HPE 5130 10GBase-T ports not working

[Michal Doležal]

Hi,

I have two JG940A switches. Both working same. SFP+ ports are working, XGT ports not. Gigabit ports working without issues. Firmware level 3208 so the latest.

According to documentation those ports are independent, so I have 4 10G ports, two metalic and two optics.

I tested multiple cables (5e/6/6a) of different types with shield or without. I tested 1G NIC cards and 10G NIC cards. I tested multiple devices. In all cases nothing. I tested full reset of the switch (so not working also in factory default), I tried shutdown/undo shutdown command.

Problem is probably at layer 1 because there is no activity on the switch during cable insertion. I tried to run debug ethernet, I tried to virtual-cable-test I tried to setup manual parameters (speed, duplex nad MDI). Nothing.  There is no combo command and in display menu all looks normal. In logbuffer there is no data at given time.

I will contact support but I would like to hear if somebody knows this situation. In case of problem on one switch OK, understandable, HW problem, replaced, solved. But here it looks as some kind of SW problem because both are working same.

Device is connected to 2/0/27 (switch is out of IRF, only renumbered with clean config)

Brief information on interfaces in bridge mode:
Link: ADM - administratively down; Stby - standby
Speed: (a) - auto
Duplex: (a)/A - auto; H - half; F - full
Type: A - access; T - trunk; H - hybrid
Interface Link Speed Duplex Type PVID Description
XGE2/0/25 DOWN auto A A 1
XGE2/0/26 DOWN auto A A 1
XGE2/0/27 DOWN auto A A 1
XGE2/0/28 DOWN auto A A 1

interface Ten-GigabitEthernet2/0/27
#
return
[HPE-Ten-GigabitEthernet2/0/27]

Ten-GigabitEthernet2/0/27
Current state: DOWN
Line protocol state: DOWN
IP packet frame type: Ethernet II, hardware address: 5c8a-38cf-2a68
Description: Ten-GigabitEthernet2/0/27 Interface
Bandwidth: 10000000 kbps
Loopback is not set
Media type is twisted pair
Port hardware type is 10G_BASE_T
Unknown-speed mode, unknown-duplex mode
Link speed type is autonegotiation, link duplex type is autonegotiation
Flow-control is not enabled
Maximum frame length: 12288
Allow jumbo frames to pass
Broadcast max-ratio: 100%
Multicast max-ratio: 100%
Unicast max-ratio: 100%
PVID: 1
MDI type: Automdix
Port link-type: Access
 Tagged VLANs:   None
 Untagged VLANs: 1
Port priority: 0
Last link flapping: Never
Last clearing of counters: Never
 Peak input rate: 0 bytes/sec, at 2013-01-01 00:01:29
 Peak output rate: 0 bytes/sec, at 2013-01-01 00:01:29
 Last 300 second input: 0 packets/sec 0 bytes/sec -%
 Last 300 second output: 0 packets/sec 0 bytes/sec -%
 Input (total):  0 packets, 0 bytes
         0 unicasts, 0 broadcasts, 0 multicasts, 0 pauses
 Input (normal):  0 packets, - bytes
         0 unicasts, 0 broadcasts, 0 multicasts, 0 pauses
 Input:  0 input errors, 0 runts, 0 giants, 0 throttles
         0 CRC, 0 frame, - overruns, 0 aborts
         - ignored, - parity errors
 Output (total): 0 packets, 0 bytes
         0 unicasts, 0 broadcasts, 0 multicasts, 0 pauses
 Output (normal): 0 packets, - bytes
         0 unicasts, 0 broadcasts, 0 multicasts, 0 pauses
 Output: 0 output errors, - underruns, - buffer failures
         0 aborts, 0 deferred, 0 collisions, 0 late collisions
         0 lost carrier, - no carrier

[Michal Doležal]

It seems that this problem was already here. https://community.hpe.com/t5/Comware-Based/5130-firmware-versions/td-p/6987412

So I will contact support directly and will give feedback. If you have this already please post.

[VoIP-Buddy]

Michal,

I actually tested this code in my lab with probably the JG940A and it should have worked for you.  What do you have connected to the other end?

The SFP+ ports are separate.  They  have always worked.

Regards,

David

[Michal Doležal]

I tested:

Mac-Mini computer

Computer with Intel i217 card

Server with nc331T

Server with nc530T

and also both switches together with CAT6A cable

Nothing worked. Thanks for feedback. I created ticket to support.

[VoIP-Buddy]

Hi!

Interesting.  It sounds like the servers are some sort of HPE server?  Can you tell me what models?

In the switch to switch test, did you use the same model 5130 EI?  Same R3208 code?

Did you try doing a quick loopback test between both 10g ports on the same switch?  Do both ports show link?

Can you tell me what your support case number is so I can see where it is in the process?

If you didn't provide it, a display diag will be very important.

Thanks!

David

[Michal Doležal]

Hi it is interesting. I used different devices - computers with common nics (realtek, Intel i217, etc) and also server cards (as I wrote before, those are mainly Broadcom chips). But as I wrote before I tested also link between switches without IRF with proper cables (CAT6A) but also addition cable types were tested. Nothing worked.

I run debug commands, I monitored the log and no activityy from this side.

From HPE still don´t have feedback. (ticket set at wednesday). I saw that I am not alone with similar problem.

display diag attached, remove pls extension

Reference number: 5325603239

configuration is completely erased /factore default mode

[Michal Doležal]

Incredible solution. First, no contact from HPE but 3 days ago rang my phone and UPS delivery looked for me with package (no info before). It was switch for replacement. Ticket cannot be accessed at this time probably closed or something :(

Delivered switch is working as expected but I reported to HPE that both switches have this problem and I am suspecting firmware issues.

So I have working and non-working box so I tried to find where is problem and found.

It is really firmware problem but not explicitely 3208 but something was wrong in the switch. I completely delete everything from the switch (configuration, certificates, logs etc.) and upload and activate older version 3116P05. Reboot and olé switch is working as expected. I upgraded again to 3208 and still system works.

So warning for me for future. Don´t believe firmwares from factory and in case of problem try to reload all firmwares again. Tomorrow we have production start for these switches so I hope that everything will be fine.

[Michal Doležal]

Problem arrise again after some period. After firmware procedure in the lab I tested whole setup and no problem. We installed switch to customer and configured. 10-T ports stayed unconnected because we were waiting for new servers from factory. Today servers came and we connect all 10G ports (we have Quadport 10Gig nc536FLR-T card). None of ports came up..

I will contact support and give feedback here. Still suspecting firmware.

[VoIP-Buddy]

Folks,

I have some new information as I am currently working a case for another customer.  Engineering is about to release a new version of software for the 5130 EI switches that has an additional fix for the 10G Copper ports.  It is not yet available on the My Networking Portal but should show up sometime in the near future.  You want to look for R3208P03.  I have tested that in the lab with rebooting and power cycling and the ports come up with no issues.

My previous guidance was to use R3208 but my customer found that that was broken as well in regards to an unexpected power failure that upon restart made the 10G Copper ports unable to connect to anything.

You can also use R3116P05 as that works as well if you need something right away.  Both versions are in the same code stream.  But as I said, if you can hang on for a little while the R3208P03 should be available soon.

Regards,

David