I'm not sure I made this clear: with the change from my homegrown
format to PCAP, the file names changed from <address>.bin to
<address>.pcap. So you'd fetch
https://gw.ampr.org/private/errors/62.147.189.164.pcap
for example.
- Brian
Hi,
I added a gateway to one of my servers and set up a packet capture
system on it. It is running a webserver inside a container with the
relevant gateway configuration. The prefix is using the "BGP routed
subnets" configuration whereby the gateway is inside the prefix, but
using a /31.
https://u4477715.ct.sendgrid.net/wf/click?upn=Ki4chJONuNfM0VomxEE-2BoZH6yGO…
It should auto-detect the client IP and captures all packets to/from
that address. If the client IP is within 44/8 then it will check the
routing table for a gateway and if so include that IP as well. I am
not sure if there would be concerns with allowing the user to type an
IP address to capture to/from, given that it's a non-production
gateway.
I have the subnet 44.131.14.252/31 registered on the portal with a
gateway address of 44.131.14.253. 252 should send encapsulated packets
and 253 should send directly. Both addresses are on the same host.
I have removed my previous route for 44.131.14.0/24 because nested
gateways don't work properly. I have tested to several destinations
and it seems to work, but if anyone finds something I've missed let me
know!
If it works properly and is useful then a hostname under ampr.org
might be more appropriate, but for now I’m just using a hostname under
my domain.
Thanks,
Mike, M6XCV
Hi there
Is there a simple way to know when new encap file created ? beside downloading the file itself and looking on the file header ?
My ISP replace my IP an hour ago and i still dont have connection ..
in the portal there is no time stamp when looking on the gateways list when the new update made ....
Im using DYNDNS method and the portal still show my old IP of course that the DYNDNS new ip is already updated
Of course i can edit it by hand in the portal but that is not the intention
Regards
Ronen - 4Z4ZQ
With all the probes going on to various AMPRNet hosts, it might
be wise for you to configure your hosts to NOT reply when an attempt
is made to access a port on which nothing is listening. Ordinarily,
a host system will reply with a TCP reset or a UDP port unreachable,
and this contributes to the outbound traffic from your host. There
are probably 'blackhole' options you can set. On FreeBSD, they are
sysctl net.inet.tcp.blackhole=2
sysctl net.inet.udp.blackhole=1
I don't know if Linux has similar options. I rather assume it does.
Note that the latter option may break traceroutes to your system.
It's probably also wise to limit the number of replies to pings so
that rabid pingers won't pollute your outbound connection.
sysctl net.inet.icmp.icmplim=5
Which limits replies to 5 per second, down from 1000 default.
In the Linux kernel, there is what seems to be a similar option:
sysctl net.ipv4.icmp_ratelimit=5
Perhaps someone can confirm that this is correct.
- Brian
After fighting with my system off and on for months I'm finally on
AMPRNET but I'm not sure If I'm on the mesh yet. I can get to google
and it looks like anywhere on the Internet and can get to my cox email
using the thunderbird email client from my workstation.
I think I had problems in the past because I had a pre-existing strong
iptables firewall and I tried to adapt the Linux configuration on the
wiki to the existing strong firewall. Yesterday I decided to start from
scratch and build using the Linux gateway instructions on the wiki. I
had it working but it seems it failed after I restarted the gateway, I
had no connectivity from my gateway to any of my local systems after the
reboot. Today I rebuilt from scratch again and it seems to be working
except it seems I can't get to anything in the 44/8 network except my
own IP block.
I've installed the latest ampr_ripd available as of today but need to
know how to tell if it's adding routes into the routing table.
My current setup is a Linux router, three Raspberry Pi and a Linux
Desktop that serves as my workstation (Yesterday it was a headless
Raspberry Pi).
Tests I've done:
1. A query on Google for "What's my IP address". I got back 44.98.63.3
(my workstation) proving I'm going through the AMPR gateway. When I
attempt to connect to some of the services linked from the wiki such as
http://n1uro.ampr.org/do.shtml and http://whatismyip.ampr.org I don't
get responses back. So my first question is how do I test to see if I
have mesh routing up to the rest of the 44Net?
2. I need to learn how to set up iptables to only accept ipencap packets
from AMPR gateways. I suspect it requires using ipset which I've used
in the past for dropping traffic from systems trying to crack into my
router which leads me to my second question. Is there anyone out there
willing to show sample code how to allow ipencap traffic only from AMPR
gateways?
3. Last night before I restarted and lost ability to use my workstation
(remotely, I have not figured out why it failed yet) I was able to log
on to my VPS at Linode from my ISP provided space and then SSH into my
workstation on my 44/8 address through the tunnel... At the time the
workstation was the headless Raspberry Pi, this worked perfectly. I also
notice my google traffic uses HTTPS and my email client is using port
465 and 993... all of these are encrypted. My third question: I know
they aren't allowed over the air, so how do we account for/deal with
software that insists on using encrypted protocols? Is SSH allowed for
remotely maintaining our nodes?
4. A couple of weeks ago, I ordered and received the parts to build a
Stratum 1 Time server that I intend to make publicly available to the
44Net as a service to the 44Net community. Once I get it online and the
security in place to prevent nefarious activity, How do I announce it?
I intend to have a web server, a time server, a DNS server and an NNTP
server online as well as a local packet node and a tunnel to the AREDN
and possibly a link to the HAMWAN out of Tampa if they'll allow me to
link to them. I'm open to suggestions on proper operating procedure.
--
Tom Cardinal/MSgt USAF (Ret)/BSCS/CASP, Security+ ce
Hi,
This might be an academic question in practice, but what is the
correct configuration for sending towards a BGP routed subnet, or for
2 subnets communicating with each other?
44.131.14.252 is behind 44.131.14.253
44.130.121.1 is behind 44.130.121.2
This is the current behaviour. For the purposes of testing, packets
sourced from 252 were encapsulated and 253 were unencapsulated.
https://u4477715.ct.sendgrid.net/wf/click?upn=MJaTQVDJZogYIZySndf7y-2BCWLgZ…
.1 sends directly to .252 and .253.
.2 sends encapsulated to .252 and directly to .253.
I would say packets to 1 or 252 should be encapsulated. I don't know
if there is any point adding an extra header to something addressed to
the gateway.
Thanks,
Mike, M6XCV
I'm been seeing rather a storm of incoming packets to the amprgw gateway;
there are six times as many inbound packets to routable AMPRNet addresses
as outbound. The firewall is doing its duty: in the last about 24 hours
it's discarded 9,905,154,331 incoming packets as being from bad guys,
and of the packets that did get past the badguy list, 24,974,234,978
were discarded. It's running around 30 MB/s, at about 20 million packets
a minute.
Looking at the incoming traffic, it appears to be mostly TCP connection
requests with large window sizes to varying 44.x.x.x addresses, with lots
of different destination port numbers but many are to port 23 (TELNET)
and port 80 (HTTP). There's a goodly number of UDP requests to port 53
(DNS) too.
The system seems to be about 85% idle.
- Brian
I just found and fixed a bug in the router where it would occasionally
reject a valid destination as being an encap-to-encap route and discard
the packet.
Detail: Rob's address-mask-and-index scheme with the huge global
addresses array only works if you FIRST check that the address
is on network 44. Oops.
Anyway, I restarted the packet errors tallys so they won't show the
erroneous results of this bug.
- Brian
I have also noticed that traceroutes through the amprgw router
don't seem to work, although the destination host is reachable.
I have yet to spend much time on figuring out why this is.
- Brian
On Fri, May 05, 2017 at 11:35:40PM +0100, M6XCV (Mike) wrote:
> I think this is because you are trying to send via the gateway.
>
> I am not sure why, but I have tried using the gateway to access the
> internet and my packets go missing. I assumed it should work, however the
> fact that it did not was not something I thought it was worth investigating
> as I am BGP routed. If are trying to reach me through the main gateway then
> perhaps it is worth asking Brian to look in to it?
>
> For the record, I attempted to ping you and my outbound seems fine.
> https://u4477715.ct.sendgrid.net/wf/click?upn=MJaTQVDJZogYIZySndf7y-2BCWLgZ…
> 668edee920.pcap
>
> I can ping 44.0.0.1. Traceroutes also show my packets reach the gateway but
> go no further. I added a static route for 8.8.8.8 encapped via the gateway
> and it gave me this.
>
> traceroute to 8.8.8.8 (8.8.8.8), 30 hops max, 60 byte packets
> 1 onsite.notmike.uk (44.131.14.129) 0.412 ms 0.392 ms 0.382 ms
> 2 Gateway.AS206671 (44.131.14.254) 9.972 ms 9.989 ms 9.988 ms
> 3 amprgw.sysnet.ucsd.edu (169.228.66.251) 158.942 ms 158.981 ms
> 159.397 ms
> 4 * * *
> 5 * * *
>
> I suspect a firewall rule, however there are A records under ampr.org for
> all of my IP addresses.
>
> Thanks,
> Mike, M6XCV
Hello All,
I've just implemented the newest version of Marius' Mikrotik script
which enables accessing 44net IPs using a gateway in 44 address space,
and was wondering if there is an IP which uses this configuration I can
test my setup against. The network which Marius was originally tested
with (44.130.120.0/24) seems to no longer be present in my routing table.
Cheers!
Chris
VE7ALB
