EX2200 lost management interface after ethernet loop

Hi all

I had a loop on my network. Once I sorted that out, I discovered that three of my EX2200 switches were no longer reachable through the network. They are responsive to serial console connections.

One of those switches was in a lab, so I messed around with it for a bit but failed to get anything to work. I eventually rebooted it, and of course that works.

'restart management immediately' just seems to reload the active cli process. 'management' does not appear as a offered option when pressing ? after entering "restart".

The switch does report that the interface is up:

root@sa4-39> show interfaces me0
Physical interface: me0, Enabled, Physical link is Up
  Interface index: 1, SNMP ifIndex: 33
  Type: Ethernet, Link-level type: Ethernet, MTU: 1514, Speed: 100mbps
  Device flags   : Present Running
  Interface flags: SNMP-Traps
  Link type      : Full-Duplex
  Current address: f0:1c:2d:bc:f5:bf, Hardware address: f0:1c:2d:bc:f5:bf
  Last flapped   : 2016-08-15 14:41:15 EDT (01:45:26 ago)
    Input packets : 157777142
    Output packets: 18168943

  Logical interface me0.0 (Index 3) (SNMP ifIndex 34)
    Flags: SNMP-Traps Encapsulation: ENET2
    Input packets : 157777142
    Output packets: 18178196
    Protocol inet, MTU: 1500
      Flags: Is-Primary
      Addresses, Flags: Is-Default Is-Preferred Is-Primary
        Destination: 172.x.x/24, Local: 172.x.x.x, Broadcast: 172.x.x.x

All three affected switches are running 12.3R6.6.

All three affected switches are using out-of-band management, plugged into one of the ge- ports on the same switch

When the interface is plugged/unplugged, the switch records up/down activity on both the me interface and the ge interface that it is plugged in to.

Swapping the connection to another, similarly configured ge port has no effect; plugging other things into the ge port works normally. The switch appears to be switching properly; configuration changes made through the console appear to be taking effect properly.

Only one of the affected switches was directly involved in the network loop, although all switches saw the resulting storm.

Naturally, I would prefer not to reboot these switches. Does anyone know how I might resurrect the network connectivity of the management interface without a reboot?