Stratix 5400 and 1783 ETAP DLR issues

Greetings,
I have a network topology that is using 5 Stratix 5400 switches that has 2 separate DLR networks configured. The DLR ports are set to only allow 1 specific VLAN for each ring. Both VLANs have the same IP address range but are segregated by using 255.255.255.128 for the subnet.

DLR 1 has no ETAPs but does have 5 Stratix 5700's in the network and I've had no issues throughout testing various network breaks. Stratix switch 1 is set for supervisor and Stratix switch 2 is set for backup. DLR 2 has 6 ETAPs in different segments of the rings network. An EN2TR is the only set supervisor and ALL ETAP dip switches are set to off. When breaking the DLR 2 link ports between Stratix switches, the network functions as expected. When breaking the link connection at any ETAP in the ring, the network initially functions as expected, 50 seconds later the network momentarily loses connection to ALL ETAPs for about 10-15 seconds, then reestablishes connection which then remains healthy.

I've since had to change the topology of the DLR 2 network to not go through the Stratix 5400's by using more ETAPs which now allows the network to function as expected when testing breaks in the ring. I then tried to use the Stratix 5400's non DLR ports as unmanaged switches to plug the DLR 2 ETAP's device port into which when doing this with 1 ETAP and 1 Stratix, functions fine. When doing this with a 2nd ETAP and a 2nd Stratix, the DLR 2 network becomes unstable and I begin to see erroneous communication losses.

Thanks for any insight to the issues I may be experiencing.

Have you confirmed that all the ETAPs have the device-level ring function enabled? Even if they are not configured as supervisors, they do need to be configured as DLR nodes. This can be confirmed with RSLinx or with the ODVA/Molex EtherNet/IP tool

I am not sure if I understood your architecture, is it possible to post a drawing?

It sounds like you have multiple rings crossing two Stratix 5400 switches. In this case it’s required to have two separate VLANs or set these two switches as Redundant Gateway.
Separation but subnet will not help.

In RSLinx, the only option I see on the ETAP under the network tab is to Enable Ring Supervisor which is unchecked. The ETAP states Supervisor Status: Normal ring node (supervisor not enabled). When selecting advanced, the only options there are:
-Enable IGMP Snooping (which is checked)
-Enable IGMP Querier (unchecked)
-Enable Device Port Debugging Mode (unchecked)
All other options are disabled due to the tap not being set to Supervisor.

Thanks for the feedback Alfredo.

I've attached a basic network drawing to give some clarity. The Stratix 5400's do NOT have redundant gateway enabled in any of the Switches as I understood that setting to only apply if I was to have 2 outside network connections to my rings.

I do have 2 separate VLAN's but the subnet of .128 will not help in this topology?

Thanks for the feedback Contr.

In this topology you must enable Redundant gateway. Because you have two exits to the same distribution switch. No exceptions here. You can’t have any devices connected to the Redundant Gateway switches.

Thanks Contr,

Unfortunately this system is back into production so I won't be able to make these modification till the next downtime.

To be clear, I need to Enable Redundant Gateway on all 9 Stratix switches. Then only select the Uplink port on switch 008 that is connected to the outside network? Precedence can be left at 0 for all switches because there is no redundant link to the outside network?

Thanks again.

I recommend that you read manual before enabling Redundant Gateway. This is not simple.
Assuming that you have two separate VLANs (must have), you need RG only on SW008 and SW009. One must be set as the active supervisor with precedence 255, another as first backup 254.
Another option is not to use redundant links to you servers (ie exit only from one switch to each server.

In any case, I believe that the same multiple rings can cross only up to two switches, you have way more than this.
I recommend to get in touch with local Rockwell office to get this reviewed by engineering.

Also, if you are planning to use DLR faceplate, then 3-Ring DLR AOI supports only a pair of 5400.
That is aligned with my concern that multiple rings can cross no more than 2 switches.

No DLR devices have been added to the controllers IO configuration at this time but the plan is to eventually bring those in. Due to limitations in testing this network and the sensitivity to this particular process, making any network modifications isn't worth the risk. Interesting that the 3 ring DLR AOI only supports a pair of 5400's due to the ability to assign multiple backup supervisor roles to the switches.

I have spoke with Rockwell Support on multiple occasions about this very issue and they did not state any concern with my current network configuration. Unfortunately, they have not been able to help resolve the issue at this time and support cases are opened, just awaiting responses.

I added a clip from the manual I've read about redundant gateways and it states that with only one connection to an outside network, redundant gateway is not needed. To further clarify the network drawing, Primary server is the PC hosting the FTV project configured as Primary with redundancy enabled and secondary is the PC running the secondary application. There is only one connection to the outside network on SW008.

Thanks again for the feedback and support on trying to get this resolved.

I am pretty sure that exiting from SW008 and SW009 to a single server is considered as a dual exit as far as I know. It creates issues is no managed via RG.

As for AOI, I know little more design insights for that AOI, so I can tell you that the reason you specified is not a valid, you can have multiple backup supervisors.
AOI documentation clearly stated that multiple rings can only cross up to two switches.

In any case, feel free to ignore my directions...

Definitely not ignoring your directions, I'm just trying to understand the issue at hand as to avoid repeating it. I'll research this AOI documentation that states multiple rings can only cross 2 switches then bring it to the design engineers attention that laid out this network originally. Rockwell Support has been lackluster and you've been very helpful.

Again, thanks for the support and once given the opportunity to test these changes I will follow up with those results.