Comware

Ok, so much of the equipment I have has this interface in the configuration

#
interface M-GigabitEthernet0/0/0
#

However I can find next to no information about its usage. I know I can do some assignments to it, but for instance, does this network get used in the "connected" routing table? Does it have other uses other than telnet/ssh/snmp queries?  What are the security concerns with the interface? Whats the difference on one of the comware routers (where there is a physical port associated with it) and switches (where it appears to be virtual only).

Unfortunately I'm sitting in London in a data center at the end of a long installation only to discover that these particular interfaces are acting differently on different pieces of gear... and with next to no mention/reference in the global internet nor in HP documentation...  Any help at this point would be appreciated...

Marcos

Howdy,

Management i/F appears in the "default" routing table instance unless you tell the switch otherwise. Many of my cutomers sacrifice a vpn-instance (vrf) to exclude it from the global table and put it on its own to make it properly "out of band" . As far as i rememebrr that's the same behaviour as the routers. Watch for any  differences between Comware v5 & v7.

I can't remember seeing a "virtual" management interface on a switch - if it shows up as M-gigx/x/x  its most probably a real port supposedly for proper OOB. IIRC it is a real L3 "routed port" with an IP instead of being a switch port with a bridge configuration (VLANs etc) on it and virtual L3 interfaces behind it.

As for the documentation

This works 99% of the time :

1. https://www.hpe.com/us/en/networking/switches.html

choose products and then switches or routers

drill down to you particular model - view all - pick a switch - product details

choose the resources tab

choose technical documentation

Ta-da !

Alternatively google for "hp manuals 5900" for example

"Setup & Install" tend to be Configuration Guides with examples rather than

"General Reference" which tend to be Command line reference stuff.

If there isn't a requirement for dedicated OOB management network - just turn them off / disconnect them. If there is then use them for SSH, SNMP v3, config backup etc.

Hope that helps.

Hi All,

I have an IRF pair of switches at the access layer and and IRF pair of switches used as an OOB Network Management tier. 

I was planning on using the M-Ethernet interfaces for true OOB MGMT but when they are IRFd the config shows (expectedly) a single management 0/0/0 for the IRF pair. 

Does this mean I can cable both but only IP address one of the M-Eth interfaces? What if I lose the switch with the management IP address configured, would this failover to the second switch in the IRF Fabric?

If this is the case then great, but if not I'd be better of using a regular ethernet interface on each and configuring and SVI on the switch for management (still OOB but not using the management interfaces which is what I really wanted to do this time round).

Any help appreciated

David

David,

You have it right. 

There is only one M-GE0/0/0 interface. It is associated with the physical port on the master in the IRF. The other physical ports are shutdown. Not only does it work but it is suggested to cable all of the management ports so that you always have access to the management port. 

This can get excessive if you have a stack of 9 switches. In that case, I recommend configuring two of the switches as a higher priority and then connect your mgmt cables to the mgmt ports on those two switches. 

Hi Eric,

old post, but maybe you're still connected ;)

I have the same issue. The problem is, as far as I understand, the M-GE0/0/0 is a logical port which is actually the mgmt port of the master switch of the IRF, whatever its up/down state is. It means that if the neighboring node goes down, or the port which HPE SW mgmt port is connected to goes down, the HPE SW is not reachable via its OoB mgmt address, right ?

I haven't found a workaround yet (except using,  for OoB mgmt, a regular port on both SW of the IRF).

Thanks for sharing your point of view!

 That would be correct.

To solve this problem you would need to use some of the "normal" interfaces and use a Management VLAN and connect them to different switches. To prevent a loop you would need to use MSTP. (Except you connect them to another IRF, then you could make a link-aggregation)

You could also use a VLAN for each different switch, to prevent L2 looping.

Yes , using regular ports was an option for me (neighboring equipment is a Juniper SRX, so I 'd use redundant-ethernet interface, it should be L2 loop free). However it means if I have a problem on traffic chain, I may loose also that dedicated O&M, because it would use the same chain, it is pretty close to inband management... I guess mgmt ports are a bit separated from the that, so if is real backup connexion.

I  found another config, but have no description of that :

interface M-GigabitEthernet0/0/0
ip address 10.153.32.17 255.255.255.240 irf-member 1
ip address 10.153.32.18 255.255.255.240 irf-member 2

I tested it : 10.153.32.18 is not reachable, even from SW IRF, even though port connected to mgmt port of SW2 is UP.

So I guess that address is active only when SW2 is master in the IRF. Do you know if this is right ?

(I can't test it because I haven't found a command to perform a switchover to SW2 ... if you have it, pls tell me ;)

If this is right, does it mean that in the original config :

interface M-GigabitEthernet0/0/0
ip address 10.153.32.17 255.255.255.240

10.153.32.17 is 'mobile', and will move to SW2 once SW2 is master in the IRF ?