LINX Manchester Peering Thursday 11th October 2018 16:00:00


A LINX Manchester member has incorrectly configured their peering device and have flooded the peering LAN. We have temporarily disabled peering at this Exchange.

Shortly before 16.00 (UTC+1) yesterday LINX were performing a routine provisioning task by moving a member from the quarantine VLAN to the production LAN on edge3-tcw. The member, through misconfiguration had enabled OSPF on their peering interface. The rate of the OSPF Hello’s was 1 packet a second, although prohibited on the LAN, ordinarily this would not cause stress on a modern router. However, in this particular instance it looks like LINX hit a known issue where a Multicast or Broadcast packet can become replicated by constantly looping around the EAPS ring.

The source IP of the OSPF packet that some members were seeing (172.21.1.90) is from the quarantine Vlan which suggests the replication may have begun slightly before, or at the time of migration into the production LAN. Our provisioning process has checks for non-MOU traffic on the quarantine LAN before moving into production, however this traffic was not picked up in this instance.

As mentioned above the issue itself looks similar to one LINX have seen in the past, but we will be raising a TAC case with Extreme for confirmation. We are also reviewing our quarantine process. The member in question has since been brought back into the production LAN without issue.

We've been advised the issue has been resolved and we have re-enabled our port at LINX Manchester.

There is no impact to end-users, traffic destine for LINX Manchester has been re-routed around Exascale's network to other Internet Exchanges in the U.K.