The issue is that Discovery needs to get more product knowledge so that it can distinguish Docker virtual interfaces, vs. others which use a similar naming scheme.
A work around would be to modify the getMACAddresses Linux script so that it checks for the docker process. If docker is present then run ip -o link show 2>/dev/null | grep -v '^[0-9]\+:\s\+veth' rather than the standard command to filter out the virtual interfaces.
There is an existing defect for this.
I had a Support case with a customer about this same problem (Also RHEL).
And, I created this Knowledge Article with the solution: Docker host constantly discovered as a new duplicate Host
Note from my previous case:
Customer changed his Platform script, and it is working fine now.
He did this:
1) edit the Linux platform script called "ip_link_mac"
2) change this line:
OLD: ip -o link show 2>/dev/null
NEW: ip -o link show | grep -v '^[0-9]\+:\s\+veth' 2>/dev/null
And, for the future, there is a Defect: DRUD1-19441
KA has more information than this. Please see the KA.
I notice now that I did not do any "if" logic in the platform script. ("if" docker host).
It worked fine without "if" logic. Though maybe there is a small performance drag for the non-docker hosts.
Looking back at the data I got from that case, I don't see how to check if it is a docker host. I only have this generic data from the docker hosts.
OS Type Type Red Hat Enterprise Linux UNIX Server
David Rough, since you have the data, if you see a way to add this "if" logic to the Platform script, please let us know.
This is wrong for a couple of reasons. First the redirection is in the wrong place. Second these interfaces can appear outside of Docker.