January 2001 - Test - lists.nanog.org

clue for the clueless please
by Peter Galbavy 23 Jan '01

23 Jan '01

This is a Microsoft Passport / Hotmail access question, so if these are alien to you, please move on... Before anyone has a go at me, I have looked on their web site(s) and cannot find any policies or pointers that are outside "forgotten password" land etc. We have turned up a new customer for BGP multihoming with a new IP block from RIPE (217.71.224/20) and in testing we are having problems getting to certain sites. Normally I would attribute this to no reverse DNS (yet), but for Hotmail it is getting weird. The reverse DNS issue is in hand and is awaiting DNS servers to be up etc. The Hotmail issues (and maybe some other sites we have not yet got to) appear to be IP block related. Do "they" do some sort of geographical redirection or filtering based on IP block or AS number ? Testing from another IP (OK, the nearend point-to-point address on one of the routers) which also does not have reverse DNS but is in an older IP block, in a different AS works fine. The problem is the redirection from authentication -> mailbox. It just hangs. Tested with multiple browsers (lynx included ;) and all the same. And pointers would be greatly appreciated. -- Peter Galbavy Knowledge Matters Ltd. http://www.knowledge.com/

2 1

Re: Involuntary outages may start at 7am PST
by Henry R. Linneweh 23 Jan '01

23 Jan '01

PG&E is caught in its own self created mess. When Australia deregulated they were paying $89.00 per megawatt hour, part of the selling off of assets, in the frame of the contract it was stipulated that they would be able to buy power back from the buyer at a fixed rate the was predictable, and per that agreement there power costs dropped to $16.00 and $17.00 per megawatt hour. PG&E on the other hand had that option and decided to forgo it and buy on the spot market same with Southern California Edison. A major miscalculation. Next but not least there is the power exchange where power can be purchased at a better rate than from PG&E. There is plenty of juice on the grid, just not enough competitive marketing alignment to make it greater public knowledge. Anyway I am pooped out and I need some sleep so I will stop here for now. Roeland Meyer wrote: > > From: Nathan Stratton [mailto:nathan@robotics.net] > > Sent: Monday, January 22, 2001 7:27 PM > > > > On 22 Jan 2001, Sean Donelan wrote: > > > > > Have any Internet providers or private data centers announced any > > > voluntary "good neighbor" measures such as wider > > temperature and humdity > > > limits, lights-out operation, off-peak use of heavy > > electrical demands > > > for laser printers, etc. > > > > Don't take this the wrong way, but frankly I would not be > > happy if my colo > > providers started implementing wider temperature and humdity > > limits. I pay > > large amounts of money for colo and I want what I am paying > > for. This mess > > was caused by California regulators and very very greedy PG&E > > who gambled > > on lower rates and lost. PG&E should be forces to liquidate > > other out of > > state assets and buy power at the market. > > Unfortunately, you are wrong. PG&E is caught between a rock and the > generators. Generator cost has been gouged up over 700% and PG&E is forced > to maintain prices. There is plently of capacity if the generator companies > want to bring it online. Obviously, they don't, because they have a pretty > good blackmail hand right now. -- Thank you; |--------------------------------| | Thinking is a learned process. | | ICANN member @large | | Gigabit over IP, ieee 802.17 | | working group | | Resilient Packet Transport | |--------------------------------| Henry R. Linneweh

1 0

Re: UUNET peering policy
by smd＠clock.org 22 Jan '01

22 Jan '01

| Arguably the "first" peering battle was whether UUnet had to pay for access | to the College Park NSS in 1989-90, which predates the CIX. No, Dave, that's only the first commercial peering battle. There are plenty of earlier fights that would not have been on anything remotely like com-priv, and were not nearly as polite as the dialogues between Rick and Al... Sean. (who used to put the NA into NANOG, but not any more :-) )

1 0

Maintenance Windows for Networks Spanning Multiple Time Zones
by tex＠delamancha.org 22 Jan '01

22 Jan '01

Folks, I am looking for input from other providers as to how they handle maintenance windows when their networks span multiple time zones. I would imagine that common windows are not expected to span more than 3 or 4 timezones due to the difficulty of finding a large enough block of time that works for all of the time zones in that window. So, what do you folks do? -tex

1 0

Re: Wierd Route (fwd)
by Richard A. Steenbergen 22 Jan '01

22 Jan '01

On Mon, Jan 22, 2001 at 12:12:47PM -0600, Jon Stanley wrote: > > I still fail to understand. If I am correct, AS3908 and AS209 are > somehow related in that Qwest was required to sell off all long haul > circuits that go into the former USWest territory. I wasn't aware > that this impacted the Qwest IP backbone as well, however this has > been brought to my attention. What I don't understand is why AS3908 > can't aggregate before advertising to C&W, which then goes into AS209. c&w only? As seen from route-server.eodus.net: Network Next Hop Metric LocPrf Weight Path *>i63.145.64.84/30 209.1.40.52 1000 0 2548 1 3908 i *>i63.145.64.212/30 209.1.40.52 1000 0 2548 1 3908 i *>i63.145.65.32/30 209.1.40.52 1000 0 2548 1 3908 i *>i63.145.65.76/30 209.1.40.52 1000 0 2548 1 3908 i ... *>i64.47.0.117/32 209.1.40.51 1000 0 1 3908 14995 ? *>i64.47.0.164/32 209.1.40.51 1000 0 1 3908 14995 ? etc etc etc 64.47.0.117 = l0.mcr1.n1.lax4.broadbandoffice.net 64.47.0.164 = l0.mcr2.dcae.broadbandoffice.net loopbacks? If for whatever reason qwest feels the need to leak their /30's and /32's to bbn, it could at least be no-export. -- Richard A Steenbergen <ras(a)e-gerbil.net> http://www.e-gerbil.net/ras PGP Key ID: 0x138EA177 (67 29 D7 BC E8 18 3E DA B2 46 B3 D8 14 36 FE B6)

1 0

Re: Wierd Route
by John Osmon 22 Jan '01

22 Jan '01

Beware of things that still call themselves 'Qwest' when you see a traceroute touch the 14 state (former) US West territory. Qwest sold all of its "Long Distance Assets" in the region to Touch America, a wholly owned telecommunications subsidiary of The Montana Power Company. Folks in the 14 state region that used to buy transit from AS209 now get that transit from AS3908. If you sit and hunt, you can find many cases where traceroutes will show that AS1 and/or AS3561 lies between different addresses where in-addr's resolve to qwest.net. Probably not all that useful of an exercise though... John On Mon, Jan 22, 2001 at 03:05:20AM -0500, Joe Budion wrote: > > This one takes the cake for weirdest route I've ever seen: > > #3 sl-gw5-kc-5-0-0.sprintlink.net (144.232.132.45): TTL Exceeded, ttl=253, > 41 ms > #4 sl-bb20-kc-0-0.sprintlink.net (144.232.23.1): TTL Exceeded, ttl=252, 40 > ms > #5 sl-bb20-fw-13-0.sprintlink.net (144.232.9.254): TTL Exceeded, ttl=251, 50 > ms > #6 Unavailable (144.232.18.162): TTL Exceeded, ttl=250, 50 ms > #7 dal-core-02.inet.qwest.net (205.171.25.49): TTL Exceeded, ttl=249, 50 ms > #8 hou-core-01.inet.qwest.net (205.171.5.169): TTL Exceeded, ttl=248, 50 ms > #9 hou-edge-07.inet.qwest.net (205.171.23.14): TTL Exceeded, ttl=247, 50 ms > #10 a6-0-6.crtntx1-ba2.bbnplanet.net (4.24.147.21): TTL Exceeded, ttl=241, > 80 ms > #11 p10-0.crtntx1-br2.bbnplanet.net (4.24.10.189): TTL Exceeded, ttl=242, 81 > ms > #12 p2-0.crtntx1-cr8.bbnplanet.net (4.24.8.198): TTL Exceeded, ttl=241, 80 > ms > #13 p5-0.toucham.bbnplanet.net (4.24.117.66): TTL Exceeded, ttl=240, 100 ms > #14 den-edge-18.inet.qwest.net (205.171.16.34): TTL Exceeded, ttl=239, 101 > ms > #15 Unavailable (63.145.64.62): TTL Exceeded, ttl=238, 150 ms > #16 SLCNTWEB01 (63.237.115.200): Echo Reply, ttl=110, 150 ms > > Statistics: Out 14, in 14, loss 0%, times (min/avg/max) 40/76/150 ms > > joe > -- John Osmon "bash awk grep perl sed df du, du-du du-du, john(a)osmon.net vi troff su fsck rm * halt LART LART LART!" -- the Swedish BOFH

1 0

multiple origin ASes
by abha 22 Jan '01

22 Jan '01

Hi All! I'm trying to get a feeling for how often ISPs announce a prefix from different origin ASes intentionally.... (and why...) I asked this question on my peering survey (*grin*), but I thought i'd hit a wider audience... Thanks. -abha ;)

4 4

Re: Wierd Route
by Sean Donelan 22 Jan '01

22 Jan '01

I hate to tell you ... In many cases, the folks in the NOC are primarily there to watch red lights and green lights. Even if you had the secret NOC phone number, it wouldn't do you much good. In a large provider the NOC has soo many rules and procedures, the NOC-level techs aren't even allowed to log onto core routers to look. And forget about making any changes to a router configuration. Even if the NOC folks wanted to help, they aren't allowed to. Whether its a large provider or a small provider, they both seem to have about the same number of people with the expertise and authority to fix "interesting" problems. It might be called 4th level support, or backbone engineering, or senior engineers. It has less to do with technical clue than management clue. I've often found a person in a NOC who understood the problem, and even understood what was required to fix it, but due to decisions by his management he wasn't allowed to fix it. There are management culture differences between the various providers which affect how well their employees can fix things. On Mon, 22 January 2001, Adrian Chadd wrote: > <off-topic> > > .. and since various companies keep hiring bright university students > right off the bat, they don't get to spend time learning how the net > works by working in a NOC. > > Which means, finding clueful people to staff a NOC is getting more > difficult each day. > > Thanks guys. :-) > > </offtopic>

2 1

Re: Wierd Route
by Richard A. Steenbergen 22 Jan '01

22 Jan '01

On Mon, Jan 22, 2001 at 03:05:20AM -0500, Joe Budion wrote: > > This one takes the cake for weirdest route I've ever seen: 10 p9-0.crtntx1-br1.bbnplanet.net (4.24.10.213) 48 msec 48 msec 44 msec 11 p15-0.crtntx1-br2.bbnplanet.net (4.24.10.114) 48 msec 48 msec 48 msec 12 p2-0.crtntx1-cr8.bbnplanet.net (4.24.8.198) 48 msec 48 msec 48 msec 13 p5-0.toucham.bbnplanet.net (4.24.117.66) 68 msec 68 msec 68 msec 14 den-edge-18.inet.qwest.net (205.171.16.34) 68 msec 68 msec 68 msec 15 208.46.255.14 120 msec 120 msec 120 msec 16 63.237.115.200 120 msec 120 msec 120 msec ner-routes>sh ip bgp 205.171.16.34 BGP routing table entry for 205.171.16.32/30, version 0 Paths: (1 available, no best path) 3908 4.24.117.66 (inaccessible) from 4.0.4.20 (4.24.0.207) Origin IGP, metric 0, localpref 100, valid, internal Community: 1:1001 209:209 209:30357 3908:900 Originator : 4.24.0.207, Cluster list: 4.0.4.20, 4.24.0.58 ner-routes>sh ip bgp 63.237.115.20 BGP routing table entry for 63.237.115.0/27, version 0 Paths: (1 available, no best path) 3908 4.24.117.66 (inaccessible) from 4.0.4.20 (4.24.0.207) Origin IGP, metric 5, localpref 100, valid, internal Community: 1:1001 209:209 209:30357 3908:900 Originator : 4.24.0.207, Cluster list: 4.0.4.20, 4.24.0.58 Gotte filters? -- Richard A Steenbergen <ras(a)e-gerbil.net> http://www.e-gerbil.net/ras PGP Key ID: 0x138EA177 (67 29 D7 BC E8 18 3E DA B2 46 B3 D8 14 36 FE B6)

5 5

Re: Wierd Route
by Jon Stanley 22 Jan '01

22 Jan '01

not particularly when you take a look at it. Misbehaved origin AS: BGP routing table entry for 63.237.115.192/27, version 3368964 Paths: (3 available, best #3) Not advertised to any peer 5056 3561 3908 167.142.3.6 from 167.142.3.6 (167.142.3.6) Origin IGP, localpref 100, valid, external 1221 3561 3908 203.62.248.4 from 203.62.248.4 (203.62.248.4) Origin IGP, localpref 100, valid, external 1221 3561 3908 203.62.252.21 from 203.62.252.21 (203.62.252.21) Origin IGP, localpref 100, valid, external, best route-views.oregon-ix.net> A router will take the most specific path that it knows about to a particular destination. Since parts of the Internet are seeing a /27 here, it's winding up going all over the place. It would appear that AS3561 is allowing such small advertisements from AS3908, and then providers that dont have filtering at the border based on prefix length get injected this very specific route. Yet more examples of what happens with a lack of filtering........ On Mon, 22 Jan 2001, Joe Budion wrote: > > This one takes the cake for weirdest route I've ever seen: > > #3 sl-gw5-kc-5-0-0.sprintlink.net (144.232.132.45): TTL Exceeded, ttl=253, > 41 ms > #4 sl-bb20-kc-0-0.sprintlink.net (144.232.23.1): TTL Exceeded, ttl=252, 40 > ms > #5 sl-bb20-fw-13-0.sprintlink.net (144.232.9.254): TTL Exceeded, ttl=251, 50 > ms > #6 Unavailable (144.232.18.162): TTL Exceeded, ttl=250, 50 ms > #7 dal-core-02.inet.qwest.net (205.171.25.49): TTL Exceeded, ttl=249, 50 ms > #8 hou-core-01.inet.qwest.net (205.171.5.169): TTL Exceeded, ttl=248, 50 ms > #9 hou-edge-07.inet.qwest.net (205.171.23.14): TTL Exceeded, ttl=247, 50 ms > #10 a6-0-6.crtntx1-ba2.bbnplanet.net (4.24.147.21): TTL Exceeded, ttl=241, > 80 ms > #11 p10-0.crtntx1-br2.bbnplanet.net (4.24.10.189): TTL Exceeded, ttl=242, 81 > ms > #12 p2-0.crtntx1-cr8.bbnplanet.net (4.24.8.198): TTL Exceeded, ttl=241, 80 > ms > #13 p5-0.toucham.bbnplanet.net (4.24.117.66): TTL Exceeded, ttl=240, 100 ms > #14 den-edge-18.inet.qwest.net (205.171.16.34): TTL Exceeded, ttl=239, 101 > ms > #15 Unavailable (63.145.64.62): TTL Exceeded, ttl=238, 150 ms > #16 SLCNTWEB01 (63.237.115.200): Echo Reply, ttl=110, 150 ms > > Statistics: Out 14, in 14, loss 0%, times (min/avg/max) 40/76/150 ms > > joe > >

1 0