SUS on VCM

[ars_59]

Hi All,

I have some trouble on setting SUS.

My development environment have 1 enclosure with 2 VC. VC1 uplink port connect to port A1 on Procurve, VC2 uplink port connect to port A2 on the same Procurve. Both using 10 Gb.

I made 2 SUS. SUS_01 for VC1 and SUS_02 for VC2.

Both have same vlan (53, 60, 100). On the server profile, 1st NIC i forced vlan mapping to SUS_01 and 2nd NIC i mapping it to SUS_02.

On the Vcenter, I've set NIC teaming.

My question are. When I disable port A1. Why it wont failover to port A2?

[Hongjun Ma]

config looks good.

when you fail A1 port on procurve, on vcenter host networking config, do u see the vmNICx mapping to VC1 go down? It should go down as you have VC smartlink configured.

what's the FW version for VC and NIC? Is it G7 server?

you can ssh into VCM and do "show server-port enc0:1:d1", which means local enclosure:first vc module:first 10G server downlink, it will show if DCC(Smartlink) is available.

for server NIC firmware/driver info, please do

ethtool -i vmnic0 (or whatever NIC you are interested)

[ars_59]

when i'm disable A1, it's all down.

Yes, I'm using VCM ver 3.3 and G7.

ethtool -i vmnic0 (or whatever NIC you are interested) --> where i can do this?

i've tried other scenario, i used 1 SUS for all port.

So i will have 1 active port and 1 standby port.

when i'm disable the active port.... again, it's fail to failover.

[Hongjun Ma]

ok. are you saying that if you disable A1 port on procurve, your SUS1 on VC always shows "up"? If so, that's something to do with DAC cable or other layer 1 stuff. it has nothing to do with VC config.

after you disable A1 port, if you wait for some time and see SUS1 status in GUI, does it still show up? Also, in VC CLI, please do "show uplinkport enc0:1:X1" some time after you disable A1 to check the link status. This is assuming you are using first 10G port on VC to connect with procurve A1, please replace it with the right port number in your case.

also, at the same time, please do "show interface brief" on Procurve to prove A1 status is down on Procurve side.

If you see procurve side is down while VC SUS1 is still up after some time, then it's some layer 1 issue likely to related with DAC cable.

what's the part number for DAC cable? is it on the VC and procurve support list?

For VC flex-10, the supported DAC cables are at:

http://h18000.www1.hp.com/products/quickspecs/13127_na/13127_na.HTML

For VC flexfabric, the supported DAC cables are at

http://h18004.www1.hp.com/products/quickspecs/13652_na/13652_na.HTML

ethtool cmd is under Vshphere cli.

[ars_59]

ok. are you saying that if you disable A1 port on procurve, your SUS1 on VC always shows "up"?

>> yes

If so, that's something to do with DAC cable or other layer 1 stuff. it has nothing to do with VC config.

 >> but if it's something wrong with DAC cable, when i used A1 or A2 port on procurve it shouldn't have connect at all, right? But it's connect just fine. The problems are when i used it to failover.

after you disable A1 port, if you wait for some time and see SUS1 status in GUI, does it still show up?

>> yes, it's still up.

Also, in VC CLI, please do "show uplinkport enc0:1:X1" some time after you disable A1 to check the link status. This is assuming you are using first 10G port on VC to connect with procurve A1, please replace it with the right port number in your case.

>> i'm using port x5 on left and port x6 on right

 when i do the config it show like this :

->show uplinkport enc0:1:x5
ID             : enc0:1:X5
Enclosure      : Enclosure-Demo
Status         : Linked (Active) (10Gb/Full)
Type           : SFP-DAC (external)
Speed          : Auto/Auto
Used By        : SUS_01
Label          : X5
LED State      : green
LAG ID         : 27
Connected From : -- --
Connected To   : -- --

->show uplinkport enc0:2:x6
ID             : enc0:2:X6
Enclosure      : Enclosure-Demo
Status         : Linked (Active) (10Gb/Full)
Type           : SFP-DAC (external)
Speed          : Auto/Auto
Used By        : SUS_02
Label          : X6
LED State      : green
LAG ID         : 26
Connected From : -- --
Connected To   : 00:24:a8:d6:3d:80 (26)

>>>> is it the problem? thw connected to and connected from on port X5 is empty.

also, at the same time, please do "show interface brief" on Procurve to prove A1 status is down on Procurve side.

 >>> yes it's down.

  A1    SFP+DA3   | No        No      Down   10GigFD          off   0
  A2    SFP+DA5   | No        Yes     Up     10GigFD          off   0

If you see procurve side is down while VC SUS1 is still up after some time, then it's some layer 1 issue likely to related with DAC cable.

 >> what should i do then? replace the 10 GB cable?

But whats wrong with the cable? it's work properly when i'm not using it for fail over. can you tell me sir?

what's the part number for DAC cable? is it on the VC and procurve support list?

>>> i'm using HP 10GBE SPF+ 3M and i'm using HP 10GBE SPF+ 5M

[ars_59]

I'm sory i forgot that i'm disabling the port A1 when performthe config. This is the result when port A1 enable.

->show uplinkport enc0:1:x5
ID             : enc0:1:X5
Enclosure      : Enclosure-Demo
Status         : Linked (Active) (10Gb/Full)
Type           : SFP-DAC (external)
Speed          : Auto/Auto
Used By        : SUS_01
Label          : X5
LED State      : green
LAG ID         : 27
Connected From : -- --
Connected To   : 00:24:a8:d6:3d:80 (25)

[Hongjun Ma]

yeah, from capture, you do have the status of A1 down while SUS up(but without "connnected to" field, which is from LLDP, which shows LLDP communication is lost after you disable A1).

If you unplug the DAC from A1, does SUS1 go down?

Also, is it possible that you move the DAC cable between A1 and VC-1 X5 to two ports on procurve switch so it's a back-to-back looped cable from Procurve? make sure you have STP enabled to prevent loop. You should see one port is active and one port is logically blocked by STP but both ports should be up physically. Now try to do "disable" on one port, does the other procurve port go down? If you have two procurve switches, then it's even better that instead to loop two ports on the same switch, you can try to use DAC to connect these two procurve switches and do the same test. I'd like to see how this "disable" work on procurve switches.

The reason I'm doubting that this is some layer layer 1 cable/transceiver issue is that even though the cable works ok in normal condition, I feel that when you disable port on procurve, somehow the laser signal doesn't get shut off so on VC side, it can still detect RX laser and think the physical layer is good. it's just my guess because this problem should have nothing to do VC design and configuration. This sounds like something not right at physical layer.

[ars_59]

Mr.. Hongjun Ma,

I'm moving my environment.

In my new environment, I'm using 1 flexnic for 1 SUS (1 SUS for 1 VC).

VC1 : SUS1 (enc0:Bay1:X5), connect to port A1 Procurve.

Vlan : 600_sus1, 601_sus1, 602_sus1, 603_sus1, 604_sus1, 605_sus1

VC2 : SUS2 (enc0:Bay2:X5), connect to port A2 Procurve.

Vlan : 600_sus2, 601_sus2, 602_sus2, 603_sus2, 604_sus2, 605_sus2

Vlan Tagging, smartlink and Teaming has correctly set.

SUS set to Active - active.

I'm using Flexfabric and 10 Gb flexnic.

I tried to power off VC1 by pressing the virtual button, failover run correctly. (SUS1 status is not active)

When i tried to power off VC2, failover running well too. (SUS2 status is not active)

It's some good sign.

But when i tried to disable uplink port on Procurve, failover failed again (SUS status is active).
Why VC fail to acknowledge that the port on procurve is disable?

[Hongjun Ma]

Hi,

Power off module failure is very different than failure of SUS uplink. Right now I feel we need to focus on the cable/transeiver part between VC and Procurve.

Did you run the test I suggested in my last email? I'd like to see if VC SUS uplink can detect failure when you unplug the cable from procurve side, also I'd like to see if "disable" CLI can bring down the remote procurve switch through the same DAC cable.