up_the_irons: no packet loss so far hmm, I see v6 traffic but not v4 traffic (v4 gw at my vps is 'ping: sento: Host is down' ugh weird i had some weird thing with my /48 that made me think i screwed up configuration, but turned out it was something else unrelated to me fatfingering anything, which was nice odd I have one vps that works and another which doesn't huh bad vlan config or something? aka I get icmp replies from the router from a different host on the same vlan yeah, sounds like it could be can ping from hosta -> hostb and hosta -> gw, but hostb does not reach the gw manually setting arp yields little success either, hrmph guess I'll try powering off the vps and powering it back on incase it re-plumbs something v6 works so I think its external to the vps itself though woo hmm, still no IPv4 access from my kvr22 vps .. anyone else on kvr22 ? yep % ping google.com PING google.com (173.194.33.2): 56 data bytes 64 bytes from 173.194.33.2: icmp_seq=0 ttl=53 time=27.247 ms when did you lose ipv4? toddf: so is this a new host on an existing account/vlan? Did you get another /30? Otherwise, there's only one usable IP in the /30 you normally get.... nothing changed. it is my 2nd oldest host in a group of 4 in this one account/vlan brycec: just woke up this morning to no response on v4 but v6 works stranger yet I can hit v4 on the other vps'en just not the gateway from this one host toddf ? haha, nice to see you over here too! :) Sounds like a good time to email up_the_irons I don't even get arp is-at from the gateway from my arp who-has request on the afflicted vps robonerd: you culled me from here to #coindev, remember? brycec: done so a while back fun reality is this is my mail server, so unless responses from tinder come via IPv6 he can't reach me back, emailed a followup from gmail as well toddf yea just forgot because you've become such a regular :D huh. so .. anybody else on kvr22 ? yes able to ping 8.8.8.8 ? able to ping whatever can you ping 208.79.89.90 ? % ping 8.8.8.8 PING 8.8.8.8 (8.8.8.8): 56 data bytes 64 bytes from 8.8.8.8: icmp_seq=0 ttl=50 time=26.977 ms no how about .91 ? no reply from that yes 64 bytes from 208.79.89.91: icmp_seq=0 ttl=253 time=0.486 ms .90 is on kvr22 .. v4 issues .. .91 is on kvr14 .. v4 works fine same vlan so its not a cabling issue to the gateway, you're able to do v4 fine from kvr22 and its not blocking on my host (I can tcpdump and see traffic flowing out, but not in) and I can ping between my vps on v4 just the gateway doesn't respond to my .90 ip time to try an IPv4 alias I guess incase somehow my .90 ip got blacklisted for $automated_bogus_reason or somesuch not that I'm aware of any $automated_reasons ever but I'm grasping at straws at this point grmph, even an inet alias doesn't speak v4 for this system, what gives I wonder? I was wanting to know if there is a way to find out why my computer is acting like it is on dailup when I am on cable and getting my full speed Yes. Next question? brycec: is there a santa claus/ I was wanting to know if there is a way to find out why my computer is acting like it is on dailup when I am on cable and getting my full speed LLKCKfan the past few years i've seen the US internet really slow down massive latency added to almost every back haul/transit on a lot of nodes i have a feeling it's data snooping + channel splitting .it is not my speed m0unds: As long as you believe, yes. LLKCKfan: Just curious, what makes you think #arpnetworks is a good place to ask that question (over, and over, and over)? hmm LLKCKfan: how is babby formed? up_the_irons: did you see my e-mail? tmarble: if you sent email to support@arpnetworks.com then you'll get an automated response that says it's been received; when he responds is not guaranteed .. I've been down all day waiting on a v4 issue that I've never seen before toddf: ok... it's kind of a personalized followup... not really apropos to escalate to support@ (yet :) toddf: can you tunnel your traffic from one box to the other? I've done so for mail voip is not quite so easy dns .. no tunnel required, it is resiliant when you say you're on the same vlan on both boxes i assume that means you can use a private ip address and talk between the two i dunno if you've doen that or not you should know but i'd be curious if that was working at arpnetworks there is a single vlan for all boxes under one account yeah I can do svlan(4) between the systems (providing arpnetworks removes a restriction that blocks it by default) and I do end up doing that also oh ok i fuond that vlan didn't work vlan on vlan yes you can't do vlan inside vlan when he does vlan himself but you can do svlan(4) because its intended to be nested inside vlan(4) ahh right well the only tiem i had shared vlan was when i had the beta vm but anyway, the boxes can reach each other? yes I can reach via v4 v6 has always worked but the one box cannot get even an arp entry to the v4 gateway just nothing comes in so can you proxy arp on the working box and forward traffic to the non working box? with no tunnel that sounds complicated i wonder where up_the_irons is hadn't thought of that, that would do inbound, but outbound .. route through my other vps I guess? meh busy I'm sure hmm does outbound static arp work? tried it, no go hmm I also confirmed another arpnetworks customer on kvr22 (where my afflicted vps resides) can reach the internet ew're thjinking up complicated ways aren't we :/ so I think it is something specific with my vlan between kvr22 and the v4 router i wonder if it's a mac address clash Sounds like toddf's vlan has become disconnected from the v4 router. i dunno what the automated mac creation is like (Oh wait, you said other v4 hosts are fine. hm) brycec: except the vlan works from other kvr systems .. kvr27,kvr28,kvr14 so the map is does the IP address with on your othewr host? Yeah, I ack'd that. kvr27+kvr28+kvr14+kvr22 -> v6 ok kvr27+kvr28+kvr14 -> v4 ok kvr22 -> v4 !ok if ip the other ip address works on the other host then i would anycast it :/ ie, put the same ip on both hosts then enable forwarding on the working box how do you anycast w/out a mac conflict sure I did this convoluted setup at first and set your default host i was having issues on my 22 box w/ipv6 packet loss last night to avoid getting a /29 then bounce the traffic from the working host oh hmm but VoIP love was not there, so I dropped it when I needed yet another separate v4 address it doesn't need to have an actual ip address on the working host it just needs to arp for it and route to the right place mac conflicts don't matter I have everything working but VoIP .. guess I could try to nat VoIP out another ip or somesuch, bleh well they don't most of the time as long as either host works but yeha you want proxy arp I tried putting another IP on the afflicted vps, and it didn't work any better rather than actual anycast with bgp you could just advertise the ip as a /32 guess I should check e.g. kvr14 to see if arp who-has is being broadcast for the afflicted ip from the router yeh i think it's mac address related I tried changing my mac! does that work? and it came back with v6 just fine, v4 no change gah it is as if the plubmbing between kvr22 and the v4 router for my vlan only is busted did you get your port shifted to gigabit? yes i assuem that's where the problem is then and the working one didn't? all of my vps'en are gigabit so far as I know all 4 damnit ;-) definitely makes no sense, given that working paths overlap with the nonworking path. :( I'm almost leaning towards something blowing some automated trigger in the switch for my vps traffic or something, though we've been repeatedly told no automatic triggers except port connection throttling on 22 so I'm guessing some wierd switch failure or somesuch well I'm going to presume there are multiple switches involved There's a UDP rate limit as well. But that's not a block. that's why i was wondering if it effected gigabit vps but not the oold one and that likely kvr14 is plugged into a different switch than kvr22 vs kvr28/kvr27 err port i'm on kvr15 i think and i haven't experienced issues so it could be a misfire on the vlan config between the kvr22 switch and the v4 router, but not the v6 router it coul dbe feel free to compare 0.v.freedaemon.com vs 2.v.freedaemon.com ;-) so if you do proxy arp and forward to the internal ip address can you see receive traffic ah, just realized another cluebit kvr27 cannot reach kvr22 via v4 nor v6 hmm but can via ipv6? but kvr27 can reach google via v4 and v6 well i'd g down the arp -s ... pub route and th en put it back when it's working right 16:54:24.985632 00:1f:6c:92:7f:00 ff:ff:ff:ff:ff:ff 0806 60: arp who-has 208.79.89.90 tell 208.79.89.89 I see that at kvr14 but I see this at kvr22 16:54:44.563261 52:54:00:ef:41:22 ff:ff:ff:ff:ff:ff 0806 42: arp who-has 208.79.89.89 tell 208.79.89.90 I see the 2nd packet at kvr14 also but the 1st packet I don't see at kvr22 so arp magic is not fixing my ethernet segment breakage between kvr22 and v4 router i just realised.. when iw as saying about doing arp stuff.. i was thinking you wouldn't have stateful filtering but you migt and that can break too yes I'm going to try it on a vps that doesn't have any fancy filtering ahh with openbsd you have to append no keep state toddf: can you give it another try up_the_irons: hi! workie! toddf: this is my fault; i made a change on kvr22 yesterday, and something broke :( I apologize voip phone registered and all manner of things are working now (it only affected people on the GigE router, which is very few) w00t nice kvr27 can now talk to kvr22 via v4 as well on my vlan The prodigal up_the_irons returns! lol it was fun given another user was on kvr22 and his v4 worked fine, but didn't think to ask about gige bits now my backup bogus backup mx works yeah sorry for the wasted diagnostic time I'm understanding of the scenario given I've been in your shoes more often than not (It's not like there was anything else to do while it was down, right? :P) brycec: exactly, I have all 3 dns servers working now as well as a result lol good :) toddf: I assume geographically diverse too? oh up_the_irons is around :) haha they send you invoices for $0 No... brycec: two at arpnetworks (one was misconfigured aka firewalled to /dev/null) and one at home (suceptable to power outages) oh, you actually do pay them? i figured you meant the free authoritative stuff i use route53 for critical stuff m0unds: I use their free stuff. There's no invoice involved. It's my authoritative too. oh, i was joking about the $0 invoice (cuz it's free) I highly recommend it, even if just for backup. Can't beat 5 separate anycast servers. I do IT consulting and service an isp in my client portfolio. if I can't get dns and mail and such right, why bother? sure I could outsource to dns.he.net but where's the ability to say "i do this for myself, I can do this for you too" ? ;-) i hate web interfaces for dns myself I see your point. Even so, I would suggest using them for backu/slave. Part of doing IT is knowing good vendors after all he.net is good now? :) wwere they evil before? they're good at supporting ipv6. they had heaps of power outages at one of their datacentres fremont? Ah. I've had issues with their tunneling, absolutely. and they're generally known to be subpar for ipv4 connectivity But never a DNS issue. but cheap in ipv6 they're pretty good though i dunno, i don't even know who's good :/ i mean some people say level3 are good, but level3 still have issues it's well known that cogent are bad bad bad agreed it's generally consdiered that at&t aren't that great for a wihle a lot of people seemed to be pushing web sites to cloudflare and omg, so many performance issues suddenly with sites that shifted funny, btc-e.com is on cloudfare i'd still probably consider some of the better cdn providers to have higher qaulity dns though is it randomly slow? they do have occasional hiccups, and indeed it is randomly slow I just programmed patience into my bot and don't generally notice as a result like 5 seconds to load a page slow? and was it reasonble consistent prior? i mean some sites that moved to cdn's had backend issues prior like for instance reddit so it can be hard to know if it's the cdn or the backend like I know about these errors enough to put them into the public api: https://github.com/toddfries/Finance-btce/blob/master/lib/Finance/btce.pm#L375 but generally speaking, if a site performs consistently, then goes on cdn and goes randomly slow, i blame the cdn I've never known btc-e.com before cloudfare so I can't comment hahahahaha as an end user i've found cachefly consistently good i think it's kind of expensive though yeah from $99/month with monitoring i do on page loads for clients on paid cloudflare, the page loads from a bunch of locations in the us are within 5% avg maybe it's an international thing, idk but +/- 5% speed in ms for page loading from cloudflare e.g. for eyeballs, not noticible it may be an international thing but i was getting slow performance when it hit the hangon it was slow performance when i hit the CA node.. but i think it wasn't so bad with the TX node from a vps but i suppose CA may get more ddos attacks from china :/ yea LOL the [NSA] has arrived oh the new dc thing is happening next year isn't it? [NSA]: meet [FBI] that is one of my goals, yes cool I was wanting to know if there is a way to find out why my computer is acting like it is on dailup when I am on cable and getting my full speed LLKCKfan: No, there isn't. Therehas to be No way. You do not want to help LLKCKfan: you keep asking the same damn question, over and over, without providing information. Why do you even ask in #arpnetworks at all?? And then silence... We ask questions, you don't answer. Why do you bother? 1i want to say that the answer might be because you're using a rural 'cable company' that maxes out at 512kbps and has 10gb caps after which you're throttled to 64kbps and they apparently have a single t1 total lolol I was wanting to know if there is a way to find out why my computer is acting like it is on dailup when I am on cable and getting my full speed LLKCKfan: No. I GET MY FULL SPEED THAT IS NOT WHAT I AM ASKING ABOUT LLKCKfan: No.