We're receiving multiple complaints about problems reaching anything @bt. Is anyone else experiencing this? I.E. e-mails to @btopenworld and also @btinternet.com are bouncing. The root cause seems that we are unable to resolve anything within bt.com et al. For whatever reason, this is affecting all of the netblocks that we have nameservers in, but not some others we control. I.E. working ok from 204.188.96.0/20: % dig @194.72.6.52 bt.com | grep ^bt bt.com. 1D IN A 62.7.244.126 bt.com. 1D IN NS dns1.axion.bt.co.uk. bt.com. 1D IN NS ns1.bt.net. bt.com. 1D IN NS dns0.axion.bt.co.uk. Not working at all within 208.139.192.0/20: ns2# dig @194.72.6.52 bt.com ; <<>> DiG 9.2.1 <<>> @194.72.6.52 bt.com ;; global options: printcmd ;; connection timed out; no servers could be reached At the same time... ns2# tcpdump -n -s 1600 host 194.72.6.52 tcpdump: listening on xl0 10:40:07.235635 208.139.192.2.36647 > 194.72.6.52.53: 17495+ A? bt.com. (24) 10:40:12.241697 208.139.192.2.36647 > 194.72.6.52.53: 17495+ A? bt.com. (24) ns2# traceroute 194.72.6.52 traceroute to 194.72.6.52 (194.72.6.52), 64 hops max, 40 byte packets 1 cw-gw (208.139.192.1) 1.255 ms 5.908 ms 1.216 ms 2 denver-core-fe-0-0 (206.168.230.5) 7.285 ms 2.721 ms 2.597 ms 3 den1cspdaa02-v200.denver.terabeam.net (216.137.18.229) 4.406 ms 2.355 ms 6.978 ms 4 den1mcoraa01-ge110-v101.denver.terabeam.net (216.137.17.149) 1.261 ms 4.901 ms 7.538 ms 5 den1pcinaa01-ge020.denver.terabeam.net (216.137.17.21) 5.450 ms 6.128 ms 7.533 ms 6 12.124.158.25 (12.124.158.25) 1.163 ms 1.572 ms 4.476 ms 7 gbr1-p60.dvmco.ip.att.net (12.123.36.138) 2.529 ms 1.285 ms 6.739 ms 8 gbr4-p70.dvmco.ip.att.net (12.122.5.21) 1.365 ms 3.114 ms 4.32 ms 9 12.122.10.117 (12.122.10.117) 22.911 ms 19.350 ms 19.352 ms 10 tbr1-p012301.n54ny.ip.att.net (12.122.10.1) 38.500 ms 38.203 ms 44.327 ms 11 gar4-p300.n54ny.ip.att.net (12.123.3.2) 37.844 ms 37.614 ms 42.708 ms 12 12.126.124.82 (12.126.124.82) 39.463 ms 39.453 ms 39.360 ms 13 * * * 14 * * * 15 * * * 16 * * * 17 * * * 18 * * * 19 * * * 20 * * * 21 * * * 22 * * * 23 * * * 24 * * * 25 * * * 26 * * * 27 * * * 28 * * * 29 * * * 30 * * * I've fiddled with our routes here, tried from different netblocks to no avail. Don't see any mention of bt at the puck.nether.net NOC list either. Thought about sending them mail, but wound up back at step #1 above...
Mike wrote:
We're receiving multiple complaints about problems reaching anything @bt. Is anyone else experiencing this?
Grrr....Three days later, BT is now telling their customers that somehow, this is our fault. I find it rather odd that everyone in the world can reach us, *except* BT customers, yet it is our fault. NB, this only affects some of our address space (I received one confirmation report that another AS is also affected by this). I have tried four different peers, and it doesn't make a difference where I send outbound packets. I have sent mail to every address @BT that looks like it might possess clue, to no avail. This is a general plea for help- if anyone has an idea of how I might resolve this, I would be very grateful... From inside BT, looking at one of our clients colo'd servers: C:\WINDOWS\Desktop>tracert www.indianmatches.com Tracing route to www.indianmatches.com [204.144.133.198] over a maximum of 30 hops: 1 192 ms 110 ms 27 ms rase4nrp4.kingston.broadband.bt.net [217.32.11.81] 2 14 ms 14 ms 13 ms 217.41.195.2 3 27 ms 14 ms 14 ms core1-pos4-1.kingston.ukcore.bt.net [62.6.40.193] 4 14 ms 13 ms 14 ms core1-pos5-1.ilford.ukcore.bt.net [195.99.120.198] 5 14 ms 14 ms 13 ms transit1-pos3-0.ilford.ukcore.bt.net [194.72.20.226] 6 96 ms 82 ms 82 ms transit2-pos7-0.newyork.ukcore.bt.net [194.72.20.154] 7 t2c2-ge6-2.us-nyb.concert.net [166.49.150.29] reports: Destination host unreachable. Trace complete Well, no duh.... Going to www.concert.net I read "Concert, the AT&T and BT global communications venture, was legally closed on 1 April 2002". Ok, so WTF are they sending packets for us that way then??? Mike
Mike wrote:
I have sent mail to every address @BT that looks like it might possess clue, to no avail. This is a general plea for help- if anyone has an idea of how I might resolve this, I would be very grateful...
On a hunch, I shut down one of our peering sessions which was passing our routes to AT&T. Suddenly things seem to be working again... Thanks to all who've responded, I think we have a better idea of where the problem is now. Mike
BT operate 2856 in the UK, and 5400 internationally 5400 used to be concert which was a BT/ATT venture which is now solely run by BT under their brand - hence I guess why the msg on concert.net Email contact sent offlist.. Steve On Thu, 19 Jun 2003, Mike wrote:
Mike wrote:
We're receiving multiple complaints about problems reaching anything @bt. Is anyone else experiencing this?
Grrr....Three days later, BT is now telling their customers that somehow, this is our fault. I find it rather odd that everyone in the world can reach us, *except* BT customers, yet it is our fault.
NB, this only affects some of our address space (I received one confirmation report that another AS is also affected by this). I have tried four different peers, and it doesn't make a difference where I send outbound packets.
I have sent mail to every address @BT that looks like it might possess clue, to no avail. This is a general plea for help- if anyone has an idea of how I might resolve this, I would be very grateful...
From inside BT, looking at one of our clients colo'd servers:
C:\WINDOWS\Desktop>tracert www.indianmatches.com
Tracing route to www.indianmatches.com [204.144.133.198] over a maximum of 30 hops:
1 192 ms 110 ms 27 ms rase4nrp4.kingston.broadband.bt.net [217.32.11.81] 2 14 ms 14 ms 13 ms 217.41.195.2 3 27 ms 14 ms 14 ms core1-pos4-1.kingston.ukcore.bt.net [62.6.40.193] 4 14 ms 13 ms 14 ms core1-pos5-1.ilford.ukcore.bt.net [195.99.120.198] 5 14 ms 14 ms 13 ms transit1-pos3-0.ilford.ukcore.bt.net [194.72.20.226] 6 96 ms 82 ms 82 ms transit2-pos7-0.newyork.ukcore.bt.net [194.72.20.154] 7 t2c2-ge6-2.us-nyb.concert.net [166.49.150.29] reports: Destination host unreachable.
Trace complete
Well, no duh.... Going to www.concert.net I read "Concert, the AT&T and BT global communications venture, was legally closed on 1 April 2002".
Ok, so WTF are they sending packets for us that way then???
Mike
On Thu, 19 Jun 2003, Mike wrote:
I have sent mail to every address @BT that looks like it might possess clue, to no avail. This is a general plea for help- if anyone has an idea of how I might resolve this, I would be very grateful...
Point the customer at www.traceroute.org? HTH, Rich
Mike wrote:
We're receiving multiple complaints about problems reaching anything @bt. Is anyone else experiencing this?
Grrr....Three days later, BT is now telling their customers that somehow, this is our fault. I find it rather odd that everyone in the world can reach us, *except* BT customers, yet it is our fault.
You appear to be excessively deaggregating your space. Perhaps they are doing the responsible thing by filtering it? As announced, 13345 has 92 prefixes originated. After aggregation, there are 44 prefixes. There appear to be a couple of holes which you aren't announcing, which would further reduce this to 38 (after aggregation.) The space from the "Colorado Internet Cooperative Association" seems to be somewhat haphazard leading to the bulk of the remainder, and they aren't originating 207.174.0.0/16 at all, which may lead to issues. In this day and age of many people not doing the right thing, it is likely that one only see's the effects when looking at someone actually doing the right thing. I'ld suggest cleaning up your announcements and seeing if the problem persists. If nothing else, you become a part of the solution.
bdragon@gweep.net wrote:
You appear to be excessively deaggregating your space. Perhaps they are doing the responsible thing by filtering it?
I had a /20 from which BT was unreachable, and a /24 working just fine, so this seems doubtful, unless they are doing it to be spiteful and punitive alone. I say "had" because we have since withdrawn almost all announcements that AS7018 was learning from AS19694 and temporarily "fixed" the problem. It still exists, and it is clear now that the problem lies somewhere between AS7018 and AS2856. AS7018 says that my traceroutes are "confusing" to them, so I have pretty much given up hope of ever seeing a resolution to this.
As announced, 13345 has 92 prefixes originated. After aggregation, there are 44 prefixes.
Policy for most of those were set prior to my involvement with BGP here. I have managed to prevent the (otherwise forgone) deaggregation of 204.188.96.0/20 and 199.45.236.0/22 since my involvement at this level. I am also working to prepare an application for PI space from ARIN and the end result will be to reduce our total # of prefixes. Due to topology, aggregation of the remainder at this time would require several hundred clients to renumber immediately. Yes, I know this sucks. Yes, I am working towards a solution. No, it isn't going to happen tomorrow. Yes, I feel the pain every time I look at what we advertise. Mike
participants (5)
-
bdragon@gweep.net
-
Mike
-
Mike Lewinski
-
Stephen J. Wilcox
-
variable@ednet.co.uk