> I'm not at all sure that Shodan is blocked on amprgw. There are
> more than 2,000 IP addresses that are blocked, with more being added
> from time to time, plus there are a number of tcp and udp destination
> ports that are blocked from all IP addresses, but there's no way
> to be sure that these lists include all Shodan and other scanners.
I have blocked some known Shodan addresses and subnets, and indeed even
a hoster that is known to be a cesspool and accomodates Shodan and the likes:
66.240.192.138 # census8.shodan.io
66.240.205.34 # malware-hunter.census.shodan.io
66.240.219.146 # burger.census.shodan.io
66.240.236.119 # census6.shodan.io
71.6.128.0/17 # cesspool! (including shodan.io, project sonar)
80.82.64.0/20 # ECATEL/QUASI (including shodan.io 80.82.77.139)
82.221.105.6 # census10.shodan.io
82.221.105.7 # census11.shodan.io
89.248.160.0/20 # ECATEL/QUASI (incl shodan.io 89.248.167.131 89.248.172.16)
93.174.88.0/21 # ECATEL/QUASI (incl shodan.io 93.174.95.106)
94.102.48.0/20 # ECATEL/QUASI (incl shodan.io 94.102.49.190 94.102.49.193)
107.6.151.192 # security.census.shodan.io
107.6.151.193 # security.census.shodan.io
107.6.151.194 # security.census.shodan.io
107.6.151.195 # security.census.shodan.io
185.163.109.66 # goldfish.census.shodan.io
185.181.102.18 # turtle.census.shodan.io
198.20.69.72/29 # shodan.io
198.20.69.96/29 # shodan.io
198.20.70.112/29 # shodan.io
198.20.87.96/29 # shodan.io
198.20.99.128/29 # shodan.io
(of course many others, these are just the shodan.io entries)
I also have some iptables rules that capture TCP SYN to addresses that are not registered in DNS and
forwards them to an nflog socket to be picked up by some scripts that finds those that are repeat
offenders. Those are logged as candidates for blocking. But I don't bother to block everything,
I run reverse-DNS on them to see if it has some signature patterns like "research", "scan" etc or
one of the known names like shodan.io stretchoid.com etc.
And irregularly I just sort the entire list and glance over it to see if there are clusters of
addresses and do a whois to see if they belong to some common network. Names like DigitalOcean
pop up quite regularly but of course they are just cloud hosters that could also host bonafide
services.
Rob
I run the gateway at 75.101.96.109 and have assigned my network,
44.4.39.8/29 to it in the portal a long time ago (more than a year ago). I
receive packets destined for the first allocated address (44.4.39.8) but
I'm not receiving traffic for any of the other hosts in my range. Did I
miss a crucial update about network allocations?
-- Longer version below:
I've made sure to log into the portal every now and then to keep my
assignment fresh, but just today I decided to actually try to start
connecting hosts inside my allocation.
To do some tests, I tried pinging the first assignment in my net,
44.4.39.8, from an outside host.
That worked fine. I successfully received the tunneled IPIP packet from
amprgw.ucsd.edu, de-encapsulated it, found that the inner destionation was
44.4.39.8, and sent it on its way to my inside network.
When I attempt to hit other hosts in my allocation, however, I never even
receive the IPIP packet for them. Traceroute from my test host seems to
indicate that the packet never even makes it to UCSD; it stops one hop
short, at sdsc-7710-7--mcore-vl2995-p2p.ucsd.edu (132.239.255.50).
WORKING HOST
Traceroute to 44.4.39.8:
traceroute to 44.4.39.8 (44.4.39.8), 64 hops max, 40 byte packets
1 50-0-193-1.dsl.static.fusionbroadband.com (50.0.193.1) 23.641 ms
19.205 ms 18.704 ms
2 xe-0-0-20.cr2.colaca01.sonic.net (70.36.228.117) 39.927 ms 72.901 ms
30.780 ms
3 0.ae0.cr3.colaca01.sonic.net (198.27.244.130) 37.454 ms 35.465 ms
33.255 ms
4 * * *
5 50.ae4.gw.pao1.sonic.net (50.0.2.5) 19.397 ms 19.728 ms 20.012 ms
6 palo-b22-link.telia.net (213.248.81.254) 19.673 ms 19.190 ms 19.935
ms
7 sjo-b21-link.telia.net (62.115.125.1) 21.363 ms 20.989 ms 20.647 ms
8 las-b21-link.telia.net (62.115.116.41) 26.836 ms
las-b21-link.telia.net (62.115.136.46) 37.521 ms 33.204 ms
9 limelight-ic-320599-las-b21.c.telia.net (213.248.96.65) 68.480 ms
28.275 ms 28.070 ms
10 198.32.251.189 (198.32.251.189) 31.314 ms 32.787 ms 28.098 ms
11 137.164.11.25 (137.164.11.25) 33.761 ms 36.236 ms
137.164.11.23 (137.164.11.23) 33.997 ms
12 dc-sdg-agg4--tus-agg3-100ge.cenic.net (137.164.11.9) 40.166 ms
137.164.11.11 (137.164.11.11) 34.933 ms 31.510 ms
13 dc-ucsd-1--sdg-agg4.cenic.net (137.164.23.54) 36.912 ms 36.172 ms
32.772 ms
14 mcore-flow-bypass-mx0-p2p.ucsd.edu (132.239.254.61) 39.436 ms 38.211
ms 42.078 ms
15 sdsc-7710-7--mcore-vl2995-p2p.ucsd.edu (132.239.255.50) 37.436 ms
37.872 ms 36.955 ms
16 amprgw.ucsd.edu (169.228.34.84) 32.790 ms * 33.361 ms
17 ke6jjj-8.ampr.org (44.4.39.8) 35.951 ms 36.368 ms 32.583 ms
NON-WORKING HOST
Traceroute to 44.4.39.9:
traceroute to 44.4.39.9 (44.4.39.9), 64 hops max, 40 byte packets
1 50-0-193-1.dsl.static.fusionbroadband.com (50.0.193.1) 69.583 ms
19.148 ms 19.244 ms
2 xe-0-0-20.cr2.colaca01.sonic.net (70.36.228.117) 120.800 ms 68.925
ms 88.207 ms
3 0.ae0.cr3.colaca01.sonic.net (198.27.244.130) 91.626 ms 34.418 ms
76.632 ms
4 * * *
5 50.ae4.gw.pao1.sonic.net (50.0.2.5) 23.116 ms 19.177 ms 19.248 ms
6 palo-b22-link.telia.net (213.248.81.254) 19.647 ms 19.396 ms 19.452
ms
7 sjo-b21-link.telia.net (62.115.125.1) 20.457 ms
las-b24-link.telia.net (62.115.119.91) 36.542 ms
sjo-b21-link.telia.net (62.115.125.1) 20.644 ms
8 las-b21-link.telia.net (62.115.136.46) 33.996 ms
las-b21-link.telia.net (62.115.116.41) 26.532 ms
las-b21-link.telia.net (62.115.136.46) 33.467 ms
9 limelight-ic-320599-las-b21.c.telia.net (213.248.96.65) 36.186 ms
28.669 ms 28.199 ms
10 198.32.251.189 (198.32.251.189) 30.786 ms 33.225 ms 31.040 ms
11 137.164.11.25 (137.164.11.25) 32.792 ms 33.481 ms 32.755 ms
12 dc-sdg-agg4--tus-agg3-100ge.cenic.net (137.164.11.9) 38.265 ms
137.164.11.11 (137.164.11.11) 31.252 ms 32.276 ms
13 dc-ucsd-1--sdg-agg4.cenic.net (137.164.23.54) 37.461 ms 37.917 ms
36.964 ms
14 mcore-flow-bypass-mx0-p2p.ucsd.edu (132.239.254.61) 36.682 ms 33.464
ms 36.222 ms
15 sdsc-7710-7--mcore-vl2995-p2p.ucsd.edu (132.239.255.50) 33.497 ms
37.393 ms 36.970 ms
16 * * *
17 * * *
18 * * *
Got my allocation online and have reconfigured elproxy to provide 150
public proxies, in addition to the existing 4 private proxies. Got a
geolocation issue causing incorrect allocation of proxies, but have
submitted change requests with 3 of the major geolocation providers. :)
--
73 de Tony VK3JED/VK3IRL
http://vkradio.com
> Sorry, I can't agree on that. Without manual interaction Echolink
> Repeaters behind traditional IPIP-Mesh gateways are not reachable
> through the NLD Echolink Relays/Proxies.
That is NOT CORRECT!
It will work just fine when the repeater is behind a traditional IPIP mesh gateway
which sends traffic to internet via amprgw (the setup described in the Wiki).
It will only fail when shortcuts are made and traffic to internet is sent via some
local NAT router instead of via the IPIP mesh. As the Echolink protocol cannot
handle this situation (as described above) one needs to make sure that Echolink
traffic is not sent via this NAT shortcut.
The Echolink system is perfectly happy with clients, proxies and relays on NET44,
it is only the bad configuration of gateway systems that will break it.
Rob
Description:
The network 44.190.0.0/16 is available to host amateur radio internet
services using direct BGP. Allocations will be made usually in /24
chunks. The focus of this network is to transfer data as direct as
possible (no radio, no tunnels) to keep reliability high. AMPRNet
systems with no dynamic routing protocol support are able to route
44.190.0.0/16 via their ISP.
Background:
AMPRNet allocations are not partitioned by connectivity type. Each
individual network can either be of type "radio", "tunnel" or "direct".
AMPRNet systems may be "dual homed" and have a line based and a radio
connection, thus there needs to be a decision whether a packet should be
forwarded by line or by radio.
This _could_ be achieved by AMPRNet systems *with* dynamic routing
protocols by learning the routes and treating each route by its type (or
any other available "flag"). However this will _never_ be the case for
AMPRNet systems *without* support for dynamic routing protocols.
Endusers running a *radio* connection to AMPRNet infrastructure such as
HAMNET User Acess Points, IPIP-Mesh Gateways or even direct-BGP Gateways
might ask how to integrate the radio connection into their home-LAN to
have permanent access using any device on their LAN. Unfortunately
standard home routers do not provide dynamic routing protocols, but
there is often the functionality to add additional static routes like
"44.0.0.0/8 via radio" and "44.190.0.0/16 via ISP". Sometimes customers
are *forced* to use the router provided by the ISP to access the
Internet (such a requirement is prohibited by law in Germany). Those
might have no functionality to set static routes at all and you might
end up attaching another router to the router to gain more functionality...
As of today the standard setup for "dual homed" AMPRNet systems
*without* support for dynamic routing protocols is to add 44.0.0.0
netmask 255.0.0.0 via <radio device within the home-LAN>. Outbound
connections to 44.x.x.x will be made by their radio with the source-44
address and outbound connections to the Internet will be made with the
source address obtained from the ISP.
This worked without any issues as long as there were no direct BGP
allocations within the network44. Nowadays we have several amateur radio
internet services provided on the Network44 by direct BGP and the
accessibility for the above mentioned AMPRNet systems highly depends on
a proper radio connection and the backbone infrastructure to transfer
the packet to its destination.
The idea of 44.190.0.0/16 is to give willing stakeholders the
opportunity to improve the situation. Amateur radio internet services
can be hosted in that range while AMPRNet systems can make use of the
additional route 44.190.0.0/16 via their ISP.
The following issue has much more impact:
Once direct BGP allocations started within the Network44 we encountered
situations of full incompatibility. It is the case for amateur radio
internet services (e.g. Echolink) working with a directory to map
callsigns to IP addresses. Echolink does learn the IP address of an
Echolink station (respectively of the used Echolink Proxy or Relay) from
the inbound connection to the directory service.
This leads to the problem, that AMPRNet systems forwarding 44.0.0.0/8 to
the radio device on the roof will not have a slow or weak connection to
a Echolink station using direct BGP but even *no* connection at all.
Since they are registered on the Echolink network with their IP address
obtained from the ISP but appearing with their Network44 address used
over radio at the Echolink station using direct BGP, the connection is
dropped due to a mismatch between "IP address <-> callsign".
This will not happen if Echolink Relay/Proxy Servers or Echolink
Stations will make use of address space from 44.190.0.0/16 while AMPRNet
systems route 44.190.0.0/16 via their ISP. There will be no IP mismatch
anymore.
I made a diagram explaining the situation:
http://db0fhn.efi.fh-nuernberg.de/~dg8ngn/echolink-amprnet.pdf
73,
Jann
--
Jann Traschewski, Faber-Castell-Str. 9, D-90522 Oberasbach, Germany
Tel.: +49-911-99946898, Mobile: +49-170-1045937, E-Mail: jann(a)gmx.de
Ham: DG8NGN / DB0VOX / DB0FOX / DB0ZM / DB0DBA / DB0HZS
For those who use my Amprnet DNS management web tool, just a note to let
you know it's been completely rewritten. Your login credentials will
still work, and it still forces SSL/TLS, but now reCaptcha has been
added. I tried to keep the look and feel of it somewhat similar to the
old one. Don't be shocked when you see a captcha box on the screen now.
--
Did you know that dolphins are so smart that within a few weeks
of captivity they can train people to stand on the very edge of
the pool and throw them fish?
-----
73 de Brian N1URO
IPv6 Certified
SMTP: n1uro-at-n1uro.ampr.org
IPv4: 44.88.0/28
IPv6:2001:470:8a1e::/48
https://uronode.n1uro.com (dual stacked)
https://n1uro.ampr.org (dual stacked)
> Each 44.190.x.0/24 subnet arranges its own BGP advertising,
> so there isn't just one point. They are spread all over the
> world.
> - Brian
Note that due to this, the approximate location in IP geolocation databases has
to be set for each of the /24 subnets. The default location for 44.0.0.0/8 is
San Diego, California, USA. I have set the location for 44.137.0.0/16 to Amsterdam,
Netherlands and other country ip coordinators have done similar for their countries.
And of course an individual amateur can set a more accurate location for their smaller
subnet.
It comes into play for some services that use IP location aware DNS to direct users
to a geographically closest (and hopefully this translates to closest in network topology)
service. With all the 44.190.0.0/16 networks located in San Diego this of course isn't
going to work. Echolink is such a service that users IP geolocation.
You can enter an address from your subnet in lookup services like this:
https://www.iplocation.net/
It shows what some of the more important services return for your location. And when
clicking on the link for each service it is usually easy to submit location data for
a subnet to that service provider.
Rob
> One thing I'm unclear on with Mikrotik is their different generations of
> hardware. I DO want to make sure I get the newer generation so, in
> theory, I get the longer supported OS support. Does anyone know if the
> CCR1009-7G-1C-1Splus is a new generation of hardware or is it older?
> Regarding buying a box that can run "The Dude" on an internal SSD, do
> Wifi, or other stuff on the side. That was something I was planning on
> running on a separate machine with say TICK, Zabbix, etc. Not sure but
> I seriously worry if something goes south in that system, can it harm
> the router. That's NOT acceptable in my book.
The CCR1009-7G-1C-1Splus is the latest model in the CCR series, but the RB1100AHx4
is a newer router that could well be the root of more new models in the future.
Of course one never knows such things for sure, but the Tilera TILE processors in
the CCR series appear to have less current development than the processor used in the 1100.
I would expect MikroTik to support the CCR in software updates at least for some years.
Remember that all MikroTik routers include software support for the support lifetime
of the device, no need for a separate support contract for that.
The selection also depends on the network architecture. The RB1100AHx4 has two hardware switch
chips each driving 5 ports so you can have hardware switching between those ports, while
the CCR series ports are all independent router ports, they are expected to be connected
to an external switch when you want hardware switching. So the CCR is a true router
while the RB1100AHx4 is a router-switch combination. And the CCR has an SFP and SFP+
slot for fiber etc.
The CCR does not have internal M.2 SSD like the RB1100AHx4 Dude edition, but it has
a slot for an SDHC card and a USB port that both can be used for external storage.
Rob
Hello Everyone,
Considering there is a good chunk of routing-savvy HAMs here, I thought
I'd use you as a sounding board on what would be a good router to buy.
Specifically, I have a project to consolidate the current adhoc setup of
three consumer grade "routers" to one larger, better router. I'm
considering something like a:
https://mikrotik.com/product/CCR1009-7G-1C-1Splus
<https://mikrotik.com/product/CCR1009-7G-1C-1Splus>
or maybe https://mikrotik.com/product/rb1100ahx4
<https://mikrotik.com/product/rb1100ahx4>
I'm looking for something that is:
- very stable
- offer long term software updates (a support contract might be fine)
- Has strong support for IPv4 NAT (to better the consumer routers
mentioned above) for the three IPs we have onsite
- maybe some L2 segmenting and vlan'ing support for traffic isolation
- has performance to grow into
- has a decent GUI UI for others in the club who can't / won't cope
with a CLI
- ACLs to limit incoming traffic to specific hosts (say limit RDP
traffic to just some people to some hosts, etc)
- maybe.. just maybe support SSL VPNs or IPSEC
- maybe dual power supplies
- stretch goal: native support for IPv6
- I have no need for dynamic routing protocols. This is a single
site and statics are fine
For background on our needs, the site supports a multi-RF link repeater
system has:
- two unique IRLP nodes (low use)
- one Echolink node (low use)
- one WIresX enabled Yaesu System Fusion repeater (decent use)
- One three band Icom Dstar stack (1.2Ghz DD system as well) (decent
use)
- One Internet enabled Motorola DMR repeater (decent use)
- backhaul of rarely used multi-county 3.4Ghz wifi network
- other random needs for remote management (SSH, RDP, etc)
I believe something like a Miktrotik would be fine for our low-end needs
but maybe something from Ubiquiti or others would be fine. I'm perfectly
comfortable with a CLI and I'm decently versed in Mikrotik (a bit weird
of a UI), IOS (but I don't want to pay for Cisco prices, JUNOS (same
point), etc. I personally think a lot of the lower tier vendor's
products have come a LONG way so I don't need/want/care for "carrier" grade.
If you have any other recommendations for a quality but not too
expensive router, I'd love to hear it!
--David
KI6ZHD