HPE 1920-24G: 4-port Trunk Link w/ 2 ports Errdisable - cabling configuration issue?

[burnbyAdmin]

Good morning all,

We have a HPE-1920-24G switch with a trunk link (4 Dynamic LACP ports) connected to an HP 1820. All 4 links worked previously, however due to expansion some cabling work was done (4 inline couplers similar to the following: http://imgur.com/1hvNQIB) were used to extend the current trunk links to the newly placed HPE-1920 which feeds 3 Netgear GS108T access switches.

2 trunk links work without issue (which currently feed the 3 access switches) but the other 2 links are disabled by the switch with the following error: 

The port's physical state (down) is improper for being attached.

Cable seemed to pass testing using a VDVMapMaster so I assumed the cabling job was adequate. I have tried setting both sides to dynamic/both to static but neither seems to resolve the issue.

Any suggesttions for the error message received?

[Vince-Whirlwind]

Is your testing testing all 8 wires?

Have you tried swapping the questionable cabling into known-good ports on both sides of a known-good link to see if the problem follows the cabling?

I wouldn't worry about what the switch is saying too much - if the only thing that's changed is the cabling, then the cabling is what needs looking at.

Have you tried nailing the ports on thwe 1920 to 100Mb instead of auto to see what happens to the dodgy links?

[parnassus]

It looks like a Layer 1 issue: what's about to verify LACP and each Port Trunk's member port status (particularly speed/duplex values) on both ends? maybe new cabling caused a speed value (re)negotiation - if speed setting is Auto - on a ports couple and that new value mismatches with speed value nogotiated on the other remaining ports couple...in other terms all eight involved ports, 4 per side, must share the same speed/duplex auto negotiated (or manually set) values.

Are all four physical links long the same lenght?

[burnbyAdmin]

I assigned speed/duplex settings on both sides but LACP negotiation still failed. 

To confirm I had settings set correctly I used a test switch (Netgear) to confirm LACP negotiation using new cables (1920 connected to NG SW using new cables with Dynamic LACP) and everything negotiated correctly; LAG was visible an in an up/up state.

Given 2 of the 4 links negotiate dynamic LACP with auto speed/duplex settings, I'm back to assuming the cables are faulty and will have to replace 2 of the 4 couplers. I do think it's unusual the cable seemed to pass the test - unless it was a fluke.

Out of curiousity, are inline couplers with punchdown blocks the norm? Or would stanard practice be to terminate both ends and use a female-to-female coupler?

[burnbyAdmin]

After setting up an additional test switch to confirm the software/configuration side was set correctly I tend to agree with you that it's a Layer 1 issue.

Will have to replace 2/4 couplers and see if it resolves the issue. All 4 cable runs are the same length from our server room to the warehouse HPE-1920. If I had to guess the total run length, I'd say all 4 CAT6 cables run no further than 250ft. 

Also tried manually setting speed/duplex but LACP negotiation still failed.

[Vince-Whirlwind]

I *really* wouldn't be troubleshooting this by looking at LACP negotiation.

You need to establish each link one by one and just get each port up and passing traffic before you look at any link aggregation.

[parnassus]

You wrote: "Out of curiousity, are inline couplers with punchdown blocks the norm?"

IMHO No, at least when used to "extend" a data cable run...simply because, once you use them (no matter if the total cable run length is less than 90 meters threshold), you will lose any Cat.5E/Cat.6 cabling certification eventually obtained on that cabling...

[Vince-Whirlwind]

I've used couplers before when I had no other choice. They don't always work - like any bit of physical wiring, there is always the potential for it to not provide full connectivity.