This is chutney. It doesn't do much so far. It isn't ready for prime-time.

If it breaks, you get to keep all the pieces.

It is supposed to be a good tool for:

Configuring a testing tor network
Launching and monitoring a testing tor network
Running tests on a testing tor network

Right now it only sorta does these things.

You will need:

A supported version of Python 3
- (we support Python versions that are still getting updates), and
Tor binaries.

Chutney checks for Tor binaries in this order:

If you run chutney's tools/test-network.sh from a tor build directory, (or set the environment variable $TOR_DIR to a tor build directory,) chutney will automatically detect the tor binaries, or
If you put the location of the 'tor' and 'tor-gencert' binaries in the environment variables $CHUTNEY_TOR and $CHUTNEY_TOR_GENCERT, respectively, chutney will use those binaries, or
You will need tor and tor-gencert installed somewhere in your path.

Stuff to try:

Automated Setup, Verification, and Shutdown: ./tools/test-network.sh --flavor basic-min ./tools/test-network.sh --coverage ./tools/test-network.sh --tor-path ./tools/test-network.sh --tor --tor-gencert (--tor-path and $TOR_DIR override --tor and --tor-gencert.) (The script tries hard to find tor.) ./tools/test-network.sh --chutney-path (The script is pretty good at finding chutney.) ./tools/test-network.sh --allow-failures

test-network.sh looks for some tor binaries (either in a nearby build directory or in your $PATH), configures a comprehensive tor test network, launches it, then verifies data transmission through it, and cleans up after itself. Relative paths are supported.

You can modify its configuration using command-line arguments, or use the chutney environmental variables documented below:

Verification Options:

repeats bootstrap and verify

--allow-failures CHUTNEY_ALLOW_FAILURES=N

repeats verify

--rounds CHUTNEY_ROUNDS=N

makes multiple connections within verify

--connections CHUTNEY_CONNECTIONS=N

Timing Options: --start-time CHUTNEY_START_TIME=N --min-start-time CHUTNEY_MIN_START_TIME=N --bootstrap-time CHUTNEY_BOOTSTRAP_TIME=N --stop-time CHUTNEY_STOP_TIME=N

Traffic Options: --data CHUTNEY_DATA_BYTES=N --hs-multi-client CHUTNEY_HS_MULTI_CLIENT=N

Address/DNS Options: --ipv4 CHUTNEY_LISTEN_ADDRESS=IPV4 --ipv6 CHUTNEY_LISTEN_ADDRESS_V6=IPV6

Chutney uses /etc/resolv.conf if none of these options are set

--dns-conf CHUTNEY_DNS_CONF=PATH --offline CHUTNEY_DNS_CONF=/dev/null

Use tor's compile-time default for ServerDNSResolvConfFile

--dns-conf-default CHUTNEY_DNS_CONF=""

Sandbox Options:

--sandbox CHUTNEY_TOR_SANDBOX=N (0 or 1)

Warning Options: --all-warnings CHUTNEY_WARNINGS_IGNORE_EXPECTED=false CHUTNEY_WARNINGS_SUMMARY=false --no-warnings CHUTNEY_WARNINGS_SKIP=true --only-warnings CHUTNEY_WARNINGS_ONLY=true --diagnostics CHUTNEY_DIAGNOSTICS=true --only-diagnostics CHUTNEY_DIAGNOSTICS_ONLY=true

Expert Options: --debug CHUTNEY_DEBUG=true --coverage USE_COVERAGE_BINARY=true --dry-run NETWORK_DRY_RUN=true --quiet ECHO=true

--controlling-pid CHUTNEY_CONTROLLING_PID=N --net-dir CHUTNEY_DATA_DIR=PATH (These are advanced options: in the past, they have had long-standing bugs.)

Standard Actions: ./chutney configure networks/basic ./chutney start networks/basic ./chutney status networks/basic ./chutney wait_for_bootstrap networks/basic ./chutney verify networks/basic ./chutney hup networks/basic ./chutney stop networks/basic

Bandwidth Tests: ./chutney configure networks/basic-min ./chutney start networks/basic-min ./chutney status networks/basic-min

Send 100MB of data per client connection

CHUTNEY_DATA_BYTES=104857600 ./chutney verify networks/basic-min ./chutney stop networks/basic-min

If chutney sends at least 5 MB of data, and it takes at least one second, verify produces performance figures for:

Single Stream Bandwidth: the speed of the slowest stream, end-to-end
Overall tor Bandwidth: the sum of the bandwidth across each tor instance The overall bandwidth approximates the CPU-bound tor performance on the current machine, assuming tor, chutney, and the OS are multithreaded, and network performance is infinite.

Connection Tests: ./chutney configure networks/basic-025 ./chutney start networks/basic-025 ./chutney status networks/basic-025

Make 5 simultaneous connections from each client through a random exit

CHUTNEY_CONNECTIONS=5 ./chutney verify networks/basic-025 ./chutney stop networks/basic-025

Run 5 sequential verification rounds

CHUTNEY_ROUNDS=5 ./tools/test-network.sh --flavour basic

HS Connection Tests: ./chutney configure networks/hs-025 ./chutney start networks/hs-025 ./chutney status networks/hs-025 CHUTNEY_HS_MULTI_CLIENT=1 ./chutney verify networks/hs-025

Make a connection from each client to each hs

Default behavior is one client connects to each HS

./chutney stop networks/hs-025

Bandwidth File Tests: ./tools/test-network.sh --flavour bwfile

Warning: Can't open bandwidth file at configured location: /tmp/bwfile

Create a bwfile with no bandwidths, that is valid for a few days

date +%s > /tmp/bwfile ./tools/test-network.sh --flavour bwfile

Multiple Tests:

Chutney can allow a certain number of failed tests. You can either set CHUTNEY_ALLOW_FAILURES or use an --allow-failures command-line option to control this. Chutney will then reattempt the test, from bootstrap through shutdown, until either it succeeds, or until it has failed $CHUTNEY_ALLOW_FAILURES+1 times. The default value is zero, so the default behavior will not change.

You can also use CHUTNEY_ROUNDS=N to run multiple verification rounds, or CHUTNEY_CONNECTIONS=N to make multiple connections within each verification round. Any round or connection failure will fail the current test.

Bootstrapping the network:

Chutney expects a tor network to bootstrap in these stages: 1. All directory authorities (DirAuths) bootstrap to 100%. 2. The DirAuths produce the first consensus. Usually, this consensus only contains authorities. 3. The DirAuths produce a bootstrap consensus. This consensus has enough relays for: * clients and relays to bootstrap, and * relays to perform reachability self-tests. Usually, this consensus needs at least 3 nodes. This consensus is usually the first or second consensus.

4.  Relays bootstrap to 100%.
5.  Relays with "AssumeReachable 1" publish their descriptors to the
    DirAuths.

6.  Relays perform ORPort reachability self-tests.
    If the consensus contains at least 1 exit, relays also perform DirPort
    reachability self-tests.
7.  Relays publish their descriptors to the DirAuths.
8.  The DirAuths produce a complete consensus, microdesc consensus, and
    microdescriptors. A complete consensus contains:
      * the authorities,
      * any bridge authorities, if present, and
      * all relays (including exits).
    Bridges, clients, and onion services are not included in the consensus.

9.  Bridges publish their descriptors to the Bridge Auth.
10. The Bridge Auth produces a bridge networkstatus.

11. Relays and bridges download all consensus flavours, then download
    descriptors and microdescriptors.
12. Bridge clients download the descriptors for their bridges.
13. Clients (including bridge clients, and onion services), download the
    most recent microdesc consensus, and microdescriptors.
14. Clients bootstrap to 100%.
    (Clients can bootstrap as soon as the consensus contains enough nodes,
    so this step can depend on step 3, not step 13.)

15. Onion Services publish their descriptors to Onion Service directories
    (otherwise known as hidden service directories, or HSDirs).

The tools/test-network.sh script uses the chutney wait_for_bootstrap command to wait for the network to bootstrap.

wait_for_bootstrap waits up to CHUTNEY_START_TIME seconds (default: 120), checking whether: * the logged bootstrapped status for every node is 100% (steps 9 and 14), and * directory information has been distributed throughout the network (steps 7-8, 11-13).

When waiting for dir info distribution, wait_for_bootstrap checks if: * each relay descriptor has been posted to every authority (step 7), * each relay is in the consensus, and the microdesc consensus, at every authority (step 8), * a complete consensus and microdesc consensus has been distributed to relays and bridges (step 11), * all authority and relay descriptors have been distributed to relays and bridges (step 11), * all bridge descriptors have been distributed to all bridge clients (step 12), and * a complete microdesc consensus has been distributed to clients (step 13).

wait_for_bootstrap does not currently check the following dir info: * microdescriptors (steps 8, 11, and 13, chutney ticket #33407), * bridge descriptors at the bridge authority (steps 9-10, tor ticket #33582, chutney ticket #33428), and * onion services have published their descriptors to the HSDirs (step 15, chutney ticket #33609).

After bootstrapping and dir info distribution, wait_for_bootstrap waits until the network has been running for at least CHUTNEY_MIN_START_TIME seconds (default 0 seconds for tor > 0.3.5., 65 seconds for tor <= 0.3.5.), to compensate for microdesc download issues in older tor versions.

In addition, wait_for_bootstrap also waits an extra: * 10 seconds for clients to download microdescs, and * 30 seconds for onion services to upload their descriptors. We expect that these delays will be removed, once the relevant checks are implemented in chutney.

If the CHUTNEY_START_TIME has elapsed, and some nodes have not bootstrapped, or there are some nodes missing from the consensus, wait_for_bootstrap dumps the bootstrap statuses, and exits with a failure.

Verifying the network:

Commands like "chutney verify" start immediately, and keep trying for CHUTNEY_BOOTSTRAP_TIME seconds (default: 60). If it hasn't been successful after that time, it fails. If CHUTNEY_BOOTSTRAP_TIME is negative, the script leaves the network running, and exits after CHUTNEY_START_TIME (without verifying).

Shutting down the network:

The tools/test-network.sh script waits CHUTNEY_STOP_TIME seconds after verifying, then exits (default: immediately). If CHUTNEY_STOP_TIME is negative, the script leaves the network running, and exits after verifying.

If none of these options are negative, test-network.sh tells the tor processes to exit after it exits, using CHUTNEY_CONTROLLING_PID. To disable this functionality, set CHUTNEY_CONTROLLING_PID to 1 or less.

Changing the network address:

Chutney defaults to binding to localhost. To change the IPv4 bind address, set the CHUTNEY_LISTEN_ADDRESS environment variable. Similarly, change CHUTNEY_LISTEN_ADDRESS_V6 for IPv6: it defaults to "no IPv6 address". Setting it to some interface's IP address allows us to make the simulated Tor network available on the network.

IPv6 support for both Tor and Chutney is a work in progress. Currently, chutney verifies IPv6 client, bridge client (?), hidden service, and exit connections. It does not use IPv6 SOCKSPorts or HiddenServicePorts.

Using DNS:

Chutney verify uses IP addresses by default. It does not need to look up any hostnames. We recommend that chutney users disable DNS using --offline or CHUTNEY_DNS_CONF=/dev/null , because any DNS failures causes tests to fail. Chutney's DNS queries also produce external traffic in a predictable pattern.

If you want to use a hostname with CHUTNEY_LISTEN_ADDRESS[_V6], or you want to run tests that use DNS, set CHUTNEY_DNS_CONF to the path to a file in resolv.conf format. Chutney's default of /etc/resolv.conf should be fine for most UNIX-based operating systems. If your tor is compiled with a different default, use --dns-resolv-conf-default or CHUTNEY_DNS_CONF="".

When the CHUTNEY_DNS_CONF file does not exist, or is a broken symlink, chutney uses /dev/null instead. This is a workaround for bugs in tor's use of eventdns. For example, macOS deletes the resolv.conf file when it thinks the network is down: this can make tor exits reject all traffic, even if a working DNS server is running on 127.0.0.1:53.

When tor has no working name servers (including --offline mode), it can crash on SETCONF. (Chutney does not use SETCONF, but some external tor controllers do.) To avoid this crash, set CHUTNEY_DNS_CONF to a file containing a working name server address. For your convenience, chutney provides a local resolv.conf file containing IPv4, IPv6, and "localhost". Use --dns-conf resolv.conf (relative paths work).

The tor sandbox:

Chutney can run with the tor seccomp sandbox enabled. But if tor's sandbox is broken on your local version of glibc, you can set CHUTNEY_TOR_SANDBOX=0 to disable the sandbox. If CHUTNEY_TOR_SANDBOX is unset, Sandbox defaults to 1 on Linux, and 0 on other platforms.

The configuration files:

networks/basic holds the configuration for the network you're configuring above. It refers to some torrc template files in torrc_templates/.

Chutney uses a templating system to produce torrc files from the templates. These torrc files can be modified using various chutney options.

The working files:

chutney sticks its working files, including all generated torrc files, data directories, log files, etc, in ./net/. Each tor instance gets a subdirectory of net/nodes.

You can override the directory "./net" with the CHUTNEY_DATA_DIR environment variable.

Test scripts:

The test scripts are stored in the "scripts/chutney_tests" directory. These Python files must define a "run_test(network)" function. Files starting with an underscore ("_") are ignored.

Test scripts can be run using the following syntax: ./chutney networks/

The chutney verify command is implemented using a test script.

Test scripts in the test directory with the same name as standard commands are run instead of that standard command. This allows expert users to replace the standard chutney commands with modified versions.

adventureloop / chutney