test.py is a regression testing harness shipped along with
scylla.git, which runs C++, unit, CQL and python tests.
This is a manual for test.py.
To run test.py, Python 3.7 or higher is required.
./install-dependencies.sh should install all the required Python
modules. If install-dependencies.sh does not support your distribution,
please manually install all Python modules it lists with pip.
In order to invoke test.py, you need to build first. ./test.py will
run all existing tests in all configured build modes:
$ ./test.py
If you want to specify a specific build mode:
$ ./test.py --mode=dev
If you want to run only a specific test:
$ ./test.py suitename/testname
Build artefacts, such as test output and harness output is stored
in ./testlog. Scylla data files are stored in /tmp.
On start, test.py invokes ninja to find out configured build modes. Then
it searches all subdirectories of ./test/ for suite.yaml files: each
directory containing suite.yaml is a test suite, in which test.py then looks
for tests. All files ending with _test.cc or _test.cql are considered
tests.
A suite must contain tests of the same type, as configured in suite.yaml.
The list of found tests is matched with the optional command line test name
filter. A match is registered if filter substring exists anywhere in test
full name. For example:
$ ./test.py cql
runs cql/lwt_test, cql/lwt_batch_test, as well as
boost/cql_query_test.
The ./testlog directory is created if it doesn't exist, otherwise it is
cleared from the previous run artefacts.
Matched tests are run concurrently, with concurrency factor set to the
number of available CPU cores. test.py continues until all tests are run
even if any one of them fails.
The main idea of CQL tests is that test writer specifies CQL
statements run against Scylla, and (almost) everything else is done by
test.py: statement output is recorded in a dedicated file, and
later used to validate correctness of the test. The initial validation
must be of course performed by the author of the test. This approach
is sometimes called "approval testing" and discussion
about pros and cons of this methodology is widely available online.
To run CQL tests, test.py uses an auxiliary program,
test/pylib/cql_repl/cql_repl.py.
This program reads CQL input file, evaluates it against a pre-started
Scylla using CQL database connection, and prints output in tabular format to
stdout. A default keyspace is created automatically.
test.py invokes cql_repl.py as a pytest providing the test file
and redirecting its output to a temporary file in testlog directory.
After cql_repl.py finishes, test.py compares the output stored in the
temporary file with a pre-recorded output stored in
test/suitename/testname_test.result.
The test is considered failed if executing any CQL statement produced an
error (e.g. because the server crashed during execution) or server output
does not match one recorded in testname.result, or there is no
testname.result. The latter is possible when it's the first invocation of
the test ever.
In the event of output mismatch file test/suitename/testname_test.reject
is created, and first lines of the diff between the two files are output.
To update .result file with new output, simply overwrite it with the
reject file:
mv test/suitename/testname.re*
Note Since the result file is effectively part of the test, developers
must thoroughly examine the diff and understand the reasons for every
change before overwriting .result files.
To debug CQL tests, one can run CQL against a standalone
Scylla (possibly started in debugger) using cqlsh.
The same unit test can be run in different seastar configurations, i.e. with
different command line arguments. The custom arguments can be set in
custom_args key of the suite.yaml file.
If a test fails, its log can be found in testlog/${mode}/testname.log.
By default, all unit tests are built stripped. To build non-stripped tests,
./configure with --tests-debuginfo list-of-tests.
test.py adds some command line arguments to unit tests. The exact way in
which the test is invoked is recorded in testlog/test.py.log.
test.py supports pytest standard of tests, for suites (directories)
specifying Python test type in their scylla.yaml. For such tests,
a standalone server instance is created, and a connection URI to the
server is passed to the test. Thanks to convenience fixtures,
test writers don't need to create or cleanup connections or keyspaces.
test.py will also keep track of the used server(s) and will shut
down the server when all tests using it end.
Since there can be many pytests in a single directory (e.g. cql-pytest)
test.py creates multiple servers to parallelize their execution.
Each server is also shared among many tests, to save on setup/teardown
steps. While this speeds up execution, sharing servers complicates debugging
if a test fails.
Specifically, you should avoid leaving global artifacts in your test, even
if it fails. Typically, you could use a built-in keyspace() fixture
to create a randomly named keyspace.
At start and end of each test, test.py performs a number of sanity checks
of the used server:
- it should be up and running,
- it should not contain non-system keyspaces.
To have a full picture for a failing pytest it is necessary to identify
the server which was used to run it and the relevant fragment in the server
log. For this, test.py maintains this link through relevant log
messages and preserves Scylla output on test failure.
A typical debugging journey should start with looking at test.py.log in
testlog where, for each test it runs, test.py prints all relevant paths
to server log and pytest output.
To extend test.py logging, you can use the standard 'logging' module API.
Individual pytests are programmed to not gobble stdout, so you can can also
add prints to pytests, and they will end up in the test' log.
For example, imagine cql-pytest/test_null.py fails. The relevant lines
in test.py.log will be:
21:53:04.789 INFO> Created cluster {127.101.161.1}
21:53:04.790 INFO> Leasing Scylla cluster {127.101.161.1} for test test_null.1
21:53:04.790 INFO> Starting test test_null.1: pytest --host=127.101.161.1 -s ...test/cql-pytest/test_null.py
21:53:05.533 INFO> Test test_null.1 failed
To find out the working directory of instance 127.101.161.1 search
for its initialization message in test.py.log:
10:05:51.722 INFO> installing Scylla server in /opt/local/work/scylla/scylla/testlog/dev/scylla-1...
10:05:51.722 INFO> starting server at host 127.159.235.1 in scylla-1...
10:05:52.688 INFO> started server at host 127.159.235.1 in scylla-1, pid 2165602
Next, we can take a look at the server log, which is at
/opt/local/work/scylla/scylla/testlog/dev/scylla-1.log:
The log contains special markers, written at test start and end:
INFO 2022-08-18 10:05:52,598 [shard 0] schema_tables - Schema version changed to 8b5e9c73-7c1c-3b28-8c31-c1359210c484
------ Starting test test_null.1 ------
...
INFO 2022-08-18 10:05:53,297 [shard 0] schema_tables - Dropping keyspace cql_test_1660806353124
INFO 2022-08-18 10:05:53,304 [shard 0] schema_tables - Schema version changed to 8b5e9c73-7c1c-3b28-8c31-c1359210c484
------ Ending test test_null.1 ------
Most often there are no errors in Scylla log, so next we inspect
the test' log, which is next to the server's at
/opt/local/scylla/scylla/testlog/dev/test_null.1.log:
cql.execute(f"INSERT INTO {table1} (p,c) VALUES ('{p}', '3')")
> assert False
E assert False
What does number 1 mean in the log file name? Since test.py can
run parallel jobs and run each test multiple times with --repeat,
each execution is assigned a unique sequence number, allowing to
distinguish artifacts of different execution.
When finished debugging, you don't have to worry about deleting the remains
of a previous run, test.py will clean then up on the next execution
automatically.
When pooling and running multiple servers, we want to avoid host/port or
temporary directory clashes. We also want to make sure that test.py
doesn't leave any running servers around, even when it's interrupted
by user or with an exception. This is why test.py has a special
registry where it tracks all servers, in which each server gets a unique
address in a subnet of network 127.*.*.*. Unless killed with SIGKILL,
test.py kills all servers it creates at shutdown.
The servers created by the pool use a pre-defined set of options
to speed up boot. Some of these options are developer-only, such as
flush_schema_tables_after_modification: false. If you wish to
extend the options of a used server, you can do it by adding
extra_scylla_cmdline_options or extra_scylla_config_options
to your suite.yaml.
In addition to the standard 'Python' suite type, test.py
supports an extended pytest suite, Topology. Unlike
Python tests, Topology tests run against Scylla clusters,
and support topology operations. A standard manager
fixture is available for these. Through this fixture,
you can access individual nodes, start, stop
and restart instances, add and remove instances from the cluster.
The manager fixture connects to the cluster by sending
test.py HTTP/REST commands over a unix-domain socket.
This guarantees that test.py is fully aware of all topology
operations and can clean up resources, including added
servers, when tests end.
test.py automatically detects if a cluster can not be shared with a
subsequent test because it was manipulated with. Today the check
is quite simple: any cluster that has has nodes added or removed,
started or stopped, even if it ended up in the same state
as it was at the beginning of the test, is considered "dirty".
Such clusters are not returned to the pool, but destroyed, and
the pool is replenished with a new cluster instead.
If any of the tests fails, test.py returns a non-zero exit status.
JUNIT and XUNIT execution status XML files can be found in
testlog/${mode}/xml/ directory. These files are used
by Jenkins to produce formatted build reports. test.py will
try to add as much context information, such as fragments of log
files, exceptions, to the test XML output.
If that's not enough, a debugging journey, similar to a local one, is available in CI if you navigate to 'Build artifacts' at the Jenkins build page. This will bring you to a folder with all the test logs, preserved by Jenkins in event of test failure.
Testing is hard. Testing ScyllaDB is even harder, but we strive to ensure our testing
suite is as solid as possible. The first step is contribuing a stable (read: non-flaky) test.
To do so, when developing tests, please run them (1) in debug mode and (2) 100 times in a row (using --repeat 100),
and see that they pass successfully.
For command line help and available options, please see also:
$ ./test.py --help