openobserve/tests/api-testing

OpenObserve API Tests

End-to-end pytest suite against the OpenObserve REST API. Runs in CI on every PR (.github/workflows/api-testing.yml); also runnable locally.

Quick start

# one-time setup
brew install rye           # or: curl -sSf https://rye.astral.sh/get | bash
rye sync                   # install Python + deps from requirements.lock
cargo build                # build the OO debug binary (~10 min first time)

# every run
ZO_ROOT_USER_EMAIL=root@example.com \
ZO_ROOT_USER_PASSWORD='Complexpass#123' \
ZO_BASE_URL=http://localhost:5080/ \
WS_ZO_BASE_URL=ws://localhost:5080/ \
target/debug/openobserve &
rye run pytest             # full suite
rye run pytest -m framework        # just the new-framework self-tests
rye run pytest tests/test_kvstore.py -v   # one file

Layout

tests/api-testing/
├── pyproject.toml          # pytest + ruff config
├── conftest.py             # legacy session-wide fixtures (do not extend)
├── support/                # NEW FRAMEWORK — write new tests against this
│   ├── client.py           # OpenObserveClient: HTTP entry point
│   ├── endpoints/          # one wrapper class per resource group
│   │   ├── search.py       # client.search.sql(...), .hits(...), .count(...)
│   │   ├── streams.py      # client.streams.list(), .delete(...), .ingest_json(...)
│   │   ├── users.py
│   │   ├── dashboards.py
│   │   ├── alerts.py
│   │   ├── folders.py
│   │   └── kvstore.py
│   ├── factories.py        # payload builders (user_payload, dashboard_payload, …)
│   ├── wait.py             # wait_until(predicate, timeout, ...) — REPLACES time.sleep
│   └── fixtures.py         # client, temp_stream_name, temp_user_email, …
├── tests/                  # test files
│   ├── conftest.py
│   ├── test_framework_selfcheck.py   # 14 tests that prove support/ works
│   └── test_*.py           # 40+ legacy test files (in revamp progress)
├── helpers/                # non-test shared code
│   ├── sourcemap_helpers.py
│   └── workflow/           # workflow chain helpers (with embedded pages/)
├── fixtures/               # static test data
│   └── rum/                # RUM ingest scripts
└── README.md               # this file

Writing a new test — the rules

Use the new framework. Don't copy-paste from old test_*.py files; they are scheduled for rewrite (see revamp plan).

# tests/<area>/test_my_feature.py
from support.client import OpenObserveClient
from support.factories import user_payload
from support.fixtures import client, temp_user_email   # noqa: F401
from support.wait import wait_until


def test_create_user_returns_complete_record(client: OpenObserveClient, temp_user_email: str):
    """Create a user — verify the response has the email AND role we sent."""
    payload = user_payload(email=temp_user_email, role="admin")
    resp = client.users.create(payload)

    assert resp.status_code == 200, resp.text
    body = resp.json()
    # VALIDATE THE BODY, not just status_code
    assert body.get("data", {}).get("email") == temp_user_email
    assert body.get("data", {}).get("role") == "admin"
    # cleanup happens automatically via temp_user_email fixture

Mandatory patterns

✅ Do	❌ Don't
from support.fixtures import client	session = create_session; url = base_url; org_id = "default" (307 occurrences in legacy code)
client.streams.list()	session.get(f"{url}api/{org_id}/streams")
body = resp.json(); assert body["foo"] == ...	assert resp.status_code == 200 and stop
wait_until(lambda: client.search.count(sql) >= N, timeout=30)	time.sleep(5)
unique_email("user") / temp_user_email fixture	hardcoded pytests@gmail.com
logger.debug("got %s", resp.status_code)	print(resp.content)
start, end = time_window(minutes=15)	hardcoded start_time = 1700629279639000

Anti-patterns that fail review

• assert resp.status_code in [200, 400, 404, 500] — pick one. If the API behaves wrongly, file a bug; don't paper over it.
• pytest.skip(...) when the test discovers the bug it's meant to catch — use @pytest.mark.xfail(strict=True).
• Commented-out test functions — delete them.
• pages.* direct imports — that subtree is legacy. New tests use support.client.

Lint

rye run ruff check support/ tests/test_framework_selfcheck.py  # strict mode
rye run ruff check --fix support/                              # autofix

Existing legacy tests have a per-file ruff-ignore (so print etc. don't fail the whole suite); new tests under support/ and any new well-formed test file should pass strict ruff.

Test markers

Marker	Meaning
framework	Self-tests for support/ framework. Run with -m framework.
order(N)	Run order pin (legacy — avoid in new tests).

Troubleshooting

• KeyError: 'ZO_BASE_URL' on collection — conftest.py reads env vars at import. Export them before running pytest (even --collect-only).
• test_workflow.py fails with SMTP error — ZO_SMTP_ENABLED=true not set. CI does this; locally you'd need to wire an SMTP env.
• Test_organisations.py warning about yield — test_create_organization uses pre-pytest-4 yield-style and is silently ignored. Scheduled for rewrite in Phase 4.

Revamp in progress

This suite is being incrementally reworked. See:

• ../../tests/ui-testing/MD_Files/api tests revamp/API_TESTS_REVAMP.md — the plan
• ../../tests/ui-testing/MD_Files/api tests revamp/oss_rework_progress/ — phase-by-phase progress