This is a tool similar to arpwatch. It main purpose is to monitor network and log discovered ethernet/ip pairings.
Main features of addrwatch:
- IPv4 and IPv6 address monitoring
- Monitoring multiple network interfaces with one daemon
- Monitoring of VLAN tagged (802.1Q) packets.
- Output to stdout, plain text file, syslog, sqlite3 db, MySQL db
- IP address usage history preserving output/logging
Addrwatch is extremely useful in networks with IPv6 autoconfiguration (RFC4862) enabled. It allows to track IPv6 addresses of hosts using IPv6 privacy extensions (RFC4941).
The main difference between arpwatch and addrwatch is the format of output files.
Arpwatch stores only current state of the network ethernet/ip pairings and allows to send email notification when a pairing change occurs. This is fine for small and rather static networks. In arpwatch case all the history of pairings is saved only in administrators mailbox. When arpwatch is used for monitoring dozen or more networks it becomes hard to keep track of the historic address usage information.
Addrwatch do not keep persistent network pairings state but instead logs all the events that allow ethernet/ip pairing discovery. For IPv4 it is ARP requests, ARP replies and ARP ACD (Address Conflict Detection) packets. For IPv6 it uses ICMPv6 Neighbor Discovery and (DAD) Duplicate Address Detection packets (Neighbor Solicitations, Neighbor Advertisements).
The output file produced by addrwatch is similar to arpwatch. Example of addrwatch output file:
1329486484 eth0 0 00:aa:bb:cc:dd:ee fe80::2aa:bbff:fecc:ddee ND_NS
1329486485 eth0 0 00:aa:bb:cc:dd:ee 192.168.1.1 ARP_REQ
1329486485 eth0 0 00:aa:bb:ff:00:11 192.168.1.3 ARP_ACD
1329486486 eth0 7 00:11:11:11:11:11 fe80::211:11ff:fe11:1111 ND_NS
1329486487 eth0 7 00:22:22:22:22:22 fe80::222:22ff:fe22:2222 ND_DAD
1329486488 eth0 7 00:33:33:33:33:33 192.168.2.2 ARP_REQ
For each pairing discovery event addrwatch produce time-stamp, interface, vlan_tag (untagged packets are marked with 0 vlan_tag), ethernet address, IP address and packet type separated by spaces.
To prevent addrwatch from producing too many duplicate output data in active networks rate-imiting should be used. Read more in 'Ratelimit' section.
Since version v1.0 addrwatch was rewritten to be more modular. Different output modules can be configured and started independently from the main data collection service.
Application architecture:
+------------------+
+-->| addrwatch_stdout |
| +------------------+
|
| +------------------+
+-------------+ +-->| addrwatch_syslog |
network | | shared memory | +------------------+
---------> addrwatch +-------------->|
| | | +------------------+
+-------------+ +-->| addrwatch_mysql |
+------------------+
In the diagram boxes represent separate processes. Main addrwach process is responsible for listening on all configured network interfaces and dumping all data to a shared memory segment. Output modules have be be started separately, they poll shared memory segment for changes and writes data to a specific output format. Current version supports stdout, syslog and mysql output formats.
Note: in addrwatch version v1.0 mysql output schema was changed to an more efficient one, by storing IP and mac addresses as binary values. To migrate existing addrwatch v0.8 installations to v1.0 there is a migration script migrate_0.8_to_1.0.sql in the main repository directory.
To compile addrwatch you must have following shared libraries:
- libpcap
- libevent
- mysqlclient (optional)
To compile addrwatch with mysql support:
$ ./configure --enable-mysql
$ make
$ make install
To compile basic addrwatch version:
$ ./configure
$ make
$ make install
If you do not want to install addrwatch to the system, skip the 'make install' step. You can find main addrwatch binary and all output addrwatch_* binaries in 'src' directory.
If sources are obtained directly from the git repository (instead of
distribution source package) project has to be bootstrapped using
autoreconf/automake. A helper shell script bottstrap.sh is included in the
repository for that. Note that bootstraping autotools project requires autoconf
and automake to be available on the system.
Example command to bootstrap autotools:
./bootstrap.sh
To automatically check source style before commiting use provided pre-commit git hook:
git config core.hooksPath .githooks
To simply try out addrwatch start it without any arguments:
$ addrwatch
When started like this addrwatch opens first non loopback interface and start logging event to the console without writing anything to disk. All events are printed to stdout, debug, warning, and err messages are sent to syslog and printed to stderr.
If you get error message: addrwatch: ERR: No suitable interfaces found!
It usually means you started addrwatch as normal user and do not have sufficient privileges to start sniffing on network interface. You should start addrwatch as root:
$ sudo addrwatch
You can specify which network interface or interfaces should be monitored by passing interface names as arguments. For example:
$ addrwatch eth0 tap0
To find out about more usage options:
$ addrwatch --help
In production environment it is recommended to start main addrwatch binary in a daemon mode, and use separate output processes for logging data. Example:
$ ./addrwatch -d eth0
$ ./addrwatch_stdout
If used without ratelimiting addrwatch reports etherment/ip pairing everytime it gets usable ARP or IPv6 ND packet. In actively used networks it generates many duplicate pairings especially for routers and servers.
Ratelimiting option '-r NUM' or '--ratelimit=NUM' surpress output of duplicate pairings for at least NUM seconds. In other words if addrwatch have discovered some pairing (mac,ip) it will not report (mac,ip) again unless NUM seconds have passed.
There is one exception to this rule to track ethernet address changes. If addrwatch have discovered pairings: (mac1,ip),(mac2,ip),(mac1,ip) within ratelimit time window it will report all three pairings. By doing so ratelimiting will not loose any information about pairing changes.
For example if we have a stream of events:
| time | MAC address | IP address |
|---|---|---|
| 0001 | 11:22:33:44:55:66 | 192.168.0.1 |
| 0015 | 11:22:33:44:55:66 | 192.168.0.1 |
| 0020 | aa:bb:cc:dd:ee:ff | 192.168.0.1 |
| 0025 | aa:bb:cc:dd:ee:ff | 192.168.0.1 |
| 0030 | 11:22:33:44:55:66 | 192.168.0.1 |
| 0035 | 11:22:33:44:55:66 | 192.168.0.1 |
| 0040 | aa:bb:cc:dd:ee:ff | 192.168.0.1 |
| 0065 | aa:bb:cc:dd:ee:ff | 192.168.0.1 |
With --ratelimit=100 we would get:
| time | MAC address | IP address |
|---|---|---|
| 0001 | 11:22:33:44:55:66 | 192.168.0.1 |
| 0020 | aa:bb:cc:dd:ee:ff | 192.168.0.1 |
| 0030 | 11:22:33:44:55:66 | 192.168.0.1 |
| 0040 | aa:bb:cc:dd:ee:ff | 192.168.0.1 |
Without such exception output would be:
| time | MAC address | IP address |
|---|---|---|
| 0001 | 11:22:33:44:55:66 | 192.168.0.1 |
| 0020 | aa:bb:cc:dd:ee:ff | 192.168.0.1 |
And we would loose information that address 192.168.0.1 was used by ethernet address 11:22:33:44:55:66 between 30-40th seconds.
To sum up ratelimiting reduces amount of duplicate information without loosing any ethernet address change events.
Ratelimit option essentially limits data granularity for IP address usage duration information (when and for what time period specific IP address was used). On the other hand without ratelimiting at all you would not get very precise IP address usage duration information anyways because some hosts might use IP address without sending ARP or ND packets as often as others do.
If NUM is set to 0, ratelimiting is disabled and all pairing discovery events are reported.
If NUM is set to -1, ratelimiting is enabled with infinitely long time window therefore all duplicate pairings are suppressed indefinitely. In this mode addrwatch acts almost as arpwatch with the exception that ethernet address changes are still reported.
It might look tempting to always use addrwatch with --ratelimit=-1 however by doing so you loose the information about when and for what period of time specific IP address was used. There will be no difference between temporary IPv6 addressed which was used once and statically configured permanent addresses.
Ethernet/ip pairing discovery can be triggered by these types of events:
- ARP_REQ - ARP Request packet. Sender hardware address (ARP header) and sender protocol address (ARP header) is saved.
- ARP_REP - ARP Reply packet. Sender hardware address (ARP header) and sender protocol address (ARP header) is saved.
- ARP_ACD - ARP Address collision detection packet. Sender hardware address (ARP header) and target protocol address (ARP header) is saved.
- ND_NS - Neighbor Solicitation packet. Source link-layer address (NS option) and source address (IPv6 header) is saved.
- ND_NA - Neighbor Advertisement packet. Target link-layer address (NA option) and source address (IPv6 header) is saved.
- ND_DAD - Duplicate Address Detection packet. Source MAC (Ethernet header) and target address (NS header) is saved.