting the properties of the GMS
> service:
>
> Properties props = new Properties();
> props.put(ServiceProviderConfigurationKeys.BIND_INTERFACE_ADDRESS,
> ipAddress);
> return (GroupManagementService) GMSFactory.startGMSModule(serverName,
> groupName, GroupManagementService.MemberType.CORE, props);
> The ipAddress field is coming from the command line. And it is set to
> 192.168.11.91 on Machine1 and 192.168.11.124 on Machine2.
>
> And I could also use command
> telnet 192.168.11.91 9701
> telnet 192.168.11.124 9701
> on the two machines and get the output like the following:
> Trying 192.168.11.91...
> Connected to hadoopsh11091 (192.168.11.91).
> Escape character is '^]'.
> JXTAHELLO tcp://192.168.11.124:56594 <http://192.168.11.124:56594>
> tcp://192.168.11.91:9701 <http://192.168.11.91:9701>
> urn:jxta:uuid-59616261646162614A787461503250337EA20FB25AC44D0F88B6884346DE904203
> 0 1.1
>
> So I thought it should not be a firewall issue in tcp ports.
>
> Is there anything else I missed to get it work on the two different
> physical machines?
>
> Best wishes,
> Stanley Xu
>
>
>
> On Tue, Jan 4, 2011 at 2:24 PM, Stanley Xu <wenhao.xu_at_gmail.com
> <mailto:wenhao.xu_at_gmail.com>> wrote:
>
> Dear All,
>
> I am currently trying to use the shoal to make our application
> running on multiple clusters. I am using the sample code and it
> works well on the dev machine with two different JVM. But when I
> put the program to two different machine. It didn't work.
>
> Is there any settings or constraints if I wanted to use the shoal
> to do clustering on two different machine? Shall I set any ip or
> port in the program? Or I should open a specific port on the firewall?
>
> What does the "same subnet" means in the shoal? Does that means
> the two machine should have the same gateway ip?
>
> Thanks.
>
> Best wishes,
> Stanley Xu
>
>