hp 2920 - i can't reach gateway whenever i connect the switch

[dbarroco]

planar network with no vlans or trunks, just a bunch of 3com switches uplinked to another switch where a cisco router (the gateway) also connects...

I preped 2 hp2920 with latest firwmare upgrade (incremental by the book until 16.01.0004), stacked them, created some vlans, added an ip to the vlan 1 and configured a default gateway (the router). also lowered the stp priority.

whenever i connect a single port to another switch, immediatly no one can reach the default gateway.... 

this port is in vlan 1 as well, no tags, no nothing, but impacts all the network in a 1 sec, 2 sec time at most. But still can reach ip's inside the network, except the gateway, wich in turn blocks access to the outside.

If it were stp, i guess it would block other lan segments. could it be arp related? stp? no routing enabled...

thank you

[16again]

Hard to tell without topology or configs.
What do you mean by "lowered STP priority."   Actually, setting this priority to lower value, increases the priority, which might cause the new switch to become STP root.

[dbarroco]

Hello,

The intention is to make it the stp root, and aggregate all the uplinks on this one. I was expecting to give a stp root value (4096 increment), instead it asked me to chosse priority, which I might have confused and give a lower value, which could have functioned the other way around, but then it would not make it root.

The topology is simple 1G uplinks, and I guess the router also. There are multiple routers. The main one (lan's gateway) chooses the path according to destination (eg, italy goes via router 2, germany via router 3, etc). No access to these...

if it is stp, the port enters in blocked state just by connecting this switch... that is the only explanation i find but i saw the routers with 1 uplink only, so i conclude that there is no physical loop...

I'm not seeing what cenario could make a router not to answer... i will back there on wednesday, i'm trying to rule out any firmware bug, or storms... etc. 

[Vince-Whirlwind]

I would double-check the config of the current "core" switch and the router, especially see if there isn't maybe some VLAN config you're not aware of.

Otherwise, we will need to see configs.

The first thing to rule out is something silly, like an IP address conflict.

The second thing would be to view the logfiles on the new switch, old "core" and router, when you connect the new switch and things go wrong.

[16again]

Indeed, without configs or logs , it's guessing what goes wrong. 
My guess is stp protection like root guard is used, blocking ports on receiving superior BPDUs

[dbarroco]

tried different ip's to rule out duplication. core isn't more than a 'dumb' switch where all get's uplinked, not much engineering going on ther.

Logs on the switch don't give anything away. I will ask to look into the router logs and the core switch.

Thank you for the hints

[dbarroco]

of course without loggs this is a shot in the dark. the routing is controlled by 3rd party on the HQ and by the weekend we had no  response from them...

i'm also inclined to stp guard, it behaves like rstp when you close the loop....i'm intentionally lowering the root value so this one is elected as root, but didn't realize you could protect this from happening. i guess this only happens on the router, bcause every other part of the network is ok. just the router goes off... 

i'll check it and update this. 

thank you

[dbarroco]

turns out to be STP allright. Still couldn't get an answer from the router team....for now going on with stp disabled.... tried to enable it with different priorities, all of them knocked the router out....stil not clear to me what it would be. I'll schedule a session with the team and log the events when enabling stp.

[16again]

You could try using bpdu filter on the port where the router is connected.  This effectively blocks all in/outgoing STP traffic on that port,  so the router will be unaware of you running your own STP instance