***: mercutio has joined #arpnetworks
ChanServ sets mode: +o mercutio
mkb has joined #arpnetworks mkb: anyone else have ipv6 problems? mhoran: Yeah all sorts of problems since Friday. I think they're still getting things settled into the new cage. mkb: yeah I figured it was that mhoran: Yeah, my IPv6 seems to be down completely.
Looks like since 5:15am UTC.
I sent a support ticket. IPv4 also seems to be quite laggy and my alerting keeps going off. ***: qbit has quit IRC (Quit: WeeChat 2.4)
vom has joined #arpnetworks vom: ever since yesterday my VPS has had broken ipv6 - i can't ping my def. gw from it and inbound traces die one hop prior
anyone else experiencing this ? should i open a ticket ?
hmm - interesting - i had the unicast gw in my config for years (::1) - and it was working - but i just did a tcpdump and i see RA's with a LL
so i removed the static gateway and rebooted - and now I have my def. gw via RA and all is well
dunno if this is a known issue / related to the maintenance or what ***: tmarble has quit IRC (Remote host closed the connection) mhoran: Ah interesting. I opened a ticket because I encountered the same. Something must have changed ***: anisfarhana has joined #arpnetworks anisfarhana: Hello guys :D
Are we having an issue with ipv6 connection right now? mike-burns: anisfarhana: http://irclogger.arpnetworks.com/irclogger_log/arpnetworks?date=2019-04-21,Sun&sel=5#l1 - seems like it. anisfarhana: What a relief.
At least i know the problem is from arpnetworks. mike-burns: Haha yeah -- it's not (necessarily) that you broke a config! anisfarhana: I already sent support ticket too just now.
mike-burns: This is first time i had problem with ipv6!
But i am expecting there will be a problem with ipv6 due to maintenance. mhoran: Maybe up_the_irons did this purposefully to get everyone back on IRC.
Yeah oddly v6 was working fine after the migration but broke around 05:11 UTC. anisfarhana: mhoran: I'm always here! :P
Pity up_the_irons. mhoran: Ha. I can't really tell because I hide joins and parts anisfarhana: He must be busy right now inside the new cage. mike-burns: Despite all our rage he is still just an admin in a cage. anisfarhana: I can handover my force to him :P
Use the force, Gary! :D
So we don't have any ETA right now for this issue right? ***: ziyourenxiang__ has quit IRC (Ping timeout: 246 seconds) anisfarhana: my ipv6 connections still okey on Friday though.
I got replied from support team already :D
Thanks arpnetwoks!
Okey i got my ipv6 back :) mhoran: Same!
anisfarhana: do you have native ipv6 at home? Or using a tunnel? anisfarhana: I disabled ipv6 at home.
v4 to v6 :D ***: brycec has quit IRC (Ping timeout: 250 seconds)
brycec has joined #arpnetworks brycec: Alright, now _I'M_ having IPv6 issues >:(
mtr from my VPS to google.com http://pastebitch.com/PVDS mercutio: brycec: looking into it brycec: thx mercutio: can you ping fe80::1 ? brycec: Yes mercutio
Some really erratic trip times: 0, 5, 20, 0, 50 etc (ms) mercutio: it's cisco brycec: k, didn't know if it might be indicative of network load [too] anisfarhana: I thought ipv6 is resolved now. brycec: Oh and for reference, from my home (tunneled) to my Thunder http://pastebitch.com/EHZb mercutio: oh hmm
i have a diffrent subnet for you brycec: hm? mercutio: a650
that hop say a6c0 brycec: Ah, yeah I'm a650
No idea about that hop *shrug* mercutio: that's really odd brycec: ::/0 fe80::1 UG 1024 3463630290 eth0
My default route remains fe80::1 so... *shrugging harder* ***: acf_ has quit IRC (Remote host closed the connection) mercutio: yeh it might not know what to pick brycec: Maybe my VLAN is misconfigured with the switch/router? (I'm more or less guessing, so I'll just shutup and let you do your thing) mercutio: i moved you to the cisco earlier, but i tested and it was fine
and now it's not!
so not sure what's up with that brycec: Computers were such a mistake :P mercutio: i traced to ::3 fine before
i can mtr to it now
better brycec ? brycec: Something changed... 1 sec
It does seem to be working, yes. Though a6c0 still shows in traceroutes mercutio: hmm i wonder if it's possible to change that
because that's confusing :) brycec: Personally, I don't think I care because I never see it, so long as things work :P mercutio: heh
yeh it doesn't happen with non routed subnets brycec: Me and my /48 are "special" eh mercutio: there's quite a few special ones brycec: Well FWIW I appreciate having a /48, I'm using 0.0061% of it! mercutio: heh
i don't have ipv6 at home still brycec: (by subnet, not address allocation) mercutio: i tested from arp, and your subnet was accessible from arp still anisfarhana: mercutio: oh really? how come? mercutio: isp doesn't support it anisfarhana anisfarhana: Interesting! mercutio: ipv6 isn't very common here anisfarhana: Change your ISP
:D brycec: (Or get a tunnel, yadda yadda. Not really the point, or important, at this juncture though.) anisfarhana: I have options here for my ADSL, ipv4 only or with ipv6 mercutio: i don't actually have anywhere local to get tunnel from brycec: Thank you so much for the speedy assist, mercutio ! And best of luck with the ongoing migrations. mercutio: otherwise i probably would
brycec: yeh sorry about that anisfarhana: up_the_irons just fixed my ipv6 too :) -: anisfarhana is happy customer! brycec: Sydney isn't "local" to you? :O /s mercutio: so yeh you both are on cisco termination for ipv6 rather than openbsd now
brycec: not really, it is further from the US than here
so it will increase latency for all US traffic by about 35 msec or so
and 135 vs 170 msec is actually quite noticably different brycec: I imagine you could go through Hong Kong and have a decent trans-pacific latency. mercutio: my friend visited china when he was living in england. latency was terrible between the two
you'd think that china could have decent latency to england, but all the paths go via the US brycec: No, I wouldn't think that at all.
There's a whole goddamn continent between China and Europe, EITHER WAY YOU GO :P mercutio: true brycec: And I suspect the connections crossing the US are better than whatever crosses Russia and friends. mercutio: europe is pretty good for connectivty, same with russia
places like bulgaria are really cheap for high speed connections brycec: The US has good economic factors for making good connections across the country. I don't imagine Russia has the same factors. But this is just conjecture on my part. mercutio: the US has terrible pricing for internet
for home users
there are a few monopolies which hardly compete
like comcast is like $80 USD for 100 megabit or something isn't it?
on a dated cable network brycec: You don't have to tell me :P anisfarhana: Come here mercutio mercutio: yeh so places like bulgaria are like $10 USD/month for internet brycec: TIL there's undersea cable directly between New Zealand and the US, much to my surprise. anisfarhana: You can enjoy my ipv6 here
:D mercutio: running fibre is mostly about labour... brycec: (well, direct to Hawaii, then from Hawaii to mainland US making landfall in Silicon Valley) mercutio: yeh there's more than one cable
they introduced a new higher latency one
but it still keeps the pricing better
it's 125ish msec from auckland to los angeles
then latency to auckland, and to your pop brycec: Yep I just read about that one (the third such, went live last June) mercutio: our latency to singapore is similar
but bandwidth to singapore tends to be much worse brycec: Still disppointing that 15,000km only gets you 125ms mercutio: yeh it's hard to do more direct paths
110 msec or something would be noticably faster
it's still reasonably direct
gpon has a bit more latency than i'd like too brycec: Quite surprising how many submarine cables come ashore just 100km from where I live mercutio: it's like 0.6 to 1.4 msec or something
like variable
i think our cable lands between san jose and los angeles
so it ends up being transported to one or the other brycec: Hawaiki (the newest one) is the one that lands near me
https://www.youtube.com/watch?v=5F_aWegm_ok video of the boat arriving, apparently BryceBot: YouTube video: "Pacific City Cable Landing" by Hawaiki Cable brycec: Oh including the cable actually coming out of some sort of ground conduit, this is neat mercutio: oh their latency to Oregon is pretty good, 126msec it says
heh they offer transport to one wilshire
it's so much easier to terminate at a popular location
it makes it much easier to sell wavelengths to smaller companies brycec: I've studied the video, matched it up to aerial photography, and subsequently realized I've driven past its landing station several times without ever realizing it was there. (Tucked back from the road, nondescript, etc)
So that was a good use of my Sunday afternoon ***: ksthomas has joined #arpnetworks brycec: "The fiber optic cable may be buried at less than one meter, but not less than three feet, in hard ground such as portions of the "reef area" along the cable route." Lol, that's a window of about 3.5in/9cm mercutio: haha
remember google abandoned a city becuase they were laying fibre too shallow? brycec: I hadn't heard that mercutio: https://www.bizjournals.com/atlanta/news/2019/02/11/google-fiber-to-make-its-first-market-exit-in.html brycec: It figures there would be someone else who'd already done what I was doing The fiber optic cable may be buried at less than one meter, but not less than three feet, in hard ground such as portions of the "reef area" along the cable route.
derp, wrong clipboard
https://cryptome.org/eyeball/cablew/cablew-eyeball.htm mercutio: they lay my fibre in my neighbours garden
it seems a stupid place to put cables to my mind brycec: Better than in YOUR garden :P mercutio: well it's more likely to have a spade through it
new zealand is doing pretty well with fibre though brycec: Fiber bear me is all in underground service ducts and the like, fortunately. But there are areas with it strung pole-to-pole too, both out of the way at least.
*Fibre near mercutio: like there is affordable 1000/500 megabit fibre connections now
in most of the country, from multiple providers
there will only be one provider to get the actual bit to your house
but head ends are reasonably affordable too, so it makes it more accessible for ilke say a backup provider to start up and have their own fibre links brycec: Good for y'all! mercutio: that said most people will only have one provider
yeh it sounded kind of iffy at first, with 100/50 and 30/10 plans standardised by the government brycec: Agh I'm reading through this marine website's report of the laying of a cable and they keep using "nm" for nautical miles and it's really infuriating me. ("NM" is the proper abbreviation for nautical miles, AND YOU'RE A DAMN MARINE WEBSITE WHY DON'T YOU KNOW THIS) mercutio: but the fibre providers came out with faster plans and now 100/20 is pretty much teh default for a decentish price brycec: Nothing wrong with 100/50 (or even 30/10) those are very serviceable connections for Most People (TM) mercutio: it's like $45 for 100/20 and $55 for 1000/500 or something for residential for wholesale cost though brycec: Meanwhile, I've got 100/100 and life is fanatstic and everythign is backed-up In The Cloud! BryceBot: TO THE CLOUD!!! mercutio: NZD which is about 0.7x USD brycec: @exch 55 NZD USD BryceBot: 55 NZD -> 36.763206847849 USD (as of Sun, 21 Apr 2019 15:00:00 -0700) mercutio: well 100/20 is a lot better than 30/10 as a base offering
so yeah it's affordable enough that providers can put a little markup on it
and buy transit etc
base plans are moe like $80 or $90 i think
@exch 80 nzd usd BryceBot: 80 NZD -> 53.473755415053 USD (as of Sun, 21 Apr 2019 15:00:00 -0700) brycec: "These patrols continued until the cable was completely buried in the fishing grounds. Little fishing activity occurred in the area during the cable installation, and no conflicts occurred. One humpback whale was observed in the vicinity during post-lay burial." I appreciate the "one whale was seen" note in their report lol mercutio: there's a cable from perth to singapore ish
and it gets hit all the time
https://en.wikipedia.org/wiki/SEA-ME-WE_3 BryceBot: SEA-ME-WE 3 :: SEA-ME-WE3 or South-East Asia - Middle East - Western Europe 3 is an optical submarine telecommunications cable linking those regions and is the longest in the world. Completed in late 2000, it is led by France Telecom and China Telecom, and is administered by Singtel, a telecommunications operator owned by the Government of Singapore. The Consortium is formed by 92 other investors from the telecom industry. It was commissioned... mercutio: i think it's around singapore somewhere it gets hit, but it goes all over the place
hah it has two fibre pairs with 2.3 terabit/sec per pair
you know that cable should give better china to england latency if it was used :)
i suppose it's still pretty indirect brycec: Reading reports from circa-2000 and they're describing cables with 560Mbps/pair and like wow we've come a long, long way in a short time. mercutio: yeah mkb: you ever read that neil stephenson article for wired? brycec: Also, 283kb file for a nav chart "Note: These images are large, and will take several minutes to load." mercutio: hahaha mkb: http://archive.wired.com/wired/archive/4.12/ffglass_pr.html mercutio: 2000 is when things were rapidly taking off though brycec: I think I sneeze slower than it took to load. mercutio: in 2011 i had like 16 megabit adsl
now i have 1000/500 fibre
16/1 adsl
in 2002 i had like 256kbit/128kbit cable
in 2001 i still had dialup at home :) brycec: Ditto, that's right around the time I went from dial-up (MPPP, two phone lines baby!) to cable (10/1 I think) mercutio: you jumped straight to 10 megabit? brycec: May not have even been that fast, it was a long time ago! mercutio: i had faster net at work in 2001 at least
i never could web browse on dialup comfortably coming from bbs's
even ssh was laggy on dialup brycec: I remember it was a white Motorola SurfBoard modem... and I remember having all kinds of fun fiddling with it and the HFC network (before ISPs actually started to care about "security") mercutio: 64kbit isdn was a lot lower latency brycec: Yeah, I remember relying greatly upon a caching proxy mercutio: heh
i've actually found browsing has got slower over time
but i think it's the browsers...
like i had an old version of chromium somewhere stored and i loaded it and it was stupidly fast brycec: mkb: Interesting (albeit LONG! Over 40k words) read, thanks
(Typically, that works out to 80+ pages in book form!) mercutio: wow sri lanka had a huge bomb
killed more than 200 people brycec: Oh you mean actual explosion, Shit.
(Thought "a huge bomb" like a failed cable, tech boom crash, etc) ***: ziyourenxiang__ has joined #arpnetworks
ksthomas has quit IRC (Remote host closed the connection)
ziyourenxiang__ has quit IRC (Ping timeout: 245 seconds)
acf_ has joined #arpnetworks acf_: hmm so my arp metal box just rebooted
after roughly 2 years of uptime
> Physical move of ARP Metalâ„¢ servers to new cage
I'm guessing that was this? mhoran: Yup. acf_: is everything still in wilshire annex, just a different cage? mhoran: Not sure, but I think so. mercutio: different cage, different floor ***: vom has quit IRC (Quit: leaving) mnathani: Does traffic trombone the SMF link between the cages? Ie: external request comes in and terminates on the old floor /cage then travels via smf to new cage to the VPS / Thunder End point and back over the SMF to the old cage and back out the transit link the request initially came in on?
or does each cage have its own Transit / Peering connections as well ***: qbit has joined #arpnetworks mercutio: some traffic traverses
we're advertising prefixes on both sides atm
and some people are still terminated in the old cage mnathani: Is the MTU on the 1gig SMF run 1500 or more like 9/10,000 Jumbo frame? I assume its an 802.1Q Trunk link carrying multiple VLANs? mercutio: nah it doesn't tunnel vlans, it just has tagged vlan pass trhough so 1500 per valn
so 1500 through to vlan i should say
it's when you want vlan inside vlan inside vlan etc you need larger mtu, and even then people usually run 1600 or such rather than 9k
unless they want larger mtu for traffic.. but high mtu isn't very popular still
basically large mtu is more useful on higher bandwidth links but then the mtu overhead isn't that big a concern, and there's lots of hardware offload now days.. acf_: I'm guessing the plan is to eventually move everything to your "new" cage mercutio: yeah mnathani: thanks mercutio! acf_: who are the transit providers we have now?
last I remember it was NTT and level3, but it seems things have changed
right now everything seems to be routed through "atlanticmetro"