Friday, August 17, 2007
Manchester to London Transit
Customers have made us aware this morning of connectivity issues from selected ISP's within the UK, which appears to have started at 8am this morning.
Investigation into the matter has proved difficult but has now been traced to one of our upstream providers advertising network routes to our network but then failing to deliver the traffic.
Under normal circumstances where a router connection fails, the advertisement to that route is removed automatically. The new routing table should then be updated to other routers that peer with it thereby preventing traffic being passed off to routers which cannot handle it.
We have closed down the peer with the London provider concerned and have raised the issue with them. When we have an explanation for the failed routes we will post further information here.
We would like to apologise to customers affected by this partial routing outage and the length of time it has taken to identify the source of the issue.
*** UPDATE ***
Further investigation suggests that the issue was restricted to a single /25 subnet used for routing transit to three customers and may have started some time after 11pm last night.
At this stage it appears that a reset filter on a single interconnect may have been the cause of the issue.
Investigation into the matter has proved difficult but has now been traced to one of our upstream providers advertising network routes to our network but then failing to deliver the traffic.
Under normal circumstances where a router connection fails, the advertisement to that route is removed automatically. The new routing table should then be updated to other routers that peer with it thereby preventing traffic being passed off to routers which cannot handle it.
We have closed down the peer with the London provider concerned and have raised the issue with them. When we have an explanation for the failed routes we will post further information here.
We would like to apologise to customers affected by this partial routing outage and the length of time it has taken to identify the source of the issue.
*** UPDATE ***
Further investigation suggests that the issue was restricted to a single /25 subnet used for routing transit to three customers and may have started some time after 11pm last night.
At this stage it appears that a reset filter on a single interconnect may have been the cause of the issue.
# posted by Netserve @ 9:55 AM 
