Wheee look at those NTT hops
...there are more hops to Google over v6 than over v4 :(
12 vs 7
brycec: I'll see if ASN 11799 can peer with us over IPv6; that's how we're getting 7 hops in v4
I really need to get on this v6 bandwagon.
i just sent 'em an email about it
Cool
the route to google looks fine
ignore the hops
it's ~ 10 msec
well to www.google.com
oh and < 1 msec by ipv4
i see why you care :)
heh I didn't even pay attention to the latency... I was just curious about the new path taken was all
Also yeah I see those same numbers, 11ms over v6, .7ms over v4
kinda trivial latency-wise
but less hops is moar bettar i guess
fewer
Also *more and *better ...
nah
intentional
Hah.
is anybody having networking issues at arp at the moment? I'm discovering my vps'en are hard down, no networking between any of them let alone to the gateways
MOAR BETTAR
?
toddf: reference to something earlier.
toddf: And I'm not seeing any issues with my lone VPS.
whether it is new or I'm blind and just noticed, I'm very glad that portal.arpnetworks.com links to support.arpnetworks.com with the login to portal and I can view and create tickets that way. mail is down for me as a result of the vps issues mentioned above .. ;-(
Heh, so it does. Neat
that sucks - i had a blip early this morning that dropped ipv4 connectivity to a couple places and triggered monitoring notifications
it seems routing via any2ix/coresite is half-working now
i don't rememer as11799.net appearing before
seeing a lot more NTT in v6 routes to ARP now :)
oh ntt has better routing too i think
oh ntt are japanese aren't they
yep
so japanese destinaions aren't good tests :)
struggling to find ipv6 sites now
facebook, kame, ..
asininetech.com
google
dundeadly.org
vps[13].cobryce.com
i did google yesterday
asinetech.com is via he.net
(spelt correctly)
(as is vps1. vps3 is on ARP)
bryce: that won't help
i was seeing what goes over ntt :/
oh what
you are on tunnel
huh?
it goes via he.net to you too bryce
mercutio: to which, vps1? I already said it's on he.net.
(vps3 is native on ARP.)
oh right
vps1
<-- creative VPS naming
i thought you said vps 1 to 3 are on arp
yeh ok
yeh i know right
you should see my old naming
i had sj.meh.net.nz, la.meh.net.nz, chi.meh.net.z, da.meh.net.nz, uk.meh.net.nz, am.meh.net.nz
etc
but then what happens if you have more than one in la :/
so i have arp.meh.net.nz now :)
the old la.meh.net.nz was on budgetvm, who had heaps of ddos issues
lol I thought you were using meh.net.nz as a dummy
now i should really cull it from dns
haha
(but then I looked it up)
i have the.net.net too
oops
the.net.nz
but i'm not using it for much atm
argh that reminds me i have to update my nameservers :/
is anyone following the debian upstart vs systemd debate?
is there a ML I should look on to see this?
Either way I'll be annoyed; I just want BSD's init.
I value my sanity... That's why I listened to the Bill Nye/Ken Ham "creationism" "debate".
there's many mailing lists i think
but you shouldn't
it's been going on for months
it's like watching politics on tv :/
srsly
you may be able to do it, but it's much easier if someone else watches it for you and gives you an overview
unless you like watching grown men bickering between each other
Well, what's the overview?
well they've had months of debate, and have finally voted on... further discussion
http://www.phoronix.com/scan.php?page=news_item&px=MTU5NDk
i pasted some url to a humurous video outlining various things in it though
http://aceattorney.sparklin.org/jeu.php?id_proces=57684
which at least helped me understand in more deatil
i hope they go with systemd myself
same
i don't really like systemd
but i'd much rather see something that everyone uses
It looks like the upstart license is incompatible?
mike-burns: but there are upstart people on the board
the upstart license is .. complicated.
https://wiki.debian.org/Debate/initsystem/upstart - the con is confusing.
I agree with3 mercutio
but it's like the fsf license i think
where you have to assign copyright or something
Does anyone use Debian/BSD?
GNU userland and FreeBSD kernel, I guess?
nope
lol nope!
That's what I thought.
i grew to like systemd when i messed with arch
i have tried ferebsd 10 now
m0unds: how do you do /etc/rc.local?
eh?
Never really understood that... If you want the FreeBSD kernel, just run FreeBSD. If you're soooo attached to Debian-isms, you have personal issues :p
to have some local stuff executed after all the systemd crap :/
oh, no idea. i used just 100% systemd crap
i lvoe how much faster pacman in arch is than apt-get in debian/ubuntu
mercutio: Write a systemd service :p
yea, it's super fast
I never understood GNU userland+FreeBSD kernel: the Linux kernel has tons more hardware support, and GNU userland is awful.
systemd is really fast to init too
i'm using arch now :)
i don't really like the system journal vs discrete logs
mike-burns: except when it comes to very specific things, like ZFS :p
but it doesn't bother me /that/ much
i'm using ssd
but yeah it boots pretty quick
brycec: Right, true; and then I'd just use FreeBSD.
I hear that, m0unds
mike-burns: which is precisely what I'm doing too :p
I would, however, use Linux + BSD userland.
for some reason freebsd is insanely slow to boot
-ENOTENOUGHRICERS
i have it on an awful little amd netbook (arch) and it boots in ~15 sec w/a 5400rpm disk
but really ther'es not a huge diff between ubuntu and arch boot times
but freebsd seems to take about 10 times as long to boot
i imagine it's not doing enough parallel starting etc
how did you time it m0unds?
not sure how long it takes to start freebsd; i don't really reboot the systems i have that run it
I think that during the migration, Arch had a systemd service that just executed /etc/rc.local too. Easy enough to do. 14:36:50 < mercutio> m0unds: how do you do /etc/rc.local?
ah, ok - i tested arch after it was already in use
i'm not a linux desktop person
And you could always just use cron + @reboot
i timed it after post completed - post is 3 seconds
ok
i'm about 1 second to grub
something like that anyway
about 5 seconds booting
i just booted my other pc
it's uefi
my win8 box is to the login screen in 5-8 seconds
ssd, uefi
yeah this is ssd/uefi
dual boot windows/linux
i also don't boot into X :/
yeah, i have arch on a second disk in that box, but arch is on a platter drive
i just login and type startx
yeah, same w/the little laptop i have
i hate lightdm
desktop arch install boots to x because i don't care
rarely ever boot it anyway
i reckoen if soemthingshould be optimised it's video card
uhh getting the resolution thingy
[drm:drm_edid_block_valid] *ERROR* EDID checksum is invalid, remainder is 128
like crap like that in dmesg
yuck
ikr
but it seems to get the data ok
it just seems to slow it down
huh.
 radeon 0000:01:00.0: DVI-I-1: Ignoring invalid EDID block 1.
my mbp takes ~20 seconds to boot from power button w/an SSD
i think it's cos i only using one monitor
m0unds: wow
my hackinstosh boots faster than that :/
are you using mavericks?
Ooh burn
yep, mavericks
has lots of issues
weird
windows 8 definitely boots faster than macos though
my wife's mba boots in about the same time, but she's still on mountain lion
2013 gen (haswell) mba
oh
i didn't know they did haswell on lion
i don't hate macos
i just prefer linux and ion
i like it, there's just lots of improvements they could make to ux
i like that it installs zsh by default
i had to teach myself to use expose instead of cmd tab
and ssh
because i kept minimizing windows and it's annoying to go to the window menu to find what i'm trying to switch back to
yea, that was why i went with a macbook - better specs than an ultrabook, lots of tools i use on a regular basis baked in
2.9GHz i7, 8gb ram, 200-someodd gb ssd
SSDs sure are great
yea, price/gig is coming down a lot too
i got one for my desktop
you think that now
wait until they get 1/10th of the latency
256GB Samsung 840 EVO
there's some room for improvement
i think the first improvements are coming with parallel requests and more direct access
the only thing that sucks about ssds is having to use machines that don't have one
bringing latency down
m0unds: haha yeh
best part is no noise tbh
taht's why i like the idea of these hybrids
but would never want to use one myself
mechanical drives get noisy
yeah i dunno why some drives are much noisier than others
but i kept wanting to have raid at home
but having nowhere to stick it
yea, my env at work isn't quiet enough to notice disks
it's pretty easy to noitce 3 or 4 disks
yep
i used to use 4 disk raid at home, it's kind of annoying/intrusive
but it was pretty fast
so now i'm doing raid on ssd with small disks
and just not storing too much at home
with one backup hard-disk
so at least my important data can be on  raid
one cool thing about ssd's is it's easier to do backups
because it forces people not to store  too much data in one place
heh true
yeah, makes it a lot easier to use time machine
hahaha
and i have some files i access remotely
and they're on hdd raid far away from me
other thing i noticed is 2.5" hard-disks tend to be quieter
indeed
so i was thinking about raiding 2.5" hard-disks
like sas disks?
nah
or like notebook disks
like 7.2k laptop hard-disks
they also have high latency often though
and higher than avg failure rates
yeah
if they're on 24/7
hence i only thought about it
yeah, haha
you can get 7mm ones
the same size as ssd's
it's insane when you think about it
sas disks aren't normally that thin
all that mechanical stuff in that small of a chassis
pretty wacky
ssd's are actually in oversized cases atm
even when they 7mm ;/
yea, they're tiny
if you see pictures of people who have opened them up
those little msata ones are funny
yeah
http://ts2.mm.bing.net/th?id=H.4733093869193817&pid=1.7
err
wasn't sure if that link would work but it does
it said bing
i assumed it was a mistaake
haha
haha yeh
was that mastaa
http://cdn.pcper.com/files/imagecache/article_max_width/review/2013-07-25/130721-132916-7.65.jpg
ahh yeh it is
but see even the normal ssd's you buy are smalleer than the case
shit man, that's 1TB...
yep
yeah they're hell expensive
but out there
i'd go for 3x500gb in raid 5
over 1tb
i'm doing 3x120 in raid10 and raidz
mercutio: I'd just thought "gee, look at all that room if they wanted to stuff it with more flash, make it a 1TB... oh, it already is"
haha
i think 500gb is more popular
typically use duckduckgo, but image searching goes to bing via !bimages bang syntax @ ddg
duckduckgo?
https://duckduckgo.com
https://duckduckgo.com/goodies
it's on amazon
i wouldn't count on it not being tracked
i wouldn't suggest that it being on ec2 means it's being tracked
it may be
i doubt it
it's shared storage
by a company that is into tracking
at any rate
not why i use it
ahh ok
https://duck.co/help/company/architecture
http://highscalability.com/blog/2013/1/28/duckduckgo-architecture-1-million-deep-searches-a-day-and-gr.html if you're curious
that is interetsing
dunno why they use amazon though :/
why wouldn't they?
low network speeds
bad routing
high cost
low performance
hasn't been my experience at all with aws
well architected stuff will run well on aws
maybe network speeds are good for you, but id on't get good speeds to amazon singapore, amazon east coast, amazon ireland?
for some reason lots of stuff is on amazon east coast, and lots of ti tends to be slow
that's the default region, iirc
and benchmarks clearly show the performance is worse than real machines
but, like i said: 16:20:01        m0unds | well architected stuff will run well on aws
if you have global reach, architect like you have global reach
not like you have users in the US and nowhere else
i had a friend with amazon, and his disk performance and cpu performance were shocking
was he using ebs?
they throttle disk i/o apparently
no idea
he had a local hard-disk
and ubuntu
if io sucked, it was probably ebs
and ssh was sluggish
ebs isn't intended to be used like that, but lots of people treat ec2 compute instances like VPS'
oh ok
yeah he was treating it like a vps
EBS is like network storage
it's not actually in the host
so IOPS are generally pretty awful
he did get some boost by upgrading to higher plan
i told him he should just get vps :/
yea, if your app isn't gonna need to scale a ton, it makes more sense to get some semblence of dedicated resources
lots of people that want to use amazon dont' seme to consider other options
yea, they have a lot of mindshare
http://docs.aws.amazon.com/AWSEC2/latest/UserGuide/AmazonEBS.html
With Amazon EBS provisioned IOPS (input/output operations per second) volumes, you can provision a specific level of I/O performance, up to 4000 IOPS per volume.
he was way under 4000 iops
i'd estimate more like 20 to 30
micro instance maybe?
dunno
yeah i think it was micro
at any rate, the point is that storage is independent of the compute instance
and he went to small
it was unsable for casual use
it had 768mb of ram i think
ah
and he went to 1.7gb?
not sure on numbers exact, but it was werid numbers :/
that said i'm on 768mb on arp
1.7 seems to beee small
where's micro
oh 0.615gb
so I hope this is not too far off topic, but I think its worth mentioning .. arpnetworks is the best vps hosting provider I've encountered to date. yet, anybody who aims to be highly available knows you can't put all your eggs in one basket. thus, I've found https://my.iniz.com/aff.php?aff=289 as another basket to put some eggs in. (yes thats a referral link). can't beat roughly $20 USD/year for 256mb/250gb xfer/10gb disk kvm vps ...
... .. but they've no serial console, no secure vnc option, no ssh menu option, no /48 of IPv6 (you get 1 address). fwiw.
solusvm?
I've seen that term. what exactly does it mean?
vps management thingie
it's a control panel with security issues
and haphazard development
most vps providers that aren't doing custom stuff are using it
it can only assign multiple single ipv6 addrs rather than contiguous blocks, iirc
it's vnc is exposed to the world  as well
and they use centos
and they've had actual security issues befeore
and their mysql database is messy
but it's like $10/month per server
so it's "cheap" for the provider
like i imagine  up_the_irons  has put MANY hours into doing his control panel
heh. like I said 'best'. there are many security things at arpnetworks I'm happy with.
one switch taking out all my vps systems for the better part of today makes me want _something_ to show signs of life elsewhere.
yeah i just recommend to stay away from solusvm if you can if you care about your data etc
toddf: yeah, gotcha
you seem to have more network problems than most toddf
i don't run any critical stuff anymore, just personal crap mostly
mercutio: I am on the test 1gbps ports, granted
maybe you should shift back :)
I consider it a kick in the pants that all eggs in one basket is bad period.
toddf: did you get your net stuff from earlier sorted out?
m0unds: yes, up_the_irons kicked the switch that had all the 1gbps ports and it booted up fine
mercutio: 'if I care about my data' aka .. solusvm is like putting a bitcoin wallet on a windows server on the net, expect it to be r00ted soon? ;-)
kind of
haha, windows is reasonably secure if you're not a dummy
i'm not aware of anything happeniing yet
not since the last big exploit thing last year?
but basically it's reasonably likely that in the not too distant future someone will find an exploit for solusvm that gives them root
and can go into all servers, or put trojans in servers automatically
and reboot :/
but the same as any problem
if you're not a normal target, then you'll probably find them just trying to doos or irc fromy our host if they get into it
mercutio: I'm planning on having 2ndary dns and perhaps some mx traffic but .. humm .. if only secure vnc or spice were possible so I could have secure access to console I'd just setup cryptoraid and require me to enter the password everytime I rebooted
hmm
why don't you just run backup at home?
I'm trying to get servers away from my home office. at least temporarily. need to re-org and redo.
ahh
well it up to you :)
i'm hestitant about dns
well, I have home business, but if any of my 3 public ip systems reboots, I have to reboot the cablemodem and ping a v4 address for them all to have internet access.
that's not reliable for hosting ;-(
beacues if someone exploits your dns
then they can put huge stupid long ttl''sgoing to wrong locations
yea, that's been happening more and more
but again if you're not a target that is unlikely to happen
maybe i just paranoid :)
well, what sucks is that it's hard to know whether you're gonna be a big target or not
I think you've made some very valid points.
like that dude who had a short twitter id - personally, no one cared about him or his blogs or work or whatever
yeah i just think people should know the risks
it's ok to know the risks and still take risks
I'll just not put my bitcoin exchange reflector on the iniz vps then .. ;-)
m0unds: oh yeh that was scary
godaddy ftw
yeah, hahaha
but it's one of those risks that you juts don't really directly consider til you read about it
oh i'd considered it many times
just like stealing bgp routes
it's not like it's a new problem
it probably doesn't help that i was aware that phone hacking was big in years gone by
new vector for a problem
and that often that could involve things like "social engineering"
well, that part of it is 30 years in the making
beforep eople really had security policies
didn't hotmail.com's dns expire once?
and someone else paid it?
http://news.cnet.com/2100-1023-234907.html
ok i didn't reailse it was in '99
now i feel old
haha
read the comment
yes hotmails dns did expire and the check for reimbursement from hotmail got auctioned off for charity iirc
oh so they did reimburse
orly? lolol
$500
still tyring to find out how much the auction went for
http://news.cnet.com/2100-1023-236181.html
ahh $7100
quite a bit
yea
and i assume that was worth more in 99
oh '99 was before the crash?
Yes. And when Hotmail was still on FreeBSD
hoh
windows 2000 eww
https://www.wolframalpha.com/input/?i=%247100+1999+dollars+today
(!wolfram bang syntax via ddg, haha)
@wa $7100 1999 dollars today
:P
$7100 (1999 US dollars) in 2014;$10120 (2014 US dollars), (based on Consumer Price Index);2.39% per year;42.56%
oh btw
@ddg cats
Cats: Together with BZN, they were key figures of what came to be called the Palingsound, an umbrella for artists residing in Volendam.
[Related] Cat Meanings http://duckduckgo.com/d/Cat?kp=%2D1
[Related] Cats (musical), an Andrew Lloyd Webber musical http://duckduckgo.com/Cats_(musical)?kp=%2D1
hah
oh that's damn cool
it's (@ddg) surprisingly limited :)
@wa $7100 1929 dollars today
$7100 (1929 US dollars) in 2014;$98270 (2014 US dollars), (based on Consumer Price Index);3.14% per year;13.84 ×
@ddg how do I banana?
brycec: I'm sorry, the DDG Zero-Click API returned no results. As this is a Zero-click Info API, most deep queries (non topic names) will be blank. Quite frankly, I suck cocks.
hahahah
It's worth pointing out that it was passport.com that expired, not hotmail.com
i wonder how long ago they phased that out
the live.com stuff still exists, and it's mostly gone
mercutio: AS 11799 is The BGP Collective, which we turned up several months ago
such a funny name
haha
is the BGP Collective anything like the Borg collective?
:)
yeah, lol
RESISTANCE IS FUTILE. YOU WILL PEER WITH US.
we will...add your network distinctiveness to our own?
mercutio: yeah, there was a reason I put in TONS AND TONS of hours into my own control pane, which is to provide something nobody else has.  i had a theory that people can smell a copy-cat, and I didn't want to be that
It may not be "pretty" or shiny, but ARP's control panel is definitely useful and unique. :)
^
brycec: staticsafe : ty!!!
:)
It definitely feels "better" to have a homegrown control panel. You're not "some guy" who just bought a license for SolusVM or whatever and is running this in his spare time. You're comitted and you really do work on it.
go up_the_irons !
brycec: yeah that's exactly what I was thinking too.  i didn't want to feel like a reseller.
one day i want to migrate it to a nice Bootstrap admin template, then it'll have some polish too!
http://imgur.com/SNq98Rx
up_the_irons: oh, i thikn you had direct routes before?
maybe it's part of the migiration
i think it's going via the new router now
brycec: i have more faith in a bad cp homegrown than the average solusvm host
at least they must know what they're doing a bit to get the cp going
for a moment once i thought it'd be cool to try making my own cp and try and sell it like solusvm
then i remembered who the customer base would be
lol
i would like to see things done better everywhere
but it's complicated
anyway, it's the support system needing top level posting, which is my biggest gripe, rather than anything about cp or anything
but that's not a huge deal
mercutio: direct routes b4?
i don't remember seeing that other ip in there
i think it just went via s1?
to any2ix
maybe ip just changed or reverse changed and so looks diff
mercutio: ASN 11799 also has Any2 peers, some of which I don't have, so that's where that hop comes from
ahh ok
well at least it's working again
(there's a general ntt issue effecting me, but not from arp, cos arp is going via any2ix)
forward path is via ntt and seems fine.
weirdly arp seems fine weird
oh it came right anyhway
*affecting
heh
s/is/his/
 >>> brycec tips his grammar nazi hat
bah typo
seems like there is some sort of issue between Verizon and NTT
http://paste.unixcube.org/k/1d1fd7
i'm not going verizon<->at&t
and sitll hitting issues
i think it's a ntt issue
and is it still happening
NTT says no
i've seen multiple different ntt issues recently
"We are not seeing any packet loss within our network."
haha
it sure looks like it though
on diff providers, at diff times
it may be something like link aggregation with one link having issues
yeah. it sure is annoying though
well that traceroute suggests reverse path issue
can you get the reverse path traceroute?
basically because it goes 3.7% to 0 to 3.6%
to 3.2%
the fact it was fine from one point in the middle, and the numbers are diff enough suggetss enough samples to suggest it's sending in other direction being problematic
and like i had find traces to arp over ntt with any2ix return path
but some bad routes to other destinations coming back over ntt for at least one of them
(hard to check reverse path with random destinations)
the reverse: http://paste.unixcube.org/k/67f9a6
hmm
may need more traces
also, to ntt.net from Verizon: http://paste.unixcube.org/k/3f3b0
err longer trace
but it gusggests that it's fine when it comes into ntt at least
looks like the return path changes in Dallas
and that it's strugglign either within ntt or ntt to verizon
one test i find useful is iperf in udp mode and doing both directions
often you'll see only one directino having packet loss
I'll try that
usually i do 2 megabit small packet size