We defined our DHCP helpers as follows
x.x.x.x with associated Tenant A/EPG aaa ( in POD1)
y.y.y.y with associated Tenant B/EPG bbb ( in POD2)
we then assigned these helpers to Tenant B/EPG ccc in POD2
the requests work fine to helper y.y.y.y, which is in the same Tenant/POD but never make it to x.x.x.x, which is not in the same Tenant/POD
when we changed to:
x.x.x.x with assiciated Tenant B/EPG bbb (in POD2 ) the DHCP requests make it to x.x.x.x
Is this a known restriction in Multipod/Tenant. Basically it appears we need one set of helpers per Tenant or POD?
Where did you configured the DHCP Relay Policies? If you want to use inter-tenant DHCP relay, you need to create a global DHCP relay policy, which is from Fabric -> Policies -> Global -> DHCP Relay.
Regarding the number of DHCP relay labels per BD, I do not see any limitations in the config guide, but with these features, unfortunately, there is not enough documentation.
Ours are defined at: Fabric > Access Policies > Global > DHCP Relay
I do not see an option in 3.2 under Fabric > Policies > Global for DHCP Relay
The path which I shared is from 4.2, but is ok, is the same thing as you mentioned.
So it could be possible that there is a limitation about the number of policies you can have associated to one BD. However, I would suggest you double check with the TAC since there is no mention about this limitation in the cfg guide. If they say it's only limited to one, then they should open a documentation defect to update the guides. If they say is not, then it's a bug in the software version you're running.
Let us know about their feedback.
What is the VRF design here with regards to where the Helpers & Clients are? If things work fine when the Helper & Clients are in the same VRF, but not when in different VRFs it's likely an issue with your DHCP server not supporting Option 82 or that you haven't leaked the Helper Subnets into the Clients VRF correctly. Shouldn't be any limitation with DHCP relay across Pods otherwise.