A5800 JC099A - Devices will not connect unless tied to an additional unmanaged switch?

[Leuthcorp]

I am encountering an interesting issue.

I repurposed an A5800 JC099A, bumped it to the most recent firmware, and proceeded to put in the simple configuration needed for the location.

There was a need for a trunked connection to an ancient (and I do mean ancient) Linksys SRW-2008. I managed to get logged into it and verify the settings. Trunks for vlans 10 & 20 off of port 8, access ports 1-5 for Vlan 10, 6&7 for Vlan 20.

I gave it a static management IP on vlan 10.

At first I thought my issues were tied directly to this ancient piece of OfficeSpace Target Practice, but I appeared to be wrong in the end.

I got everything functioning on the bench, everything looked good. I saved the configuration, saved the configuration as an actual name as a backup, waited for it to save, saved again, shut it down (pulled the power) and proceeded to wire it up.

Once it was online, I started connecting the ports, naming them as I went, and saving the configuration in the process. So far so good, until it wasn't.

Nothing would connect anymore. Everything on the network was down. Again, I thought it was the SRW2008. But the weirdness comes from devices plugged directly into the switch. There's a smart tv that could not connect, and several computers that could not connect.

Until I plugged them into an unmanaged switch tied to the A5800.

This makes literally no sense to me. With a Cisco, I would have run into problems immediately with this. But no, I have a simple configuration on 1/0/19:

interface GigabitEthernet1/0/19
port link-mode bridge
port access vlan 10
undo lldp enable

If I plug a SINGLE device into this port, it will no longer function.

But an unmanaged netgear 5 port gigabit switch seems to allow everything to connect? Don't get me wrong, I am happy that the unamanaged switch works without fail. It's great for a small network like this with a low budget, but.... what? The point fo the switch itself is to be ya know... the *main* switch.

Configuration:

https://pastebin.com/H2r8APG8

Router-->A5800-->Unmanaged switch-->Devices = Works

Router-->A5800-->Devices = Not Working

Router-->A5800-->SRW2008 (Ancient Linksys)-->Devices = Not Working

[parnassus]

Interesting.

Who is performing IPv4 Routing on your network? I suspect it is the "Router" uplinked from the Interface 1/0/23 of your HP 5800 (that interface is tagged on VLAN id 1, 10 and 20 so it carries those VLAN id up/from the Router LAN interface).

Access interfaces are untagged (port access vlan 10) with PVID id 10 so are accepting any untagged incoming traffic (and once accepted those packets are internally tagged with VLAN id 10 into the Switch 5800) and are removing the VLAN id 10 tag of any outgoing traffic (packets leaving the access ports are egressing without a VLAN id tag).

A peer Switch (Managed) should be configured with proper VLAN tagging if it is connected among the 5800 and the final host...instead an unmanaged Switch is going to not understand VLAN id tagging so it accepts whatever tagging the packets arrive with...if then the final host accepts untagged traffic (because it is VLAN unaware) then the traffic flows as you see...but, instead, if a packets will arrive with traffic tagging then the final host would drop it (if the ethernet card is not configured to be VLAN tagging aware).

I expect that a typical VLAN unaware host connected directly to, as example, HP 5800 Interface 1/0/19 will be able to ping any IP Addresses within the very same VLAN id 10 IP address space and, for routing to other subnets, the Router should chime in (return routes to VLAN 1, 10 and 200's subnets and access policies to accept traffic from - and between - those subnets once its inter-VLAN routing feature is enabled).

So when you wrote "Nothing would connect anymore. Everything on the network was down." you need to specify the what, when and who...saying an host doesn't connect anymore means nothing if you don't specify against what it was connected before (Ping KO to Router LAN interface on a particular VLAN? The internet - say 1.1.1.1 or 8.8.8.8? or what else?)...and writing the Network was down are you meaning the interfaces lost their links (Layer 1) or what else?

[Leuthcorp]

I thought for a moment I had found a solution. Sometimes switches can be finicky. I set each port to speed auto 1000, and several devices lit right up. Autonegotiation wasn't working. 

The Samsung Smart TV for the Lounge area refused to come online no matter what settings I put in... I brought it back online by plugging it into an unmanaged switch tied directly to the router.

The router itself is a custom built OPNSense, which had been working just fine (and appears to be working...just fine).

The unmanaged switch that I had several machines plugged into was working off of a drop. I swapped it out for an 8 port (another netgear unmanaged gigswitch rather than the 5 port). Everything died immediately and refused to come back. If its an unmanaged switch, it shouldn't be remembering much of anything. 

To make things more interesting: I can ping one of the computers sitting on said unmanaged switch:

.....and I now feel exceedingly stupid.

I told windows 10 to fix the connection. It forced the gateway in, and the connection worked again.

So there's that.

Still having the issue with the Smart TV and several other devices that don't want to operate properly.

When I said that nothing would connect anymore, that was everything plugged directly into the switch.