Running multiple clusters on a single server pairs on CitectSCADA 2018 R2

I had 3 clusters setup running fine in a test environment on V8.1 on the latest update. Today, I upgraded to v8.2, ever since, the alarms don't work, be the Example_SA project alarms work fine. I looked in the trace logs for the alarm server, there are countless errors with refused connections, the firewall is disabled on my machine too. It's not a privilege issue either.

2019-09-11 13:44:22.941 -07:00 21256 0 Error Transport TcpipTransport::EndConnect() [CLIENT 0.0.0.0:57459 --> 127.0.0.1:2082 #1] SocketException: No connection could be made because the target machine actively refused it 127.0.0.1:2082
2019-09-11 13:44:25.168 -07:00 27936 0 Error Transport TcpipTransport::EndConnect() [CLIENT 0.0.0.0:57472 --> 127.0.0.1:12082 #2] SocketException: No connection could be made because the target machine actively refused it 127.0.0.1:12082
2019-09-11 13:44:27.424 -07:00 15584 0 Error Transport TcpipTransport::EndConnect() [CLIENT 0.0.0.0:57492 --> 127.0.0.1:22082 #3] SocketException: No connection could be made because the target machine actively refused it 127.0.0.1:22082

I had to manually define the port numbers 12080 and 22082 for cluster 2 and 3 respectfully. Again, this work perfectly in v8.1.

I also get a hardware alarm from each of the running Alarm servers - Data Not Ready.

  • Hi Chris,
    It looks like the Alarm Server can't connect to the IO Servers. Can you confirm you have rebooted the machine after installing Citect SCADA 2018 R2?
    You also need to ensure that the Windows User accounts are added to the appropriate Citect User Groups depending on the role they play. See "Post Installation Configuration" in the help for information on configuring this.
    Did you run Configurator and setup security? If you want to disable security, you still need to set System Management Server to "Not Configured" and then untick encryption in the Encryption section.
    Finally, run the computer setup wizard and run again.
    What did you mean by manually define port numbers? In the project configuration? For each Alarm Server? Which column / field?
  • It looks like the Alarm Server can't connect to the IO Servers. Can you confirm you have rebooted the machine after installing Citect SCADA 2018 R2? Yes

    You also need to ensure that the Windows User accounts are added to the appropriate Citect User Groups depending on the role they play. See "Post Installation Configuration" in the help for information on configuring this. Done

    Did you run Configurator and setup security? Yes If you want to disable security, you still need to set System Management Server to "Not Configured" and then untick encryption in the Encryption section.

    Finally, run the computer setup wizard and run again. Done that.

    What did you mean by manually define port numbers? In the project configuration? For each Alarm Server? Which column / field? Topology>Edit>Alarm Servers>[Port,Legacy Port, Database Port].

    Oh, and I found that this is only an issue running as a Stand-Alone system, the issue goes away once I configure it to be networked, never had that issue on past versions.

  • G'day Chris,

    When you say it works in Network Mode, but not standalone. Are you making any configuration changes when switching between those modes? Is SECVS002 and 4 on the same computer (i.e. dual nic) or are they separate computers?

    Is the IO and Alarm configured to run on the same computer? (when running stand-alone)
    Are you having IO problems as well? Or just Alarm? Do trends work?

    Do you have the DNS Name field configured for the stand-alone computer? (and does it match the host name of the computer you are running on.) This is essential for encryption to work as certificate validation can only be done with host names.

    cheers,

    bradley
  • Hi Chris,
    I'm still a little confused how you could have compiled your project in Citect SCADA 2018 (8.10) without defining the Port, Legacy Port and Database Port for the additional Clusters. If you leave them blank, they assume the default value of 2080, 2076 and 5482 respectively. But in your configuration, you have 3 clusters all on the same machine, which if left blank, would all assume the same value and have a conflict. You would not be able to compile this project without receiving the error: "E2092 - A server with this network address and port already exists". So when running more than one cluster on the same machine, you must define unique ports. Have you modified the topology during the upgrade? As Bradley asked previously, it would also be good to know how many network cards you have and the number of machines in the system.
    It is very interesting that there is a different between running stand-alone ([LAN]TCPIP=0) and networked mode ([LAN]TCPIP=1). It is worth checking the DNS name field is set correctly.
  • I am using my local machine as a testing node for our upcoming deployment of a new SCADA system for our new site. Because we have two units (with identical tag names and screens) plus some common systems, we choose to go with three clusters. We will still only have two servers, physical nodes, on site once everything in in-place. So, I have setup the project to act like the primary server for all processes (I/O, Trend, Alarm, Report). SECVS002 in the primary server name and the network address is set to "localhost", SECVS004 is the secondary server and does not exist yet.

    The project configuration didn't change from testing on v8.1 to v8.2. I have in the past always set the Runtime to "Stand-Alone" in the Computer Setup Wizard, when I test/develop these new systems. When I run the project as "Stand-Alone", the Alarms don't work, as if the alarm server can't connect to the I/O server, this issue goes away when I run the project as "Networked". I don't have this issue when running the Example_SA project as "Stand-Alone".

    No IO problems, trending works fine, just no alarms. And I have discovered the PC name issue early on, so that was already remedied.
  • Olivier,

    We run a few systems with multiple clusters on the same server pairs, so we are familiar with setting the port numbers in this case. If you look at my screen shot above, you'll see the port numbers are blank for the first cluster, only the second and third cluster server pairs have the ports numbers defined. If there are better port numbers to use, please let us know, I just added a 1 front of the second cluster and a 2 in front of the third cluster.

    I have the PC name set to 'localhost', just like the 'Example_SA' project...that works in 'Stand-Alone' mode. That why I think it may be a multi-cluster issue, and maybe even those port numbers.
  • Hi Chris,
    Those port numbers look fine to me. I guess just double check the ones you configures aren't already in use on the same machine.
    Can you confirm what you have set for: Topology > Network Addresses > Address field? Is it "localhost"?
    Just so you know, "localhost" is not supported for the IP Address field. See extract from help:
    Note: "Localhost" and a fully qualified domain name (complete domain name for a computer) are not supported.
    I ran a quick test using "localhost" for the IP, and it works in networked mode, but not in stand-alone.
    I'd suggest setting the IP Address to 127.0.0.1 and the machine name in DNS Name setting in Topology > Computers.