HPE 5130 (JG936A): all four SFP+ ports stay down with JD092B modules and new OM4 cords
Second hand HPE 5130 for the rack at home, and it arrived with four HPE JD092B X130 10G SFP+ SR modules in the box. I bought new OM4 LC patch cords, wired one SFP+ port to the 10G card in the server, and got absolutely nothing.
- HPE FlexNetwork 5130 24G PoE+ 4SFP+ EI (JG936A)
- 4x HPE JD092B X130 10G SFP+ SR
- new OM4 LC duplex patch cords
- quad port SFP+ NIC in the server at the far end
web interface: all four SFP+ ports show down
no LED on any of the four cages with a module seated
the 24 copper ports on the same switch are perfectly happyWhat I have already tried:
- reseated every module and moved them around between the four cages
- used a different patch cord on each attempt
- moved the fibre to a different port on the server NIC
Same result on all four ports every time. Are these switch cages dead, or did I get handeed a box of dead JD092B?
Also available in: English
العربية
Deutsch
Español
Français
हिन्दी
Bahasa Indonesia
Italiano
日本語
한국어
Nederlands
Polski
Português
Русский
Türkçe
Tiếng Việt
简体中文
Comments 3
Do not condemn anything yet, isolate the switch first. Take one patch cord and loop SFP+ port 1 to SFP+ port 2 on the 5130 itself with two of the JD092B seated. If that pair links, you have proven the cages, the modules and that cord in one shot and the fault is on the far side. If you have a short passive DAC around, the same test with a DAC between the two 10G ports takes the fibre out of the equation entirely.
Then confirm the switch actually reads the modules:
If it comes back with vendor and part number for all four, the optics are alive and being read. A module that does not show up there is a different conversation.
Third, and this one catches people constantly: LC duplex patch cords are not always crossed. Plenty of them ship straight through, so TX lands on TX and nothing ever comes up. Swap the two fibres at one end and retry, it costs ten seconds.
Last, look at the far end with a phone camera. Point it at the transmit side of the modules in the server NIC. An SR laser usually shows up as a faint dot on a phone sensor, and no light from any of them tells you where to dig next.
On support: the JD092B are HPE's own, so the 5130 will not argue with them. Just be aware that if you ever move those modules onto an AOS-CX 6400 box they are not on the qualified list there, and you would be leaning on allow-unsupported-transceiver on AOS-CX 10.5 and later instead of a supported combination.
Worth adding a case where the modules turned out to be the guilty party, so nobody swings too far the other way. I was replacing a Brocade ICX 7250 with an identical unit, same firmware, same VLANs, same PoE settings. The SFP+ fibre uplink from a Proxmox host linked on the old switch and refused to come up on the new one, and moving it back brought it up again.
We went through the usual suspects: cable too long, a hairline crack that only just worked in one port, swapping the cable ends to see whether the fault followed the cable. In the end, replacing both Brocade SFP+ modules on the same OM3 run brought the link straight up.
So optics that link in one port and refuse in an identical port on another chhassis can still be the marginal component. Loop test first as described above, but keep a spare pair of modules to swap in before you start reading config diffs.
Results, and thanks, the order of tests was the useful part.
Looped port 1 to port 2 with one of the OM4 cords and two of the JD092B, and got nothing at all. Before condemning the cages I swapped the two fibres at one end, and the pair came up on the spot, so those cords are straight through rather than crossed, exactly as warned. Cages, modules and cords all fine, I just had to wire them the other way round.
display transceiver interfacelisted all four JD092B with vendor and part number too.That sent me to the far end. Held a phone camera in front of the server NIC with the modules seated and there was no light at all from the transmit side of any port. The quad port SFP+ card was dead. Replaced it, rebooted, ran the fibre with TX and RX swapped at one end the same way, and all four links came up.
The switch and the JD092B were innocent the whole time.