You are answering your own question but you haven't connected the dots yet. MGMT port = OOB network. Front ports = In-band network. The two shouldn't meet. You shouldn't be connecting them to one another. If you are managing "in-band" then configure an in-band SVI/IRB interface.
In Mist, it's this section:
The yellow highlighted is for configuring the in-band mgmt stuff as a virtual-interface. Select the VLAN you want MGMT to be on and configure it there. I've used both DHCP/static, works fine.
The green section is where you can adjust the OOB interface to behave differently/etc. Having both mgmt and in-band active at the same time can cause headaches for accessing the Mist portal for the switch. It will choose either the in-band or the mgmt default route to get to Mist. Pressing that tick box for "Dedicated Management VRF" will basically make the switch keep Mist traffic just to the MGMT port interface. I'm just going to point out... you're playing with fire trying any this stuff out remotely for the first time... test it out in a lab.
The switch is likely seeing itself via LLDP neighbors and is saying "we have a loop". Have I seen it before? Yes. It doesn't happen always... confusingly. Usually see it when installers plug it in as you've described (mgmt port to front ports) not realising the mgmt port plugged into the old switching network was just there for remote staging. The loop is not the same as a BPDU block (mgmt port doesn't do STP; as it's not a switchport).
Save yourself a lot of headaches, get in-band management working via the in-band SVI interfaces. Leave the mgmt port alone as a "the port I use to stage switches" or a "I totally broke the inband network and need to jerry rig a way to cloud manage this switch remotely" option.
The switch doesn't like having mgmt and in-band default routes at the same time.
------------------------------
Jonathan Bekkers
------------------------------
Original Message:
Sent: 05-14-2025 16:17
From: NICHOLAS MARSZALKOWSKI
Subject: vme.0 causing switch loops.
That aside, the MGMT port is there for OOB management, but is causing switch loops when used. Just curious if anyone else had any experience with this.
------------------------------
NICHOLAS MARSZALKOWSKI
Original Message:
Sent: 05-14-2025 15:58
From: asharp
Subject: vme.0 causing switch loops.
I've not had to use a vme.0 or mgmt port for my Mist lab setup, I'm just connecting from ge-0/0/0 on a EX4100-F to a ge port on my WAN Edge (SRX320 in this instance).
Regards,
Original Message:
Sent: 05-14-2025 13:25
From: NICHOLAS MARSZALKOWSKI
Subject: vme.0 causing switch loops.
That doesn't make sense to me.
Mist, by default, relies on the physical MGMT port in the back of the switch, which is vme.0. We use this for OoB management.
SSH (or any management) isn't even enabled by default on our IRB interfaces.
------------------------------
NICHOLAS MARSZALKOWSKI
Original Message:
Sent: 05-14-2025 01:22
From: PER GRANATH
Subject: vme.0 causing switch loops.
The 'normal' way of doing this is to not use the vmd.0 at all, but to have an irb.0 interface in VLAN 15, and as a DHCP client. That's the factory default config of the switch, although using VLAN 1.
------------------------------
PER GRANATH
Original Message:
Sent: 05-13-2025 12:57
From: NICHOLAS MARSZALKOWSKI
Subject: vme.0 causing switch loops.
Good Morning.
We currently have a setup with 2 EX2300 switches within an IDF - SW1 uplinked back to our distribution switch SW2 uplinked to SW1. Port 44 on each switch is untagged for VLAN 15 (mgmt). We have the MGMT port on each switch connected to port 44 on itself. vme.0 pulls it's IP via DHCP currently.
These switches are added to and managed through the Mist portal
Recently, I noticed that from time to time, Mist will report a switch loop between vme.0 and port 44. With vme.0 acting as an endpoint/client, i'm not sure what is causing this.
I reached out to support who originally said we shouldn't be connecting the MGMT port to what they called a "revenue port", but using a switch port is the only way we can connect these devices to the web, as we don't have direct connections back to our management switch. It was indicated that the fix is to statically assign an IP or to create a mgmt VRF for vme.0. Neither of these two solutions make sense to me as the issue appears to be at layer 2.
Any insight would be welcome.
------------------------------
NICHOLAS MARSZALKOWSKI
------------------------------