Tag: gentoo

NiFi Build Error

I’m testing NiFi out on my local Gentoo installation to prepare for an implementation at work, and after a rather lengthy build/test process (“ten minutes” my fanny) ran into this error:

$ mvn clean install
[INFO] Scanning for projects...
...
'Script Engine' validated against 'ECMAScript' is invalid because Given value not found in allowed set 'Groovy, lua, python, ruby'

This error left me scratching my head.  Nothing related to JavaScript/ECMAScript dependencies were mentioned anywhere.  How would you get it, anyway?  Webkit, I suppose…

Sudden epiphany: this is a new Gentoo installation, and this program, including the build script, is running Java.  Gentoo doesn’t install Sun Oracle’s Java by default, but instead comes with IcedTea out of the box.  It’s acceptable for some simple uses, but is buggy for any complex. (Minecraft is a great example where it just doesn’t work.)  I haven’t used Java for anything yet, so I hadn’t installed the JDK yet.  The build instructions specify JDK 1.7 or higher, but I didn’t think anything of it because I’m used to just having it installed.

echo "dev-java/oracle-jdk-bin Oracle-BCLA-JavaSE" \
  >> /etc/portage/package.license/file
emerge -av dev-java/oracle-jdk-bin
...
$ mvn clean install
[INFO] Scanning for projects...
...
[INFO] BUILD SUCCESS

Finally!

Bridging Wired and Wireless Networks, Gentoo-style

I want my wired and wireless networks to share a single 192.168.1.x address space (instead of separate 192.168.0.x and 192.168.1.x addresses).

In order to do that, we need to set up a bridge to merge disparate networks into a single space.

Part 1: The Basic Configuration

ADMtek NC100 (uses tulip driver)
Ralink RT61 PCI (uses rt61pci driver)
hostapd
linux 4.1.15-gentoo-r1
net-misc/bridge-utils 1.5
net-wireless/iw 3.17

Part 2: Making It Work

I started out creating a basic bridge, using the Gentoo Wiki as a guide:

cd /etc/init.d
ln -s net.lo net.br0

/etc/init.d/net.br0 start

There’s no need to change how hostapd starts; it still talks to wlan0 (not br0).

# /etc/conf.d/net

modules_wlan0="!iwconfig !wpa_supplicant"
config_wlan0="null"
config_eth0="null"
config_br0="192.168.1.1/24"
brctl_br0="setfd 0
sethello 10
stp off"
bridge_br0="eth0 wlan0"

The Problem

The above config is naive and doesn’t work right.  I got this error:

Can't add wlan0 to bridge br0: Operation not supported

Huh.  There’s nothing indicative in dmesg about the error, the last entry shows the bridge being created on the wired card and then being taken down.  Just to be sure, I created a bridge with just eth0 and it worked:

$ brctl show
bridge name   bridge id           STP enabled   interfaces
br0           8000.00045a42a698   no            eth0

After casting about a bit, I found a serverfault.com page that pointed to this fix:

$ iw dev wlan0 set 4addr on
$ brctl addif br0 wlan0

That works, but that won’t do me much good as a long-term solution.  I would need to pay a visit to the basement after every planned reboot and unplanned power outage, or else nobody can get onto the network.

( More about the 4addr option here. )

You can’t just add the option to modules_wlan0, it doesn’t work that way.  A quick visit back to the wiki suggested the solution, though, which is to define a preup function where we can execute arbitrary commands.

The Working Config

These statements are in addition to the WAN interface config:

# /etc/conf.d/net
modules_wlan0="!iwconfig !wpa_supplicant"
config_wlan0="null"
config_eth0="null"
config_br0="192.168.1.1/24"
brctl_br0="setfd 0
sethello 10
stp off"
bridge_br0="eth0 wlan0"

preup() {
    # br0 uses wlan0, and wlan0 needs to set the
    # 4addr option before being used on a bridge
    if echo "${IFACE}" | grep -q 'br0' ; then
        /usr/sbin/iw dev wlan0 set 4addr on
    fi

    return 0
}

Then do all the accounting to clean up:

rc-update add net.br0 default
rc-update del net.eth0 default
rc-update del net.wlan0 default

I also had to update my iptables config to refer to br0 instead of eth0 and wlan0.

Finally, a reboot to test that everything starts properly.

Setting up a Gentoo-Based Dual-Stack Router

Our network has been based around a home-built router for quite some time, ever since I got fed up with the crappy ActionTec router that Verizon bundled with our FiOS service. (If you’re going to offer high-speed internet, you should probably bundle equipment that can actually keep up.) I had originally followed a slightly older version of these instructions to get a nice basic router going. But I finally wanted better. I wanted the bright, shiny, new thing. I wanted IPv6.

So, here’s my instructions for going from an existing IPv4 router to dual-stack IPv4/6.

Note: I am using dnsmasq for DNS and DHCP, hostapd for wireless management, and an iptables firewall. Since Verizon still doesn’t widely support consumer IPv6, I’m using a tunnel broker to get my /6 address. If you’re using a different setup your mileage may vary. If you find anything that I appear to have forgotten, please let me know!

Step 1: Recompile the Kernel

This should be obvious: if you want to run ipv6 you need ipv6 support in your kernel. In order to trim as much off my kernel as possible I did not have it built in, and had to recompile.

You should also add netfilter support for ipv6 so that your firewall will work.

kernel changes
1
2
3
4
5
6
7
8
9
10
11
12
Networking support  --->
    Networking options  --->
        <*>   The IPv6 protocol  --->
            <*>   IPv6: IPv6-in-IPv4 tunnel (SIT driver)
        [*] Network packet filtering framework (Netfilter)  --->
            IPv6: Netfilter Configuration  --->
                <M> IPv6 NAT
                <M> IP6 tables support (required for filtering)
                <M>   Packet filtering
                <M>   ip6tables NAT support
                <M>     MASQUERADE target support
                ... other filtering options as you may need for your situation

Step 2: Update Your IPv6 Support

Again, it was never compiled in, in order to trim off unused bits of code. Add ‘ipv6’ to your USE variable and emerge --newuse world

Step 3: Install network tools (if they aren’t already)

emerge --noreplace sys-apps/iproute2 net-firewall/iptables

Step 4 (optional): Set up your tunnel

If your ISP doesn’t provide ipv6, and many don’t, you need to request an address range from a tunnel broker. I’m using Hurricane Electric, which is free, but there are others — see this list or just google it.

If you have multiple machines on your network (which is assumed, since this is a router guide), you may prefer a /48, so that autoconfig works nicely, instead of the default /64. This guide assumes a /48.

Going forward, replace 2001:470:891a: with your own /48 range.

Now activate your tunnel:

1
2
3
4
5
ip tunnel add he-ipv6 mode sit remote 1.2.3.4 local 5.6.7.8 ttl 255
ip link set he-ipv6 up
ip addr add 2001:470:1f06:2a3::2/64 dev he-ipv6
ip route add ::/0 dev he-ipv6
ip -f inet6 addr

Step 5: Update Your Net Config

I have two wired and one wireless card in my router. Here’s what my /etc/conf.d/net looks like:

1
2
3
4
5
6
7
8
9
10
11
# enp2s0 is my exterior wired nic (aka public facing)
# enp3s5 is my interior wired nic
# wlp3s6 is my interior wireless nic
 
dhcp_enp2s0="nodns" # we choose our own DNS, tyvm
 
config_enp3s5="192.168.0.1/24 2001:470:891a:0::/64"
 
modules_wlp3s6="!iwconfig !wpa_supplicant"
config_wlp3s6="192.168.1.1/24 2001:470:891a:1::/48"
dns_servers_wlp3s6="127.0.0.1"

After making appropriate changes, restart your NICs. If you’re working remotely, you may want to be connected via two paths instead of just one (so when you inevitably get bounced and can’t reconnect, you still have a way back in).

A properly-configured set of addresses looks like this:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
# ip addr
1: lo: &lt;LOOPBACK,UP,LOWER_UP&gt; mtu 65536 qdisc noqueue state UNKNOWN group default
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 brd 127.255.255.255 scope host lo
       valid_lft forever preferred_lft forever
    inet6 ::1/128 scope host
       valid_lft forever preferred_lft forever
2: enp2s0: &lt;BROADCAST,MULTICAST,UP,LOWER_UP&gt; mtu 1500 qdisc pfifo_fast state UP group default qlen 1000
    link/ether 00:e0:4d:bf:03:f5 brd ff:ff:ff:ff:ff:ff
    inet 108.20.118.17/24 brd 108.20.118.255 scope global enp2s0
       valid_lft forever preferred_lft forever
    inet6 fe80::cbdf:25c0:c948:f4bb/64 scope link
       valid_lft forever preferred_lft forever
3: enp3s5: &lt;BROADCAST,MULTICAST,UP,LOWER_UP&gt; mtu 1500 qdisc pfifo_fast state UNKNOWN group default qlen 1000
    link/ether 00:04:5a:42:a6:98 brd ff:ff:ff:ff:ff:ff
    inet 192.168.0.1/24 brd 192.168.0.255 scope global enp3s5
       valid_lft forever preferred_lft forever
    inet6 2001:470:891a::/64 scope global
       valid_lft forever preferred_lft forever
    inet6 fe80::204:5aff:fe42:a698/64 scope link
       valid_lft forever preferred_lft forever
4: wlp3s6: &lt;BROADCAST,MULTICAST,UP,LOWER_UP&gt; mtu 1500 qdisc mq state UP group default qlen 1000
    link/ether 00:1a:ef:07:4d:a7 brd ff:ff:ff:ff:ff:ff
    inet 192.168.1.1/24 brd 192.168.1.255 scope global wlp3s6
       valid_lft forever preferred_lft forever
    inet6 2001:470:891a:1::/64 scope global
       valid_lft forever preferred_lft forever
    inet6 fe80::21a:efff:fe07:4da7/64 scope link
       valid_lft forever preferred_lft forever
5: sit0@NONE:  mtu 1480 qdisc noop state DOWN group default
    link/sit 0.0.0.0 brd 0.0.0.0
6: he-ipv6@NONE: &lt;POINTOPOINT,NOARP,UP,LOWER_UP&gt; mtu 1480 qdisc noqueue state UNKNOWN group default
    link/sit 108.20.118.17 peer 209.51.161.14
    inet6 2001:470:1f06:2a3::2/64 scope global
       valid_lft forever preferred_lft forever
    inet6 fe80::6c14:7611/64 scope link
       valid_lft forever preferred_lft forever

Test it with a ping:

ping6 www.kame.net

Step 6: Reconfigure dnsmasq

You’ll need to add router advertisments and your new addresses. Rather than hard-coding an address, dnsmasq offers a ‘constructor’ label which figures it out automatically. Here’s the relevant section from my /etc/dnsmasq.conf:

1
2
3
4
5
6
7
8
9
10
11
12
13
domain-needed
bogus-priv
domain=jonesling.us
dhcp-authoritative
enable-ra
dhcp-range=192.168.0.20,192.168.0.100,12h
dhcp-range=192.168.1.20,192.168.1.100,12h
dhcp-range=192.168.2.20,192.168.2.100,12h
dhcp-range=::,constructor:enp3s5,ra-names,slaac,12h
dhcp-range=::,constructor:wlp3s6,ra-names,slaac,12h
resolv-file=/etc/resolv.dnsmasq
selfmx
enable-ra

And restart it: /etc/init.d/dnsmasq restart

Step 7: Configure your firewall

Pretty much every iptables reference in your firewall config will be mirrored with an ip6tables command.

Here’s my script to set up iptables (if you see an error or something stupid, I would appreciate your criticism – paired with your reasoning on why it should be changed so I can know better for next time).

iptables.sh
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
#!/bin/bash
# based on http://www.gentoo.org/doc/en/home-router-howto.xml
 
# set to '0' to lock the kids out
OPEN_INTERNET=1
 
# these systems can get shut out when OPEN_INTERNET isn't true
declare -a NO_SURFING=( 'wii-u'
                        'kids-computer'
                      )
 
# these systems never get shut out
declare -a OK_SURFING=( 'parents-computer'
                        'parents-phone'
                      )
 
# these ports take precedence over CLOSED_PORTS
declare -a OPEN_TCP_PORTS=( 'ssh'
                            'http'
                            'mail'
                            'submission'
                          )
 
declare -a OPEN_UDP_PORTS=( 'submission' )
 
# if the port is meant to be closed, we close tcp *AND* udp
declare -a CLOSED_PORTS=( '0:1055'
                          'svn'
                          'distcc'
                          'x11'
                          'nfs'
                          'icpv2'
                          'mysql'
                          'rtsp'
                          '3128' # squid
                          '3130' # squid ICP
                          '3551' # nisport
                        )
 
declare -a LAN_SERVICES=( "svn" )
 
# blacklisted IPs and ranges
# http://www.iana.org/assignments/ipv4-address-space/ipv4-address-space.xhtml
declare -a IP_BLACKLIST=( # APINIC
                          # AFRINIC
                          # LACNIC
                          ...
                        )
 
LAN=enp3s5
WLAN=wlp3s6
WAN=enp2s0
SIT=he-ipv6
 
INSIDE=( $LAN $WLAN )
 
LOCAL_RANGE_IPV4='192.168.0.0/16'
LOCAL_RANGE_IPV6='2001:470:891a::'
 
# First we flush our current rules
iptables -F
iptables -t nat -F
ip6tables -F
ip6tables -t nat -F
 
# Setup default policies to handle unmatched traffic
iptables  -P INPUT ACCEPT
iptables  -P OUTPUT ACCEPT
iptables  -P FORWARD DROP
ip6tables -P INPUT ACCEPT
ip6tables -P OUTPUT ACCEPT
ip6tables -P FORWARD DROP
 
# Then we lock our services so they only work from the LAN
iptables  -I INPUT 1 -i ${LAN}  -j ACCEPT
iptables  -I INPUT 1 -i ${WLAN} -j ACCEPT
iptables  -I INPUT 1 -i lo      -j ACCEPT
ip6tables -I INPUT 1 -i ${LAN}  -j ACCEPT
ip6tables -I INPUT 1 -i ${WLAN} -j ACCEPT
ip6tables -I INPUT 1 -i lo      -j ACCEPT
 
# block members of IP_BLACKLIST, plus any addresses passed in on the
# command line
for IP in ${IP_BLACKLIST[@]} ; do
    iptables -I INPUT -s ${IP} -p TCP --dport ssh -j DROP
done
 
for IP in $@; do
    iptables -I INPUT -s ${IP} -d 0/0 -j REJECT
done
 
iptables  -A INPUT -p UDP --dport bootps -i ${WAN} -j REJECT
ip6tables -A INPUT -p UDP --dport bootps -i ${SIT} -j REJECT
iptables  -A INPUT -p UDP --dport domain -i ${WAN} -j REJECT
ip6tables -A INPUT -p UDP --dport domain -i ${SIT} -j REJECT
 
# Explicitly allow access to services on the WAN
for SERVICE in ${LAN_SERVICES[@]} ; do
    for IFACE in ${INSIDE[@]} ; do
        iptables  -A INPUT -p TCP --dport svn -i ${IFACE} -j ACCEPT
        iptables  -A INPUT -p UDP --dport svn -i ${IFACE} -j ACCEPT
        ip6tables -A INPUT -p TCP --dport svn -i ${IFACE} -j ACCEPT
        ip6tables -A INPUT -p UDP --dport svn -i ${IFACE} -j ACCEPT
    done
done
 
# Allow access to our server from the WAN
for PORT in ${OPEN_TCP_PORTS[@]} ; do
    iptables  -A INPUT -p TCP --dport $PORT -i ${WAN} -j ACCEPT
    ip6tables -A INPUT -p TCP --dport $PORT -i ${SIT} -j ACCEPT
done
 
for PORT in ${OPEN_UPD_PORTS[@]} ; do
    iptables  -A INPUT -p UDP --dport PORT -i ${WAN} -j ACCEPT
    ip6tables -A INPUT -p UDP --dport PORT -i ${SIT} -j ACCEPT
done
 
# Drop TCP / UDP packets to privileged ports
for PORT in ${CLOSED_PORTS[@]} ; do
    iptables  -A INPUT -p TCP -i ${WAN} -d 0/0 --dport ${PORT} -j DROP
    ip6tables -A INPUT -p TCP -i ${SIT} -d 0/0 --dport ${PORT} -j DROP
 
    iptables  -A INPUT -p UDP -i ${WAN} -d 0/0 --dport ${PORT} -j DROP
    ip6tables -A INPUT -p UDP -i ${SIT} -d 0/0 --dport ${PORT} -j DROP
done
 
iptables  -I FORWARD -i ${LAN} -d $LOCAL_RANGE_IPV4 -j ACCEPT
iptables  -A FORWARD -i ${LAN} -s $LOCAL_RANGE_IPV4 -j ACCEPT
ip6tables -I FORWARD -i ${LAN} -d $LOCAL_RANGE_IPV6 -j ACCEPT
ip6tables -A FORWARD -i ${LAN} -s $LOCAL_RANGE_IPV6 -j ACCEPT
 
if (( OPEN_INTERNET )); then
    echo 'yay, everybody gets internet'
    iptables  -I FORWARD -i ${WLAN} -d $LOCAL_RANGE_IPV4 -j ACCEPT
    iptables  -A FORWARD -i ${WLAN} -s $LOCAL_RANGE_IPV4 -j ACCEPT
    ip6tables -I FORWARD -i ${WLAN} -d $LOCAL_RANGE_IPV6 -j ACCEPT
    ip6tables -A FORWARD -i ${WLAN} -s $LOCAL_RANGE_IPV6 -j ACCEPT
else
    echo "boo, only ${OK_SURFING[@]} get internet"
    for IP in ${OK_SURFING[@]}; do
        iptables -I FORWARD -i ${WLAN} -d $IP/255.255.255.255 -j ACCEPT
        iptables -A FORWARD -i ${WLAN} -s $IP/255.255.255.255 -j ACCEPT
    done
fi
 
iptables  -A FORWARD -i ${WAN} -d $LOCAL_RANGE_IPV4 -j ACCEPT
ip6tables -A FORWARD -i ${WAN} -d $LOCAL_RANGE_IPV6 -j ACCEPT
 
iptables  -t nat -A POSTROUTING -o ${WAN} -j MASQUERADE
ip6tables -t nat -A POSTROUTING -o ${SIT} -j MASQUERADE
 
# This is so when we boot we don't have to run the rules by hand
/etc/init.d/iptables save
/etc/init.d/ip6tables save
 
# fail2ban should be reloaded after flushing iptables
/etc/init.d/fail2ban reload

Step 8: Update your DNS

Add a AAAA record to your domain’s DNS record.  You may have to keep this one up-to-date yourself.

Interesting to note: you might be thinking “crap, what’s the ipv6 equivalent of these CNAME records?”  Stop worrying, there isn’t.  The CNAME is read like normal, but ipv6 clients will then look up the AAAA (instead of the A) record of the destination host.  It just works.

What?  You built your own router but you don’t have your own domain?  WTF is wrong with you?

Step 9: Reboot your clients

While I was working, I made a bunch of mistakes and my clients had multiple ipv6 addresses – making networking from them unstable as they didn’t necessarily know which address to use. Rebooting will clear them – and make sure your config is proper.

At this point your clients should be in ipv6 and you’re gonna be all excited to see if work.  Browsers take ipv6 addresses a little differently: http://[2001:470:1f06:2a3::2]/

Airprint woes

It’s all my fault, really.  This wouldn’t have been an issue if I had just let Xtina use my computer to print her boarding pass, but in my defense I didn’t know that she was doing that.  So I gave her our iPad to use.

When it came time to print, she quite logically asked me how she would do that.  I, of course, did not know how — I’ve never tried printing from iPad or smartphone, though I vaguely knew it was possible.  The issue just never came up and I hate printers.

I knew that it would require avahi, so I started installing that on our printserver while I hit Google to see what else I would need.

The first hit was a very fine article by Linux Magazine, and it explained pretty much everything.  But it’s never that simple, because nothing printed and cups started using 100% of a CPU.

Repeated in the /var/log/cups/error_log a billion times were messages like these:

Request from "192.168.1.32" using invalid Host: field "dandelion.local:631"

That took a little more detective work because I didn’t read the Linux Magazine article carefully enough.  The solution was to add an additional directive to the cups config:

--- /backup/snapshots/dandelion.0/etc/cups/cupsd.conf   2015-06-08 08:33:31.000000000 -0400
+++ /etc/cups/cupsd.conf        2015-06-24 19:29:34.410488191 -0400
@@ -1,6 +1,7 @@
 LogLevel warn
 PageLogFormat
 # Allow remote access
+ServerAlias *
 Port 631
 Listen /run/cups/cups.sock
 # Share local printers on the local network.

Summary:

Gentoo packages required:

  • net-print/cups
  • net-dns/avahi

Also download and run airprint-generate after cups is configured and running.

If you have iOS 6+, which is pretty much a given nowadays, make sure you have the correct MIME types available, and add them if not:

echo 'image/urf urf string(0,UNIRAST<00>)' > \
    /usr/share/cups/mime/airprint.types
echo 'image/urf application/pdf 100 pdftoraster' > \
    /usr/share/cups/mime/airprint.convs

Add the appropriate services to your default runlevel, and start them as well:

# rc-update add cupsd default
# rc-update add cups-browsed default
# rc-update add avahi-daemon default