[Beowulf] odd vlan issue

[Michael Di Domenico]

thanks after two days of digging, i think i finally figured out that
we have a layer 2 routing problem.  i'm not the network guy so i'm not
digging into it deeper, but it appears that there are either
malfunctioning LACP trunks or more likely a misconfigured VPC
connection inside the menagerie of switches the network team built.

the network is too complicated to describe here, but the base issue is
that there are two switches 'supposedly' operating jointly, but don't
seem to be sharing their CAM/ARP tables correctly.  for whatever
reason packets get duped to the switch that does not have the
destination machine and since there's no arp/cam entry the switch just
blasts the packet out all the ports.

its not clear why the packets are being sent to layer 2 devices where
the device doesn't exist, but it's clear there's something broken in
the spanning tree database.  it's also not clear why it only affects
one of the vlans and not all.

but again, not the network guy...  and for once it is the network... :)









On Wed, Jun 16, 2021 at 12:48 AM Greg Lindahl <lindahl at pbm.com> wrote:
>
> On Mon, Jun 14, 2021 at 12:38:50PM -0400, Michael Di Domenico wrote:
> > i got roped into troubleshooting an odd network issue.  we have a mix
> > of cisco (mostly nexus) gear spread over our facility.  on one
> > particular vlan it's operating as if it's a hub instead of switch.
>
> I have run into this situation when I have servers that have incoming
> UDP traffic and never talk or do TCP. The switches have no idea where the
> server is, so they broadcast all of the incoming packets.
>
> An ARP reply or connecting with TCP tells the switch which port to use.
>
> -- greg
>
>