#networking
1 messages · Page 315 of 1
yes, they had the -EM devices with extra memory for BGP in routeros 6.. the extra memory is probably not necessary anymore with v7
I bet unifi would do the opposite
they would add more ram, and have a 300% profit margin on the modules
their high end gear so expensive
@tender hazel Just saw the nanog email this morning of others talking about the Akamai traffic
https://mailman.nanog.org/pipermail/nanog/2021-April/thread.html
@hollow marlin 🤣 https://mailman.nanog.org/pipermail/nanog/2021-April/212976.html
thanks.. I was checking nanog yesterday and didn't see anything
@hollow marlin lots of work ahead: https://i.imgur.com/4IS1ZpU.png
AMX-IX hit new record
9.935Tbit/s
I compared some other exchanges
and turns out
this is actually the best one to be on
psh, 4 of our baby MX10003s can handle that no probs
Its connected to everything
This complex
is the heart of european internet
@hollow marlin these are all within 5-30 mins driving
I guess 1 atom bomb
and rip europe
@hollow marlin whats the most extreme bandwidth in a single unit you've seen deployed?
like, how does an IX
do the actual peering, is that just a patch in the exchange?
between two switches ?
most of these buildings
do up to ~1tbit/s
except the national institute for particle physics
they were sucking 2tbit/s
you all get an assigned IP on a common subnet
sometimes you may peer directly with others, but often you get the peering by peering with a single route server at the IX and everybody who peers with the route server exchanges routes
route servers are helpful because if you want to peer with all 100 people at an IX or however many there are, you don't have to contact them individually
thats what AMS-IX says they do
they have a single route server
and provide their services through that
but how was just beyond me
yeah but even when they have a route server, people don't always have to use it - some companies have a direct peering policy where they will only peer directly and not through the route server
and what is that then?
you get a direct fiber link between your gear and the other peer?
no, you still peer with them over the shared subnet, but you create the peering link with their IP directly, not the IP of the route server
or do they do some kind of switching with vlans on the endpoint?
@tender hazel so this subnet
that's a public range ?
yes, it is generally a public range
what if you had two ISPs that wanted to directly exchange routes?
would they still use that common gateway from the exchange?
or is that what it literally means
one big giant street
and everyone sets up shop
Nothing extreme, 80-100gbps. Majority of our customers are business but still a good amount of residential
the exchange doesn't really provide a common "gateway", it is a shared network where people can peer over, and the route server will be used by most to automate that
@tender hazel but what if I was to say, build a building with servers in it
and you get multiple x-connects with a fiber to different exchanges
you announce that route, to all three of those exchange points then?
AMS-IX is looking at RFC5549 to help conserve public IPv4 space btw
mikrotik supports that in routeros v7
you can do BGP over either v4 or v6 and advertise both v4 and v6 but it is generally advised to peer over v4 for sharing v4 routes and over v6 for v6 routes
so you would set up two peerings with each, one for all of the v4 stuff with the ipv4 peering address and one for all of the v6 stuff over the v6 peering address
Much more than dual, there are tons of AFI/SAFI
Yes, v4/v6 are typically separate just for ease of management. Having them in the same peer can have some odd limitations
I don't think RFC5549 has any specific dependencies on BGP.. but I don't think they are using it yet in production
they are just preparing for eventually going to that
It does, its based on a modified NLRI. Not sure I have seen OSPF/inter-as-inter-as allow v6 next-hop for v4
Im sure it exist in the latter
All our IX peers are /31. But at their scale it makes more sense to move to v6 next-hops
ok
ahh yes, it's been a while since I read the presentation
slide 24 covers the behavior
it runs on all devices, but it is not yet ready for production no
it is stable enough to use at home
but I certainly wouldn't put it in any sort of important role
@tame carbon Just wondering do u know if its possible to watch blu ray disk copies on vlc on ur phone? (via that network sharing) The ones that u open on vlc by clicking 'open disk > blu ray', I'm thinking no rn because they have javascript menus and stuff but wondering if u know.
@tender hazel So I'll do that fiber 1000/500 setup on v6 :P
@thick minnow you need a network stream for something like that
oh so its in this case that streaming through vlc would actually be useful>
well if you have a blueray
might just be easier to use handbrake or something
HandBrake is an open-source, GPL-licensed, multiplatform, multithreaded video transcoder.
You can use this to turn DVDs and BluRays into files
Yes.. anything production like that I wouldn't run on v7
Id like to run it on my AC2 but the forums say otherwise
I'm running v7 at home on my rb4011 wifi
it's the first version that hasnt' been rebooting like crazy on me
it's stable with my config and I have a fairly kitchen sink config running on it
AC2 is reporting at least the last 2 betas causing reboots/freezes
@hollow marlin most of the reboots/freezes in my experience are due to the config
routeros 7 doesn't properly update the config syntax when going from one version to the next like routeros 6 would
the stable v6 on the rb4011
so you can have a syntax that is invalid that gets loaded and it causes instability
I've never had any issues with
the ac2 had issues with wireless once
after 120 days
reboot fixed it
to upgrade with stability you need to export the config, upgrade, reset to no defaults and paste the config back in
that way it will be validated by the parser for that version
I don't really believe the people who stay it is not stable with the hap ac2
they probably did not reset to no defaults and paste the config back in like you should
you can always use partitions to test v7 on your hap ac2 safely
Yeah, Id recommend that regardless since v7 is almost built from the ground up. I know routing config not updating was a big problem for many.
I have not messed with that yet. Is there a second flash that the partition lives on or just single flash? AC2 only has 16mb and base package take up most of it. Only reason I ask is Cisco/Juniper have secondary flash for failover
That said, I have 2 more AC2s on standby and I could test with one
oh only 16MB.. I thought it was a bit more for some reason
yeah for partitioning you need one with 64MB at least I think
Ill give it shot on a spare. If only Mikrotik will cut it out with the 16mb flashes
microsoft is having dns issues
fun
https://downdetector.com/
You can see how everything related to microsoft is just having issues
NOO minecraft too
@peak cloak soon discord will be down then too
you drop the a bomb on microsoft
and poof, no more fortnite
discord uses google cloud, if microsoft buys it I guess they will have to move to Azure?
might not just be related to microsoft
Cloudflare is still fine tho
unless Telus uses Microsoft
itll be shoved into their global offerings clouds they set aside for public service use most likely. So yeah, Azure
anyone here familiar with IIS URL Rewrite?|
I would like to add a site into IIS (fake directory since the app doesn't use IIS)
Then create a rule so that I can access this web app from the internet. myapp.mysite.com should point back to localhost:8080
so reverse proxy basically?
<rewrite>
<rules>
<rule name="Reverse Proxy to webmail" stopProcessing="true">
<match url="^webmail/(.*)" />
<action type="Rewrite" url="http://localhost:8081/{R:1}" />
</rule>
<rule name="Reverse Proxy to payroll" stopProcessing="true">
<match url="^payroll/(.*)" />
<action type="Rewrite" url="http://localhost:8082/{R:1}" />
</rule>
</rules>
</rewrite>
I am trying to configure my Unifi Security Gateway for the first time. I am able to get through the configuration process to set it up with the Controller software, but when I click finish, it gives me this error.
"{Mac}" is not a valid Target.
And I know that the USG does not have wifi capabilities, but the setup process for the controller has me entering in a SSID and password for a wifi network. How could that be?
@plain siren great, I'll try that. thanks
you can skip the wifi step
can someone help me with a vm it keeps deciding to stop working
i got this base station for a home security system and discovered it seems to function like a mesh router; i still don't quite understand how wireless mesh networks work, and i'm worried about it potentially impacting internet bandwidth
👍 Im in the controller. Now when I go to the Unifi devices menu, I can see the USG on the list, but when I hover over it, it says "Managed by other"
did you try setting it up before?
you may have to reset it to release it
😩 http error 404
Present, would you know anything about setting up VLANs between Unifi and MicroTik?
Just got my switch setup but my VLAN WLANS and wired VLAN connections are not getting connection.
Yes but I couldn't get passed that last error I was getting
well I've set up vlans between my USW FLEX mini and my dlink switch
done it at all between unifi and mk?
don't you use mk a lot or am i thinking of someone else?
crystal
ah
but vlans are standertized
yeah, it was 12 am for him so he wasn't able to help atm lol
it's just a sfp+ link from udmp to crs326
so in the controller I would just make the link be trunked
in the controller it's the all option
ah
make sure in your networks tab you have the correct vlan id setup
i've been using the gui
i assumed, yeah
I can show my d-link config
idk why bridge is there in untagged
is that your trunk port?
oh it's by vlans like that
@tribal ferry there is nothing tagged
what do i need to tag?
your trunk port should be tagged
my uplink is sfp-sfpplus1
trunk basically means all tagged
yeah tag that
on all vlans?
yes
and i guess i need to un-untag it
wdym?
can someone help me with a vm it keeps deciding to stop working if u can help please dm or @ me
remove it from being untagged to make it tagged
oh yeah
and two, you can only have a port be tagged or untagged per vlan, not have it be vlan 12 untagged and vlan 13 untagged
but you can gave vlan 12 tagged and vlan 13 tagged
on one port
well i need vlan traffic to be able to pass through all ports
oh your connecting a vlan supported switch to distribute to the end devices?
yep
ah ok
well
then everything should be tagged
some of these are going to switches
one is going to an ap
another goes to my cable that goes upstairs
yeah so any port that is going to another device that supports vlans needs to be tagged
Oh fun VLAN Trunks
so what should i change in this config?
because this is all a vlan is
tag everything
and remove all untags?
yes
the reasoning behind this is that it needs to be tagged so the final switch or ap will know what packet is on what vlan and untag it on the right port
or ssid
so what should i use untags for then?
anything going to a device that is only on one network or client devices
so for example
I have my AP on a port that is tagged both vlan 20 (trusted) and 30 (guest). The AP untags the packets that come in and sends it out to the clients, and tags packets coming back
but I also have a roku connected to my core switch, and that is untagged on IOT, since it's a client device
so, tagging every port in use besides the trunk will allow vlans to flow freely?
well tagging every vlan is called trunking
WOOOOHOOOO!!! I got the USG to adopt. Now, since I'll be using the USG as my main router, and my Linksys WRT3200 as my WiFi access point, what do I need to do?
good diagram to wrap around for brain chewing:
put the linksys in AP mode
yeah vlans are hard to explain, they finally clicked for me once
many people confuse vlans with subnets
it's more that a subnet is carried over a vlan
^
still got to wrap my head around this lol
a VLAN Trunk basically means you dont have to run 3 cables to another switch just to have 3 networks (subnets) sent to that switch
in this case, those 3 VLAN's cant talk.
not on same networks, but you could use the router, static routes for example to allow... VLAN 1 to talk down to 2 and 3 but only allow 3 to talk back up
It keeps the "Vision" clients have on your network limited to your walled garden... the vLAN
ofc there is more to it and it has more uses
yeah unless you use a Layer 3 switch, if you want to go from subnet on a vlan to another you need to go through the router
Get 3 straws of diff colors, wrap them in something. The wrap is the trunk and each vlan is a straw. Your data can only pass through 1 straw despite being in the same wrap
A router would allow you to control mixing at what would be a "smart" junction/combine valve as needed (L3 would just be a junction valve without advanced reaching control like a router has)
if i have a friend
and i wanna send him like a 200gb file
over the internet. what is the best way for me to go around this?
for free
upload to a free file sharing site or self host a sharing service
for 200gb?
at reasonable speeds?
do you know any such sites?
@hollow marlin have you done any QoS at fairly high rates? like close to 1Gbps or so?
@hollow marlin lets talk about graceful failover again
You wouldn't use statics as they are already connected, they'd already be routable. You would need ACLs to stop them from routing from one VLAN to another
do you know if BFD over BGP kills GF?
I know I sinned by doing that but it was for the sake of simplicity despite being mostly wrong
I couldnt think of another way
Is there a way to make it where I can access the USG interface at 19.168.1.1 & Linksys interface at 192.168.1.2?
All our access edges are set with QoS. 10/40gb. Primarily for voice priority
yes
Give .1 static IP to USG and .2 static IP to linksys?
Ooooo you have an update on the issue?
i do but this is not the same thing
Do I do that in their respective interfaces?
Yus
@clear igloo I found a bug in NX-OS 9x that exists in 7x but they had no clue 9x did it. no code upgrade path
its rare to trigger it. I triggered it by flapping a VPC port channel multiple times using the interface range VS one interface at a time
@hollow marlin we've been running into issues with mikrotik doing queueing for connections nearing 1Gbps.. we have one large site with hundreds of customers and 1Gbps coming in, and the best effort queue is only able to hit about 850Mbps at the highest.. if I disable the qos, the usage immediately increases to like 950Mbps
Oh, I missed this comment. It shouldn't, BFD just informs BGP to pull the peer on loss, but by that time, BGP already knows the peers going down from GF
we're not seeing GF work 😦
we get packet loss the second BGP goes down
oh i should mention the links go down too
its complicated
there is no individual CPU core that is even close to maxing out when the clients are hitting that maximum
FW A <> FW B, then to the switches. FW A has full link failure so FW B with the same peer IP turns up all its interfaces and tries to resume
we get like 30sec packet loss
this is how Palo failover works
standby FW keeps all links hard down
I'd get a PCAP, but if it's the same IP, FWB is going to be using a different sec/dst port and the session will have to time out.
Unless the FWs are clustered
clustered
i was not able to see what BGP was doing because the second I started to try that I hit that bug and OOPS
Symptom: Ethpm got locked and causing vlan create/delete, copy r s etc relative operations fail. N9K-1# sh sys internal ethpm info all | in LOKD LOKD @ 838755 usecs after Wed Feb 17 20:06:38 2016 LOKD @ 977171 usecs after Wed Feb 17 20:11:06 2016 N9K-1# copy run st [########################################] 100% 2016 Feb 17 20:09:47 N9K-1 %SYSMG...
"quite timing relative and very hard to hit."
FU cisco
Hmmm... DM me a sketch with some dummy IPs
damnit where did my visio template file go
I have not dived into Mikrotik queuing, do you have one of the configs I could take a peak at. WRED could be kicking in when you are getting that close
yes we are using RED
Let me see if Mikrotik has the thresholds or if they're configurable.
/queue type
add kind=red name=red-queue red-avg-packet=1000 red-burst=20 red-limit=60 red-max-threshold=50 red-min-threshold=10
/queue tree
add bucket-size=0.01 max-limit=980M name=Splitlake parent=MHT_347_SPL queue=red-queue
add bucket-size=0.01 limit-at=10M max-limit=10M name=7_monitoring_and_routing_splitlake packet-mark=\
monitoring-and-routing parent=Splitlake priority=1 queue=red-queue
add bucket-size=0.01 name=0_best_effort_splitlake packet-mark=no-mark parent=Splitlake priority=6 queue=red-queue
add bucket-size=0.01 limit-at=5M max-limit=5M name=6_mgmt_traffic_splitlake packet-mark=mgmt parent=Splitlake priority=2 queue=red-queue
add bucket-size=0.01 name=5_ent_priority_splitlake packet-mark=ent-priority parent=Splitlake priority=3 queue=red-queue
add bucket-size=0.01 name=4_ent_splitlake packet-mark=ent parent=Splitlake priority=4 queue=red-queue
add bucket-size=0.01 name=3_retail_priority_splitlake packet-mark=retail-priority parent=Splitlake priority=5 queue=red-queue
add bucket-size=0.01 name=2_background_splitlake packet-mark=background parent=Splitlake priority=7 queue=red-queue
add bucket-size=0.01 name=1_scavenger_splitlake packet-mark=scavenger parent=Splitlake queue=red-queue
about 78,000 PPS going through the queue
mikrotik for the most part just uses the queue types and HTB support that is built into linux already, although they added their own in a few cases like pcq
I can change the queue type but I get the best performance from RED in general
we mark the packets by using bridge filters which are set from the MPLS experimental bits
Are you able to walk me through the process on both the USG and Linksys to set the static IP?
The USG I know you just punch in https://192.168.1.1/ On first connect and the WebUI will ask you to set it a an IP
How?
sftp, scp...
What do those stand for
Is there any way I can just send someone a file with utorrent?
Because normally when letters are involved that means hard to set up.......
can someone help me with a vm it keeps deciding to stop working if u can help please dm or @ me
I am using unraid the graphics card is a gigabyte rx590 and what’s happening is when I send it to the gpu it works for a couple min then black screens and is I try and start it again it freezes at the tianocore screen
@thorny vector I’ve also tried routing the vbios, booting in legacy, using Q35, using SeaBIOS instead of OVFM
Any errors kicked back? @sturdy knoll
The logs don’t show any errors
so i was tryna setup zerotier and i have 2 devices on a network rn both show tha status as ok but when they try to ping each other all packages end up as lost
did i miss something
also both show up as online in the website panel
anyone got any ideas?
@thick minnow no matter how you twist it, the initial transfer is limited by your internet upload speed, so self hosting is best. Look into a few options like sftp, scp or even a web server. If your upload speed is super slow, it might even be faster to just overnight ship a flash drive
I guess that last option isn’t really free
How does self hosting bypass the upload speed limitation?
@thick minnow sending something via torrent requires an online tracker managing the torrent connections (seeders and peers) in order to coordinate the flow of data
If you are just trying to send a file to one person, torrent probably would not make sense - you would have to register the torrent with some kind of tracker and have the other person get the magnet link to get the seeders and peers from the tracker
torrenting is only faster if you can potentially have a bunch of different people serving the file to one person
Nah that won't be the casw
Just need to send a single big file
Well, it won't really be quick in that case, but from my experience it'll probably be a lot more reliable
So which method do you recommend?
how big?
that's really big
But how do I make the tracker?
I haven't done it before myself
and that's a really old link
the idea should still work as long as you use a current tracker
That article doesnt explain the tracker.
It just says if your using a private tracker set it to private otherwise public like wtf?
yeah so with torrents you probably need to use a public tracker which means someone else could download the file too as long as you are offering it
nothing to protect it
it may not be suitable for what you need
sending a file that huge is not a normal thing
may I ask what it is that you are trying to send that is so huge? you don't have to answer if you aren't comfortable answering
@tender hazel summer beach photos of your mom
@thick minnow Torrents are for peer to peer sharing
eg: thousands of computers and users
I need a way now to disable the quick setup feature in the mikrotik app for our users
I'm hoping that mikrotik will be able to add some kind of feature to allow control for that
yeah this is different though
I don't care about winbox, and we already remove it via webfig
this is just about the mikrotik phone app
the mikrotik phone app uses the winbox port so I don't think any sort of filtering will work
attacking this from a developer's perspective xD
the issue for me is that we can't have our users going into quick setup on the phone app
because here's what happens
@tender hazel give them a presentation, with the threat of sodomization by baton
if I go into quick setup on the phone app with one of our routers and just next next next next finish
it disables the PPPoE interface and changes all wireless interfaces to open authentication
I always tell tik users to NOT TOUCH quickset after configs have been done xD
@tender hazel wait what about user groups?
can;t you restrict PPPoE setings and such?
there is no group for controlling quickset access
and the issue is that we actually want our users to have full control of their routers
even the ability to lock us out, if they wish, and if that helps them to avoid wanting to buy their own third party router
you just want to protect them from themselves
yes
the problem is that in the phone app the quick setup is this enticing big icon with a magic wand and sparkles
displayed very prominently
practically screaming "click me" at the user
I was looking for a click me meme
couldnt find one
Provided to YouTube by The Orchard Enterprises
Soulflakes · DePhazz
Prankster Bride
℗ 2016 phazz-a-delic
Released on: 2016-11-04
Music Publisher: Copyright Control
Auto-generated by YouTube.
I am not sure if this is the perfect channel, But I was trying to port forward on my router and it dosent seem to work. Any ideas please?
@tacit leaf have you had any port forward on this network before?
You might want to check if you are not behind a CG-NAT, and have your own public IP
Never, I had them on previous network and it worked all fine
Is there any way to check it?
@tacit leaf open commandline and run a tracert 1.1.1.1
you can look at the path it takes through the routers, if you are on a CG-NAT or not
Says trace complete
ok screenshot?
not CGNAT
@tender hazel that ISP does have some shitty backbone lol
It probably does lol.
@tender hazel you certain?
I don't recognize any of the providers, but what exactly do you mean
@tender hazel what about the 10.0.0.0/8 ?
that's not any concern - we use RFC1918 for point to point links inside our network since they aren't used for any customers and the only thing they show up in is traceroute
so if our own customers traceroute to 1.1.1.1 they will hop over a few RFC1918 links in the 10.0.0.0/8 space
@tender hazel the other points break traceroute lol
if there was CGNAT here you would see something in the 100.64.0.0/10 space and you don't
yes they do, but that looks like their upstream similarly using RFC1918 addresses internally
I am soo confused xD
when the upstream provider uses RFC1918 internally to save on IPv4 public space you will see timeouts for a few hops that are addressed via RFC1918
it is the only downside of using RFC1918 for point to point is that people who aren't on your network will see timeouts for those hops and think something is wrong
@tacit leaf go to your router config
but in actuality it is good practice for conserving public IPv4 space
@tacit leaf find that port forward config, and have a look
send us screenshot if you can
@tender hazel I was always under the impression that they use 10.0.0.0/8 internally, but thought they use some kind of encapsulation to make it appear as though it is a direct link between two routers
looks right
@tacit leaf does the router WAN IP correspond to the public IP you get if you search online for your own IP?
It’s different
@tacit leaf what does the WAN on the router report?
no, the problem with encapsulation is that it adds overhead which can impact MTU, so you don't want to do that when it isn't necessary.. you can use any sort of RFC1918 subnet to connect any two public subnets.. it works perfectly, the only downside is that you don't get a traceroute response from the RFC1918 subnet unless you are on the same provider network as the RFC1918 subnet
those 40 bytes too much? xD
Oh, is there anything I can do about it
that's RFC1918, CG-NAT shouldn't use that range
where are you seeing that 172.16.15.89?
ugh
then your ISP is using RFC1918 space as CG-NAT instead of the actual CG-NAT space
Yeah
Developing world has got the wrong end of the stick in ipv4
they started using internet widely after we already ran out of IP addresses
india has better IPv6 deployment than most of the world
yeah, v6 xD but not v4
@tacit leaf your public IP is shared amongst other users from your ISP
Is there anything I can do about it tho?
thus, you cannot port forward.
Ohhh, bruh thats shit
yes you were right - it didn't show up in the traceroute though
@tender hazel I didnt think they would have a public range on their PPPoE servers
Can I ask them to give a personal IP address?
@tacit leaf usually a business plan would give you your own static public IP
Like, sometimes feels like here in europe its so easy to get your own public IP, even as a household consumer
and that everywhere else, its utter shit
Ohh, but getting one will fix my issue?
South Africa still has a bunch of IPv4's
@tacit leaf the problem is that we are out of IPv4 addresses.
and getting your hands on one is a combination of the right ISP & some luck
Hmm, soo it’s worth a try atleast
It looks like they do
which is actually strange
afaik, there's no point in using a public IPv4 as the ISP address on a PPPoE tunnel
I mean we do it but we only do it because we have a ton of addresses and it is just one of those
and it is already routed to the device
we could change it to an RFC1918 in a few seconds and it wouldn't do anything
except it would free up some addresses
Meanwhile, I'm sitting on a /29
and not even using two of those addresses
yup in the case of the provider in this case it makes sense to use RFC1918
every last address they can give to a customer is valuable
I think we have finally convinced mikrotik to add /31 support
yeah.. /31 is a special thing
it was designed to conserve IPv4 address space
no network or broadcast address
the device has to be designed to support /31
ok but a downstream client can use a /32
when there are only two devices there is really no need for a network or broadcast address
and not care ?
yes, a /32 works too, but the problem is that /31 is so standardized, mikrotik seems like they are doing something weird by not supporting /31 and supporting /32 only
mikrotik and linux in general allow you to do /32's on both sides which in some ways is even better than a /31
because you can have point to multipoint with a /32 - one /32 address on the "hub" with multiple /32 addresses for each "spoke"
the issue is that /31 is considered so normal that to not support it, you seem to be strange
I thought /31 was a gimmick
in north america, the fact that mikrotik doesn't support /31 just makes them seem more like some tinkertoy router vendor
and that /32 and /30 are the only ones that make sense
either you have full broadcast domain & network, or none at all
every other major router vendor supports /31
time for routerOS to step up to the challenge
they been struggling because of the old kernel
the /31 support shouldn't really matter because they support /32 and it can do the same thing theoretically but it is just the optics of the situation
we had techs waste hours and hours and hours trying to figure out how to get one of our mikrotiks to connect to an upstream ISP that was giving us a /31
F
all they had to do was configure it for a /32 instead but that isn't exactly clear if you don't know what you are doing and the upstream says "assign this address with /31"
@tender hazel that was me with 0 experience about vlans and subnets
setting up fiber on my RB4011, called my ISP like 5x
I think they are glad its all working now lol
IPTV was even worse
All they gave me was a tutorial for a draytek
@tame carbon what should i exactly ask my isp?
@tacit leaf ask them if they provide public IPs to customers
probably not... but never know
its probably going to cost extra
Aight will ask
@tender hazel lol shared IPs are so annoying when IP banning. On that minecraft server we had lots of issues with people from Comcast
you ban 1 person
and 1 week later ,you get another user complaining
"I am banned"
IPv6 is the only solution to all this
@tender hazel It got so bad to the point, that I started logging IPv4 <--> User
the thing is that people are still asking what do we need IPv6 for? nat is great!
So I could get a picture of who would be using what, and if I could ban without a problem..
and you sometimes find 1 public IP
it is insane if you can't explain to someone what the problem is with NAT
used by 20 different users over 1 year's worth of time
why it is bad, always, always
@tender hazel lol "please can you unban me, my brother was being an idiot"
nice NAT you have there.
IPv6 is the solution
if two people have the same IPv6 address exactly, they are the same person 🙂
Our mc server doesn't support v6
protocol might these days, but my own database doesnt lol.
It stores 32 bit unsigned integers
well the thinking has to change to make things better
it is changing now - PS5 now actually uses IPv6
DROP PROCEDURE IF EXISTS proc_ban_ip;
CREATE PROCEDURE proc_ban_ip(
IN addr INTEGER UNSIGNED,
IN reason VARCHAR(128),
IN duration BIGINT
)
BEGIN
INSERT INTO ip_address VALUES (addr, NULL, NOW(), NOW(), duration, reason)
ON DUPLICATE KEY UPDATE
banreason = reason, banned = duration;
END;
^ :D
PS4 could get an IPv6 address but didn't use it for anything - you could ping it but it did squat
@tame carbon yeah so you will have to adjust that code to be more protocol neutral
@tender hazel the problem is the internal code that calls this procedure
it uses InetAddress
which on minecraft only applies to v4
stupid stupid stupid
@tender hazel ^ lol my schema
it logs who was the first user to use an IP
and any other users that might have used it
first time, and last use
and also tracks how often an address is used
java?
lemme find the calling code
@tender hazel yeah but the server only listens on v4.
OVH doesnt support v6 
what's OVH?
idk
it burned down
@tender hazel see the problem is here
@Override
protected KnockturnPlayer login(UUID uuid, InetAddress address, String username) {
try (Connection connection = KnockturnCore.getMySQLConnection()) {
PreparedStatement ps = connection.prepareStatement(
"CALL proc_player_login(?, UNHEX(?), ?);"
);
ps.setString(1, address.getHostAddress());
ps.setString(2, uuid.toString().replace("-", ""));
ps.setString(3, username);
ResultSet rs = ps.executeQuery();
rs.next();
Long discordid = rs.getLong("discordid");
if (rs.wasNull()) {
discordid = null;
}
return new KnockturnPlayerImpl(
rs.getString("username"), rs.getLong("firstlogin"),
rs.getLong("lastlogin"), uuid, connection, server, rs.getBoolean("banned"),
rs.getString("banreason"), discordid);
} catch (SQLException e) {
logger.error("failed to login", e);
return null;
}
}
getHostAddress() always returns v4
interface is the same for v4 and 6
50$ yearly for a public ip 
It is yes
I can get another isp and pay 2-3$ extra to get a public ip by default
@tender hazel its an interface that is subclassed based on the type
But that is moot
if the server only listens on v4.
then you can't connect via v6 yeah
any incoming connections will always be of Inetv4
Thanks a lot for helping yall 🙂
either way it is incredibly frustrating to see so many people having to set up these crazy workarounds like doing a VPN to a VPS
if (address instanceof Inet4Address) { //foo }
to get a public IPv4 address when they most likely already have a public IPv6
@tender hazel I wrote this code in 2014
and it still works :D
just needs to be dusted off for the eventual move to a new hoster
@tame carbon this illustrates how commonly used /31's are
@tender hazel could my ISP in theory use my FttH to provide a vlan which I can peer with in an exchange?
The fiber operator here can support 10G as well
in theory, but you need an AS number to peer
yeah, but I need to be on a special block don't I ?
I can't just announce a route through my WAN
lol advertise 127.0.0.0/8
BGP works over TCP so you can be several hops away from who you are peering with
like you can peer with someone halfway across the world separated from you by a dozen routers
oh really?
yeah
SO THAT. is what my african friend wanted to do
he asked me if i wanted to peer
cus he sits on a pile of v4's
I thought I couldnt
Might be a super stupid question, I guess theres no work arounds for this?
Pretoria Wireless Users Group is a South African wireless users group. It is non-profit community organisation providing a wireless community network in Pretoria, the capital of South Africa.
PTAWUG has been created as the community’s answer to South Africa’s restrictive telecommunications environment. The South African telecommunications indust...
@tender hazel he wanted to peer with this ^ xD
he's the operator for them
but just peering won't actually give you connectivity to that, you need to have routes installed outside BGP that will allow you to direct traffic there
right
@tender hazel lol some of the crazy stories I've heard from them
like getting hijacked while on the job to fix a tower
by a guy with a BB-Gun
christ
they were in the car
@tender hazel yeah unlucky for the guy with the BB-gun
they had a real 9mm
I think they shot him in the leg and drove off
Pretoria is quite a rough neighborhood
@tender hazel opiate infested regions makes people do some bad things :/
That happens to AT&T dudes in Chicago too lol
im not even joking
taking down ctOS
you might be getting an IPv6 address from your ISP and that should allow you to do the same stuff but it depends on what the games (or whatever else you are trying to access) support
I actually just had a guy tell me a story how someone was shooting at him cuz he was trying to upgrade their internet but it turns out they were trying to rob him
minecraft
chicago
@tacit leaf and minecraft doesn't have full support for IPv6 yet, and it is something that is more their problem I would say, than your ISP's
whistles https://tunnelbroker.net/
@plain siren behind CG-NAT?
Does it matter? Its point to point
I thought you had to enter your remote IP into the link
yes you can use tunnelbroker over nat
you have to enter the remote IP but it doesn't have to be the remote IP that you actually have
it can be the remote IP you are NAT-ted to
^
and if it changes?
if it changes you have to change it, or you can automate it with scripts
Neither did their IT.
Or their heads
everybody seems to treat lack of support for IPv6 as no big deal.. super frustrating
I'll use it once tunnelbroker allows me to set GeoIP
I cant watch american netflix, library is terrible
your ISP should give you v6
you're in germany right?
Netherlands
the traceroute there is fine
European datacenter colocation, Connectivity, DDoS protection, Web Application Firewall, Qbine, IAaaS, NoCaaS, servers, internet exchange.
@plain siren ooh btw, I called my soon to be employer again, and I agreed to take the job
yeah I found that before, I don't play minecraft or other online games (I only play single player games myself) but it might work
They gave me a week to decide, but I jumped the gun this morning
Ill try 🙂
we've been hit in the last few days by this stupid call of duty update
we have hospitals not getting their required service because of call of duty
here's the problem in our case
our upstream requested a second bell cross connect one year ago
due to covid bell hasn't delivered yet
a cross connect in the same building they are in where they have facilities
so the call of duty update took them down
because they knew they would be running out of bandwidth over a year ago and it's taken over a year for bell to deliver
which is insane considering that it is a colo in the biggest facility in toronto
and if bell takes over a year to deliver a simple service to a colo in the biggest colo facility in toronto
net neutrality? lel.
canada
oh
we have net neutrality
but bell says they could not install because they have to get this part that is out of stock
some like part worth a few dollars
out of stock for the past year due to COVID
do they not do incremental patches?
people are reporting anywhere between 50GB and 80GB for the update
50GB is the low end
I'm not sure why there is the 30GB difference between some users and others
probably
they have updated most of the assets in the game
it is likely easiest to just update the entire thing
this is our third night dealing with those issues
it isn't as bad tonight but it is still impacting us
right.
My 25 page report on comparison between Relational and Graph databases is FINALLY FINISHED.
omfg. I hate this course so fucking much
There are a few services online that you can peer with for BGP monitoring purposes - you don't actually use them for routing but for monitoring
QRator is one
they are this monitoring service in russia
I dont need russian service
we aren't peering with them ourselves yet but I've signed up for an account there because they've picked up a few open ports in our network that we didn't notice
all the russian blocks are on my drop list
who needs russians on their network
same is for the chinese.
they have no business on this network
@tender hazel doesn't HE support some kind of BGP monitor too?
we don't really need to peer with them, and we don't, currently
no, not the same type
and the qrator service has alerted us to ports we have had opened that we should not
My ISP did that for me
we had a BGP port open on one router to the world and we only found out because of the qrator alert
they werent happy with the exposed port 53
on my public subnet
so much so that they emailed me, and immediately dropped the route
that's fine, except we are the ISP in this case
I forgot to configure the FW on my mikrotik
to only allow 53 from the DMZ itself
not WAN
Oh shit, get the whiskey, hes an adult now
something a bout DNS servers used for DDoS >_>
DNS Reflection
@plain siren I had some junior positions at two other companies, and I am also self-employed (I have VAT number)
we wanted to do a scan of our network for ports that could be used for DDoS amplification
I know im being an asshole
<3
but one of our employees was against it
your an asshole too so doesnt work I guess
No, I just suck at detecting sarcasm lol
Same difference
he was saying that if we started scanning for open ports that we would start calling our customers every day about open ports and that eventually we would be doing nothing except calling customers about open ports and go out of business as a result
so therefore due to the slippery slope we shouldn't ever scan for open ports
@tender hazel you want to portscan my network? :P
which of course doesn't make sense at all
no.. I wouldn't portscan anybody else's network unless they asked.. our own network is different, I think it is good practice to see if our customers are vulnerable
That's why I am asking lol
I have no idea. I did an NMAP scan when I first finished the config
how do i check if I have ipv6?
or that
if you get 10/10 you have ipv6
unless you can switch ISPs your only workaround is to get a VPS somewhere and set up a VPN tunnel to that
See, I can do v6 just fine
but i dont want to set up a mailserver to finish the certification
I am too lazy
so I am limited to ~60mbit
you can do v6 just fine over an HE tunnel yes but that won't solve the netflix issue
See this kind of shit wouldnt ever trend in EU: https://i.imgur.com/q4sdNOT.jpg
that's the same reason we couldn't use netflix on HE
we actually get blocked
we can't use netflix at all
if we are on HE
so I would have to turn HE tunnel on and off every night before being able to use netflix
I got sick of it and just left it off
I assume those aren't trending in NL and are instead just trending in US and you are getting them because of that
in our case it is blocked entirely
within europe there's no difference to netflix content
netflix won't even load
because single market EU
if I turn up an HE tunnel
and go online with it
I can get to everything except netflix
netflix says sorry you are blocked because you are on a VPN
yeah probably due to netflix not tracking everything
to enable/disable v6
I am in canada so netflix seems to have created special programming for that
I hate going to germany and watching youtube
if you are in canada and connecting to HE, netflix is blocked entirely
because when I return home
all the ads on youtube are german
for couple weeks
my HE tunnel is somewhere in New Jersey
our server administrator is from russia originally
sometimes he does updates for software
but I get 7ms to 1.1.1.1 even on v6
and our software changes to russian language
it detects his preferred language on his computer and tries to be helpful and chooses that as the language
helpful
so suddenly we go into the calendar and the months are in russian
or other crazy things like that
it is work to try to figure out how to roll out IPv6 on an ISP-wide basis
and the problem is that a lot of ISPs don't really see a benefit for them
they see it as just extra work for nothing, so they ignore it
yes
the curse of IPvX
I had to push to get IPv6 rolled out at our ISP
we are completely dual stack now
but I had to push for it
I remember that time, when it took 8 months to convince my coworkers that git was better than svn
even in an ISP there are people saying "but we don't need IPv6 for years and years, IPv4 is fine"
"merges" were an all-day ordeal.
and while in progress, nobody could push code
@tender hazel how are the IPv4 reserved addresses handled in v6?
yeah it is really that, you get so stuck on what you know well (Ipv4 or svn) that you are totally oblivious to what can be done better
sorry?
If my ethernet cable is damaged can it affect my ping
I've been getting really high ping lately.
@tame carbon yes, but I only know it because I've seen a few things online that reference it.. but it isn't automatic
I think it is designed for 464XLAT
@tender hazel an ideal dual-stack would be a NAT'ed v4 locally, and a public V6
The outgoing v4 would prefer v6 addressing
both addresses can be reached on v6
v4 only on v4
dual stack and 464XLAT are two different solutions
dual stack means the client has IPv4 and IPv6 at the same time, usually CGNAT on v4 and public v6
then...
@thick minnow
what are those v4 mapped v6 addresses even for
if nobody uses them
@thick minnow wifi?
464XLAT is a way for the client to only have IPv6 but be able to access ipv4 resources without having an ipv4 address to begin with
@thick minnow I see nothing wrong
my ping has been high lately
my understanding is they are for 464XLAT
ever since it was damaged
you don't need an IPv4 address you can just have an IPv6 address, you can make a request to that IP with a system that has Ipv6 only and it will respond
@tender hazel I'd imagine that the ISP would have some kind box that did this for you then?
yes
so you can connect to v4 from a v6 network
sounds neat
why arent we using this?
or is that the million dollar question xD
we actually are using this for cell phones in many cases
most cell phones in the US get only IPv6 addresses
tmobile for instance
Nailed it
they are using 464XLAT to get to all IPv4
It was a waste of space for a small point in what couldve been replaced with many other things that are avail now
The fact the assignment table is tainted with a transitional block just looks bad on the paper 😦
but one issue is that 464XLAT works great for all websites but doesn't always work with VPN's.. that is what is starting to impact meraki, when tmobile moved their cellular to 464XLAT, the meraki customers could no longer connect to their corporate VPNs which were IPv4 only
Every time I look at this I get conversation I get more confused, people do realize networks don't work solely on just getting sent out over what standard protocol for IP addresses they're using right?
I'm about to break out my cisco textbook and do some explaining lol
Honestly this whole thing should be moving much faster
ffs
OUT
we are this small-medium ISP and we moved to v6 dual stack a long time ago
The longer some are without IPv4 and some are without IPv6....
We have a problem
that leaves regions becoming inaccessible and growing
whats v4 and v6
a lot of people think that there is no point in IPv6 that we are fine with IPv4 forever with NAT
it is horrible
192.168.0.1
4 Octects in the Address. The IP Address. IPv4
IPv6 is 128 bits, 4x as large address space as v4
@thick minnow what do you mean more complex? 
no NAT
the largest address you can have is 255.255.255.255
If you were to check the largest number in binary (11111111) for a Byte (in base8)... its 255.
4 Bytes makes up your IP Address
Ok I don't understand the problem? Is the problem scalability or security?
Sounds more like a you problem, than a protocol problem
Because they solved the scalability problem all the way back in the 90's for IPv4
But the protocol works just fine?
pretty sure you can configure your firewall to filter invalid packets
the extension headers are often ignored because there are issues with the idea
but here's really where the problem is
they came up with an improvement for IPv4 like 30 years ago
we still don't have it rolled out
Gooooot it
if you want to make a new better protocol to replace it, we wait another 40 years?
not going to happen
they can't
IPv6 is it
and the longer that people fight against it, the longer it will take to get there
@thick minnow https://ipv6excuses.com/
IPv8, it's 1024 bit addresses!
You have extensions for days!
ok I thought your problem with IPv4 was scalability.
The fact that IPv6 is actually easier to work with but its apparently hard for engineers to grasp and roll out is frankly pathetic.
so you're happy with people having to buy VPS's to be able to play online games?
Honestly I can subnet all day, but for some reason I can't figure out hexadecimal so I'm part of the problem lmao
Im happy with IPv4
Except any new website or server that is currently acquired doesnt have an IPv4 address.... So how are you going to access it?
Fade away
I think I have a bingo guys https://i.imgur.com/POXgrSh.png
yup, we have a bingo
What if Discord moved to a new host when Microsoft Purchases it and you dont have IPv6
Wait. What?
yet internet is still growing.
So anything new is out of your reach
My local exchange: https://i.imgur.com/FTeUdou.png
The internet is growing. FAST.
and the amount of people behind a NAT is already annoying enough
every single user who is behind a CG-NAT cannot port forward.
or host services
And its even worse
because they only do NAT444 and no v6 support.
We've been out of IPv4 addresses since the 90's scalability isn't the issue, we solved it literally DECADES ago
IPv6 has proven to be in both terms of Management, Reliability, and an actual reduction on a switch stack thanks to the lack of NAT, an improvement. There is 0, and not a damn single one, excuses to not want to push and implement IPv6 other than pure laziness or complacency.
So that really really shouldn't be a part of your argument lol
What...
Yes, this, exactly
Each regional IP Numbers authority recently hit P3 Ending.
They are Revoking addresses to force the change.
This is some mental gymnastics lol I'm goin' to bed.
Mental Gymnastics
Documented by the horses mouth on said website
yeah you are
Yeah like Ubiquiti and they dont even support IPv6
Point?
@tender hazel sometimes...
Those are all excuses
No there isnt. They are nullfied by the fact we have no choice.
v6 is current. v4 is old.
We are OUT of IPv4. time is UP
@thick minnow lol I dont think you know who you're talking to
Rouing is anything but pretentious
No I am pretentious,
👍
Im also an asshole and a douchebag
The problem is, what are you going to do about it and I dont care.

Yeah but that is irrelevant
Vendors have had enough time to adapt.
more than 10 years
Vendors already have adopted it. Its not vendors
its implementations
They are adapting to adopting as thats the goal but either way
Semantics
yawn I'll just wait for my ISP to provide v6
@plain siren what even: https://i.imgur.com/XiEfwBV.png
Find me an enterprise device that hasnt meant industry standards of all points (Security, Reliability, etc etc) by any one of the big 17 out there that hasnt already implemented and even begun focusing on IPv6
2017 was 4 years ago
I said enterprise, not ubiquiti tier

some customers were forced to abandon meraki because of the lack of proper Ipv6 support and the mobile networks moving to 464XLAT
entire teams of salespeople prevented from connecting to the corporate VPN because the meraki didn't support IPv6 and tmobile on 464XLAT
Not the mobile networks faults that terrestrial providers are fucking idiots.
They were faster now they have to compensate for the other sides lack of ability
Well someone just gave away their eyes.
the issue is that no single person or company has the ability to phase IPv4 out
officials in Massachusetts later said that the Oldsmar facility used an unsupported version of Windows with no firewall and shared the same TeamViewer password among its employees.
if there was someone with that control we would have been on ipv6 by now
FUCKING, Brilliant.
Japan disagrees
Rakutan is fully IPv6 only, management infra included
but if people don't have to do something, they won't feel like they have to do something
yes, but they would be risking people abandoning their own platforms for ones that still supported IPv4
uhhh does no one know about the. you know what never mind, this ones mine.
even more reasons to use v6.
Google, Amazon, and Facebook are full dual stack already
and its not worth the risk for them
You will soon enough anyways
Network Acceleration Alliance? There is a cutover agreement already?
Its this year?
Like guys?
Anyone?
no
ok
Its signed under the ONF
Although AT&T is tarded, they wont make it
just start dropping all v4 traffic
Neither will comcast
you'd be surprised how fast people would switch
we are paying a fortune for public IPv4 right now
ouch I feel that
if we actually didn't have to pay that, we would be able to expand our network so much more quickly
faster service for customers, higher packages
can I have a /4 ?