(Beacon) One server is creating a new multicast cluster. How resolve it?

[Anon Anon]

Hello All,

I having an multicast issue in my environment.

I have 8 app servers, but only one isn't including in multicast cluster created by nsa server.

All servers (A until G) are included in CLRTY-SA cluster, but server H is creating the cluster CLRTY.

 

Someone had this problem ??

Attachment(s)

log_inicializacao_servico_multicast_24112016.txt.zip   1012 B 1 version

app-ca_debug(PX02H).log.zip   7 KB 1 version

app-ca_debug_px02a(NSA).log.zip   5 KB 1 version

[Nika Hadzhikidi]

Hi Mauricio

Both CLRTY and CLRTY-SA are the names that are used by our application's jgroups to create multicast groups, and usually one does not listen to the other.
So I would suggest you check the following:

1. Multicast address and ports are the same on all servers, including the faulty one.
2. All servers should have a common CSA password
3. All servers should be in the same subnet. You can check this with ipconfig (Windows) or ifconfig (UNIX). If they are not on the same subnet this is going to be an issue.

 

If all the prerequisites are met, please confirm what made you think that the multicast does not work between them? We do not usually recommend setting org.jgroups to WARN in CSA, the recommended level is FATAL.

Did you set the logging level up to troubleshoot an issue? If not, and you don't have any other symptoms than the warnings, please ignore the messages and reset the logging level to FATAL in CSA.

 

 

I hope this helps -Nika

[Anon Anon]

HelloNika_Hadzhikidi,

Follows the answers.

1. Yes, are the same. Ip: 230.0.0.2 and port 9090

2. Yes.

3. They are in the same subnet (10.58.181.XX)

     NSA SERVER

   faulty one server

 

When we do one trace in NSA server, these not receive messages from H server.

 

Thanks.

[Aurora_Gaimon]

Try to comment "127.0.0.1" IP address (localhost) from "hosts" file in machine "H".

Restart services and check if problem is solved.

 

If it does not resolve issue, revert it back.

[Anon Anon]

HelloAurora_Gaimon,

 

We did this, but still server H not see the A.

 

 

 

Thanks.

[Aurora_Gaimon]

It looks like "A" and "H" have to 2 IPs:

 

H:

10.58.181.71
10.58.208.132

 

A:

10.58.181.72

10.58.208.123

 

You need to fill the "Bind Address" with all IP address (NSA and all client machines) coma separated, like this:

10.58.181.62,10.58.181.63,10.58.181.64,10.58.181.66,10.58.181.68,10.58.181.69,10.58.181.70,10.58.181.71

 

Update all properties.xml for all machines (NSA and client nodes) and restart CA PPM services. 

[Anon Anon]

HelloAurora_Gaimon,

 

The support told me that configuration it is only to versions earlier 13.3, i'm using the version 14.4 P4.
Is that correct ?

 

Thanks.

[Nika Hadzhikidi]

Hi Mauricio

 

You can still use the bind addresses - please do try adding all the NIC IPs in the cluster to the bind address in CSA/properties.xml, separated by commas with no spaces (as in Aurora's example above). Then save and restart and test. See if that helps. If that's not enough, we at Support will have to do some more troubleshooting with you on the case.

 

Regards -Nika