Can the network have an influence if TTL is 0 ?

Question

Hi,
We are using Coherence 3.5.3 and have 3 windows services running that form 1 cluster on a single windows 2008 machine. The TTL is set to 0 since there is no need to leave the physical machine as the 3 JVMs run on the same pc. Could it be that the connected network still has an influence? If TTL is 0, does the communication between the JVMs (and thus the nodes of the cluster) still go over a network card ?
Are there things I can do to check this?
This is the logging I get from Coherence on the cluster:
21 二月 2012 21:26:48,761 [Logger@9219882 3.5.3/465p5] INFO Coherence - 3.5.3/465p5 (thread=DefaultCacheServer, member=2): Started DefaultCacheServer...
SafeCluster: Name=tmsngCluster
Group{Address=228.1.2.3, Port=34405, TTL=0}
MasterMemberSet
ThisMember=Member(Id=2, Timestamp=2012-02-21 21:26:46.64, Address=192.168.0.100:8089, MachineId=14549, Location=machine:WIN-OJ9GGEALAL5,process:368,member:INPUT1, Role=Input)
OldestMember=Member(Id=1, Timestamp=2012-02-21 21:26:16.29, Address=192.168.0.100:8088, MachineId=14549, Location=machine:WIN-OJ9GGEALAL5,process:792,member:PROCESS1, Role=Process)
ActualMemberSet=MemberSet(Size=3, BitSetCount=2
Member(Id=1, Timestamp=2012-02-21 21:26:16.29, Address=192.168.0.100:8088, MachineId=14549, Location=machine:WIN-OJ9GGEALAL5,process:792,member:PROCESS1, Role=Process)
Member(Id=2, Timestamp=2012-02-21 21:26:46.64, Address=192.168.0.100:8089, MachineId=14549, Location=machine:WIN-OJ9GGEALAL5,process:368,member:INPUT1, Role=Input)
Member(Id=3, Timestamp=2012-02-21 21:26:46.67, Address=192.168.0.100:8090, MachineId=14549, Location=machine:WIN-OJ9GGEALAL5,process:3272,member:CLIENT1, Role=Output)
RecycleMillis=240000
RecycleSet=MemberSet(Size=0, BitSetCount=0
Services
TcpRing{TcpSocketAccepter{State=STATE_OPEN, ServerSocket=192.168.0.100:8089}, Connections=[1]}
ClusterService{Name=Cluster, State=(SERVICE_STARTED, STATE_JOINED), Id=0, Version=3.5, OldestMemberId=1}
InvocationService{Name=Management, State=(SERVICE_STARTED), Id=1, Version=3.1, OldestMemberId=1}
DistributedCache{Name=BackService, State=(SERVICE_STARTED), LocalStorage=enabled, PartitionCount=1021, BackupCount=1, AssignedPartitions=0, BackupPartitions=0}
ReplicatedCache{Name=ReplicatedCache, State=(SERVICE_STARTED), Id=13, Version=3.0, OldestMemberId=1}
InvocationService{Name=DefaultInvocationService, State=(SERVICE_STARTED), Id=14, Version=3.1, OldestMemberId=1}
DistributedCache{Name=InputBackService, State=(SERVICE_STARTED), LocalStorage=enabled, PartitionCount=1021, BackupCount=1, AssignedPartitions=1021, BackupPartitions=0}
The pc has multiple network cards: One on 10.101.16.182, one on 192.168.0.109
This is the ipconfig output (Sorry for the Chinese, I don't understand it, but this is at a customer site):
C:\Users\Administrator>ipconfig
Windows IP 配置
以太网适配器 本地连接 4:
媒体状态 . . . . . . . . . . . . : 媒体已断开
连接特定的 DNS 后缀 . . . . . . . :
以太网适配器 本地连接 3:
媒体状态 . . . . . . . . . . . . : 媒体已断开
连接特定的 DNS 后缀 . . . . . . . :
以太网适配器 本地连接 2:
连接特定的 DNS 后缀 . . . . . . . :
本地链接 IPv6 地址. . . . . . . . : fe80::783d:f432:79d0:15e6%12
IPv4 地址 . . . . . . . . . . . . : 192.168.0.109
子网掩码 . . . . . . . . . . . . : 255.255.255.0
默认网关. . . . . . . . . . . . . : 192.168.0.1
以太网适配器 本地连接:
连接特定的 DNS 后缀 . . . . . . . :
本地链接 IPv6 地址. . . . . . . . : fe80::ccaf:a85:eb9:90f9%11
IPv4 地址 . . . . . . . . . . . . : 10.101.16.182
子网掩码 . . . . . . . . . . . . : 255.255.255.0
默认网关. . . . . . . . . . . . . : 10.101.16.181
隧道适配器 isatap.{4137279E-1278-4A28-985F-A9DFCF8827BE}:
媒体状态 . . . . . . . . . . . . : 媒体已断开
连接特定的 DNS 后缀 . . . . . . . :
隧道适配器 isatap.{C2A0F541-74DF-4C45-8F96-3442ED0817CD}:
媒体状态 . . . . . . . . . . . . : 媒体已断开
连接特定的 DNS 后缀 . . . . . . . :
隧道适配器 isatap.{982A9A61-F542-4CC2-812D-6A3A4D0F7749}:
媒体状态 . . . . . . . . . . . . : 媒体已断开
连接特定的 DNS 后缀 . . . . . . . :
隧道适配器 isatap.{EFBCBB08-33F9-426A-A690-0CD8A6B69650}:
媒体状态 . . . . . . . . . . . . : 媒体已断开
连接特定的 DNS 后缀 . . . . . . . :
隧道适配器 Teredo Tunneling Pseudo-Interface:
连接特定的 DNS 后缀 . . . . . . . :
IPv6 地址 . . . . . . . . . . . . : 2001:0:4137:9e76:144b:15a4:3f57:ff92
本地链接 IPv6 地址. . . . . . . . : fe80::144b:15a4:3f57:ff92%19
默认网关. . . . . . . . . . . . . : ::
C:\Users\Administrator>
I also had a look at this documentation: http://docs.oracle.com/cd/E14526_01/coh.350/e14527/installcoh.htm
but it does not help much. The last line says "If all the preceding options fail, set up Coherence to run on a single host." but no link to documentation on how to do that.
regards,
Wim

Answer 1

user10435421 wrote:
Hi,
We are using Coherence 3.5.3 and have 3 windows services running that form 1 cluster on a single windows 2008 machine. The TTL is set to 0 since there is no need to leave the physical machine as the 3 JVMs run on the same pc. Could it be that the connected network still has an influence? If TTL is 0, does the communication between the JVMs (and thus the nodes of the cluster) still go over a network card ?
Yes, it can have an effect. Let's suppose that there is another cluster on another box which uses the same cluster address, but does not have TTL set to zero, thus its packets can reach your machine. You however won't have a chance to reply due to your sending with TTL=0.
In this case those packets can cause issues e.g. with forming the cluster on your machine because you will see that there is another cluster somewhere with the same address, but you can't communicate with it. This can prevent a cluster starting up on your machine.
Even if this happens after your cluster is formed, it may still cause error messages in the log mentioning that you received a heartbeat message from a different cluster on the same cluster address.
An easy workaround would be to choose a different (unused) cluster address for the cluster on your machine.
Best regards,
Robert