L8x Redundancy with/without IP Address swapping

arendcv

Member
Join Date
May 2018
Location
Perth
Posts
2
Hi,
Looking at installing a L8x redundancy system and have perused the manual.
Yep version 1 (beta..) on L8x, probably not the best idea, as well as the limitations associated with it, embedded Eth port disabled, single controller per chassis etc...

I am currently using produce/consume unicast across subnets for interPLC communications and this is an issue. So to avoid creating a flat layer 2 subnet across the site for communications purposes I assume I would have to revert to using messaging and some code to determine which read MSG data to use when reading from the two redundant controllers (without IP swapping). Possible the read MSG from the secondary may fail anyway.

I could create a single subnet for the two redundant controllers to permit IP address swapping. Assume the MSG function may work better (more seamless) with this approach, and possibly the SCADA with a single IP to hit. Although I am sure the SCADA will have some lag when failing over.

Is anyone on this forum running without IP swapping or using MSG rather the produce/consume who can add to this?
 
I would not recommend you to read ANYTHING from the secondary processor, it’s memory is under full control of the RM2 module and constantly been updated by crossroads.

I would use IP swapping, since you will always read from one IP address without worrying who is primary.
Produced/consumed connections will be bumpless, MSG will have some delays.

Also with L8 redundancy you need to follow Chapter 6 of the user manual to get perfomance improvement vs L7.
 
I have this setup and working in a lab.
First thoughts are the RM2 module (or the backplane) is not up to the task of an L84/L85, makes sense why AB restricting only one controller for this initial release.

I'm only using 40% of the controller memory and the scan times for each task are 10x to 20x what they were prior to enabling redundancy, and that is after disabling the sync on every program within the task. No I/O connected.
Without recoding everything the project basically follows the recommendations of the manual at a glance, only periodic tasks (4), small number of program files/task (max 8), usually 1-2, no unused tags, all UDT/Array tags. Sure I could probably sharpen up some UDTs and improve memory usage.

Had to wind all the periodics out and watchdogs, so no task overlaps. Via the L_CPU AOI task usage is showing 52% unused on the CPU. User tasks 38%.

Have updated firmware of the RM2 to 20.030, will try replacing the fibre patches between the two RM2s, just encase there's an issue there however unlikely as both channels are connecting and its not switching channels.

Any other settings worth checking that would make some decent improvements?
 

Similar Topics

Hi all. Customer wants analog faceplates really bad, even if we explained that it doesn't make much sense in his process. What he wants to see...
Replies
5
Views
134
Hi all. I'm trying to set up communication between a 1756-L72 (EN2TR) and a 1756-L82 (CPU port) using produced / consumed tags. From the L72...
Replies
13
Views
379
I noticed something that’s mostly just a curiosity. Maybe someone here knows. An IT system is reading a L82 with the on-board IP and 1,0 added...
Replies
2
Views
705
Good afternoon folks, We have twin PowerFlex 755 480V, 14A AC drives (v14.005 fw) with embedded Ethernet that perform the same function side by...
Replies
8
Views
2,517
I'm trying to find clarification from the Rockwell documents what it means to an L8x that controller connections are "not applicable" as in L7x...
Replies
0
Views
1,581
Back
Top Bottom