Files
neon/test_runner
Christian Schwarz 881356c417 add metrics to detect eviction-induced thrashing (#3837)
This patch adds two metrics that will enable us to detect *thrashing* of
layers, i.e., repetitions of `eviction, on-demand-download, eviction,
... ` for a given layer.

The first metric counts all layer evictions per timeline. It requires no
further explanation. The second metric counts the layer evictions where
the layer was resident for less than a given threshold.

We can alert on increments to the second metric. The first metric will
serve as a baseline, and further, it's generally interesting, outside of
thrashing.

The second metric's threshold is configurable in PageServerConf and
defaults to 24h. The threshold value is reproduced as a label in the
metric because the counter's value is semantically tied to that
threshold. Since changes to the config and hence the label value are
infrequent, this will have low storage overhead in the metrics storage.

The data source to determine the time that the layer was resident is the
file's `mtime`. Using `mtime` is more of a crutch. It would be better if
Pageserver did its own persistent bookkeeping of residence change events
instead of relying on the filesystem. We had some discussion about this:
https://github.com/neondatabase/neon/pull/3809#issuecomment-1470448900

My position is that `mtime` is good enough for now. It can theoretically
jump forward if someone copies files without resetting `mtime`. But that
shouldn't happen in practice. Note that moving files back and forth
doesn't change `mtime`, nor does `chown` or `chmod`. Lastly, `rsync -a`,
which is typically used for filesystem-level backup / restore, correctly
syncs `mtime`.

I've added a label that identifies the data source to keep options open
for a future, better data source than `mtime`. Since this value will
stay the same for the time being, it's not a problem for metrics
storage.

refs https://github.com/neondatabase/neon/issues/3728
2023-03-20 16:11:36 +01:00
..

Neon test runner

This directory contains integration tests.

Prerequisites:

  • Correctly configured Python, see /docs/sourcetree.md
  • Neon and Postgres binaries
    • See the root README.md for build directions If you want to test tests with test-only APIs, you would need to add --features testing to Rust code build commands. For convenience, repository cargo config contains build_testing alias, that serves as a subcommand, adding the required feature flags. Usage example: cargo build_testing --release is equivalent to cargo build --features testing --release
    • Tests can be run from the git tree; or see the environment variables below to run from other directories.
  • The neon git repo, including the postgres submodule (for some tests, e.g. pg_regress)

Test Organization

Regression tests are in the 'regress' directory. They can be run in parallel to minimize total runtime. Most regression test sets up their environment with its own pageservers and safekeepers (but see TEST_SHARED_FIXTURES).

'pg_clients' contains tests for connecting with various client libraries. Each client test uses a Dockerfile that pulls an image that contains the client, and connects to PostgreSQL with it. The client tests can be run against an existing PostgreSQL or Neon installation.

'performance' contains performance regression tests. Each test exercises a particular scenario or workload, and outputs measurements. They should be run serially, to avoid the tests interfering with the performance of each other. Some performance tests set up their own Neon environment, while others can be run against an existing PostgreSQL or Neon environment.

Running the tests

There is a wrapper script to invoke pytest: ./scripts/pytest. It accepts all the arguments that are accepted by pytest. Depending on your installation options pytest might be invoked directly.

Test state (postgres data, pageserver state, and log files) will be stored under a directory test_output.

You can run all the tests with:

./scripts/pytest

If you want to run all the tests in a particular file:

./scripts/pytest test_pgbench.py

If you want to run all tests that have the string "bench" in their names:

./scripts/pytest -k bench

To run tests in parellel we utilize pytest-xdist plugin. By default everything runs single threaded. Number of workers can be specified with -n argument:

./scripts/pytest -n4

By default performance tests are excluded. To run them explicitly pass performance tests selection to the script:

./scripts/pytest test_runner/performance

Useful environment variables:

NEON_BIN: The directory where neon binaries can be found. POSTGRES_DISTRIB_DIR: The directory where postgres distribution can be found. Since pageserver supports several postgres versions, POSTGRES_DISTRIB_DIR must contain a subdirectory for each version with naming convention v{PG_VERSION}/. Inside that dir, a bin/postgres binary should be present. DEFAULT_PG_VERSION: The version of Postgres to use, This is used to construct full path to the postgres binaries. Format is 2-digit major version nubmer, i.e. DEFAULT_PG_VERSION="14" TEST_OUTPUT: Set the directory where test state and test output files should go. TEST_SHARED_FIXTURES: Try to re-use a single pageserver for all the tests. NEON_PAGESERVER_OVERRIDES: add a ;-separated set of configs that will be passed as RUST_LOG: logging configuration to pass into Neon CLI

Useful parameters and commands:

--pageserver-config-override=${value} -c values to pass into pageserver through neon_local cli

--preserve-database-files to preserve pageserver (layer) and safekeer (segment) timeline files on disk after running a test suite. Such files might be large, so removed by default; but might be useful for debugging or creation of svg images with layer file contents.

Let stdout, stderr and INFO log messages go to the terminal instead of capturing them: ./scripts/pytest -s --log-cli-level=INFO ... (Note many tests capture subprocess outputs separately, so this may not show much.)

Exit after the first test failure: ./scripts/pytest -x ... (there are many more pytest options; run pytest -h to see them.)

Writing a test

Every test needs a Neon Environment, or NeonEnv to operate in. A Neon Environment is like a little cloud-in-a-box, and consists of a Pageserver, 0-N Safekeepers, and compute Postgres nodes. The connections between them can be configured to use JWT authentication tokens, and some other configuration options can be tweaked too.

The easiest way to get access to a Neon Environment is by using the neon_simple_env fixture. The 'simple' env may be shared across multiple tests, so don't shut down the nodes or make other destructive changes in that environment. Also don't assume that there are no tenants or branches or data in the cluster. For convenience, there is a branch called empty, though. The convention is to create a test-specific branch of that and load any test data there, instead of the 'main' branch.

For more complicated cases, you can build a custom Neon Environment, with the neon_env fixture:

def test_foobar(neon_env_builder: NeonEnvBuilder):
    # Prescribe the environment.
    # We want to have 3 safekeeper nodes, and use JWT authentication in the
    # connections to the page server
    neon_env_builder.num_safekeepers = 3
    neon_env_builder.set_pageserver_auth(True)

    # Now create the environment. This initializes the repository, and starts
    # up the page server and the safekeepers
    env = neon_env_builder.init_start()

    # Run the test
    ...

For more information about pytest fixtures, see https://docs.pytest.org/en/stable/fixture.html

At the end of a test, all the nodes in the environment are automatically stopped, so you don't need to worry about cleaning up. Logs and test data are preserved for the analysis, in a directory under ../test_output/<testname>

Before submitting a patch

Ensure that you pass all obligatory checks.

Also consider:

  • Writing a couple of docstrings to clarify the reasoning behind a new test.
  • Adding more type hints to your code to avoid Any, especially:
    • For fixture parameters, they are not automatically deduced.
    • For function arguments and return values.