This PR will update the back-end to Python 3.9, and will update the base Docker image to br3ndonland/inboard instead of tiangolo/uvicorn-gunicorn-fastapi-docker. Tests are passing, and the Docker Compose network starts successfully with all expected endpoints.
# Clone the repo
git clone git@github.com:whythawk/full-stack-fastapi-postgresql.git
cd ./full-stack-fastapi-postgresql
# Run the tests
bash ./scripts/test.sh
# Generate a sample project
bash ./scripts/dev-link.sh
# Start the Docker Compose network in the sample project
cd ./dev-link
docker-compose up
Remove references to start-reload.sh (not present in inboard)
Update pyproject.toml for inboard: Gunicorn 20, Uvicorn 0.14, and FastAPI 0.66 (and thereby Starlette and pydantic) are managed there. This should also resolve the Uvicorn 0.14 issues reported in:
tiangolo/full-stack-fastapi-postgresql#389
tiangolo/uvicorn-gunicorn-docker#74
Update Python version in pyproject.toml to 3.9
Related:
tiangolo/uvicorn-gunicorn-docker#33
tiangolo/uvicorn-gunicorn-docker#72
tiangolo/uvicorn-gunicorn-docker#73
tiangolo/uvicorn-gunicorn-fastapi-docker#53
tiangolo/uvicorn-gunicorn-fastapi-docker#67
tiangolo/uvicorn-gunicorn-fastapi-docker#86
Update Poetry build backend to poetry-core
bc8d9d5
See the Poetry docs for more info on how the poetry-core build backend works.
Restrict SQLAlchemy to 1.3
f17cb9d
The previous SQLAlchemy version constraint in pyproject.toml, sqlalchemy = "^1.3.16", allowed SQLAlchemy 1.4 to be installed. SQLAlchemy 1.4 includes many API changes, and should not be installed until the back-end Python application logic has been updated.
This commit will update the version constraint to sqlalchemy = "~1.3", which allows further updates to SQLAlchemy 1.3, but not SQLAlchemy 1.4. See semver.npmjs.com for a helpful demonstration of the version syntax. This commit will also install psycopg2 through SQLAlchemy for simplicity and compatibility.
Commit 00297f9 gitignored poetry.lock. This commit will add poetry.lock to version control with Git to avoid dependency resolution errors during Docker builds.
In this project, the Dockerfile copies poetry.lock into the Docker image, but there's no step to generate poetry.lock in the first place. Without poetry.lock, dependency resolutions are commonly seen.
Example dependency resolution error
```text
❯ bash scripts/test.sh
WARNING: The following deploy sub-keys are not supported and have been ignored: labels
WARNING: The following deploy sub-keys are not supported and have been ignored: labels
WARNING: The following deploy sub-keys are not supported and have been ignored: labels
WARNING: The following deploy sub-keys are not supported and have been ignored: labels
WARNING: The following deploy sub-keys are not supported and have been ignored: labels
db uses an image, skipping
flower uses an image, skipping
pgadmin uses an image, skipping
proxy uses an image, skipping
queue uses an image, skipping
Building backend
[+] Building 15.3s (8/10)
=> [internal] load build definition from backend.dockerfile 0.2s
=> => transferring dockerfile: 797B 0.1s
=> [internal] load .dockerignore 0.1s
=> => transferring context: 2B 0.0s
=> [internal] load metadata for ghcr.io/br3ndonland/inboard:fastapi-python3.9 0.3s
=> [1/6] FROM ghcr.io/br3ndonland/inboard:fastapi-python3.9@sha256:5591f436a37490a1569afd9e55ae 0.0s
=> [internal] load build context 0.0s
=> => transferring context: 64.67kB 0.0s
=> CACHED [2/6] COPY ./app/pyproject.toml ./app/poetry.lock* /app/ 0.0s
=> CACHED [3/6] WORKDIR /app/ 0.0s
=> ERROR [4/6] RUN bash -c "if [ true == 'true' ] ; then poetry install --no-root ; else poetr 14.4s
------
> [4/6] RUN bash -c "if [ true == 'true' ] ; then poetry install --no-root ; else poetry install --no-root --no-dev ; fi":
Skipping virtualenv creation, as specified in config file.
Installing dependencies from lock file
Warning: The lock file is not up to date with the latest changes in pyproject.toml. You may be getting dependencies. Run update to update them.
SolverProblemError
Because app depends on sqlalchemy-stubs (^0.3) which doesn't match any versions, version solving failed.
at /opt/poetry/lib/poetry/puzzle/solver.py:241 in _solve
237│ packages = result.packages
238│ except OverrideNeeded as e:
239│ return self.solve_in_compatibility_mode(e.overrides, use_latest=use_latest)
240│ except SolveFailure as e:
→ 241│ raise SolverProblemError(e)
242│
243│ results = dict(
244│ depth_first_search(
245│ PackageNode(self._package, packages), aggregate_package_nodes
------
executor failed running [/bin/sh -c bash -c "if [ $INSTALL_DEV == 'true' ] ; then poetry install --no-root ; else poetry install --no-root --no-dev ; fi"]: exit code: 1
ERROR: Service 'backend' failed to build : Build failed
```
ERROR: for flower Cannot start service flower: OCI runtime create
failed: container_linux.go:380: starting container process caused:
exec: "celery --broker=amqp://guest@queue:5672//":
stat celery --broker=amqp://guest@queue:5672//:
no such file or directory: unknown
Update Celery Docker image to Python 3.9
bbd84b8
Note that, while the Celery Docker image has been upgraded to Python 3.9, the application is still running Celery 4. Further updates to the source code will be needed before upgrading from Celery 4.
Fix Travis CI Docker ARG error
2cbf37f
Travis CI may be running an old version of Docker.
ERROR: Dockerfile parse error line 18: ARG requires exactly one argument
Description
This PR will update the back-end to Python 3.9, and will update the base Docker image to br3ndonland/inboard instead of tiangolo/uvicorn-gunicorn-fastapi-docker. Tests are passing, and the Docker Compose network starts successfully with all expected endpoints.
To try it out locally, install Docker, then:
Browse to the project's URLs:
PGADMIN_DEFAULT_EMAILandPGADMIN_DEFAULT_PASSWORDfrom .env file): http://localhost:5050FLOWER_BASIC_AUTHfrom .env file): http://localhost:5555Changes
Use inboard with Python 3.9 for back-end
df21f12
APP_MODULE,PRE_START_PATHstart-reload.sh(not present in inboard)pyproject.tomlfor inboard: Gunicorn 20, Uvicorn 0.14, and FastAPI 0.66 (and thereby Starlette and pydantic) are managed there. This should also resolve the Uvicorn 0.14 issues reported in:pyproject.tomlto 3.9Update Poetry build backend to
poetry-corebc8d9d5
See the Poetry docs for more info on how the
poetry-corebuild backend works.Restrict SQLAlchemy to 1.3
f17cb9d
The previous SQLAlchemy version constraint in
pyproject.toml,sqlalchemy = "^1.3.16", allowed SQLAlchemy 1.4 to be installed. SQLAlchemy 1.4 includes many API changes, and should not be installed until the back-end Python application logic has been updated.This commit will update the version constraint to
sqlalchemy = "~1.3", which allows further updates to SQLAlchemy 1.3, but not SQLAlchemy 1.4. See semver.npmjs.com for a helpful demonstration of the version syntax. This commit will also installpsycopg2through SQLAlchemy for simplicity and compatibility.Version-control
poetry.lockwith Git16130f0
tiangolo/full-stack-fastapi-postgresql#69 tiangolo/full-stack-fastapi-postgresql#123 tiangolo/full-stack-fastapi-postgresql#144 tiangolo/full-stack-fastapi-postgresql@00297f9
Commit 00297f9 gitignored
poetry.lock. This commit will addpoetry.lockto version control with Git to avoid dependency resolution errors during Docker builds.There is no established convention for working with Poetry in Docker, so developers have to consider each use case individually. See https://github.com/python-poetry/poetry/discussions/1879#discussioncomment-346113.
In this project, the Dockerfile copies
poetry.lockinto the Docker image, but there's no step to generatepoetry.lockin the first place. Withoutpoetry.lock, dependency resolutions are commonly seen.Example dependency resolution error
```text ❯ bash scripts/test.sh WARNING: The following deploy sub-keys are not supported and have been ignored: labels WARNING: The following deploy sub-keys are not supported and have been ignored: labels WARNING: The following deploy sub-keys are not supported and have been ignored: labels WARNING: The following deploy sub-keys are not supported and have been ignored: labels WARNING: The following deploy sub-keys are not supported and have been ignored: labels db uses an image, skipping flower uses an image, skipping pgadmin uses an image, skipping proxy uses an image, skipping queue uses an image, skipping Building backend [+] Building 15.3s (8/10) => [internal] load build definition from backend.dockerfile 0.2s => => transferring dockerfile: 797B 0.1s => [internal] load .dockerignore 0.1s => => transferring context: 2B 0.0s => [internal] load metadata for ghcr.io/br3ndonland/inboard:fastapi-python3.9 0.3s => [1/6] FROM ghcr.io/br3ndonland/inboard:fastapi-python3.9@sha256:5591f436a37490a1569afd9e55ae 0.0s => [internal] load build context 0.0s => => transferring context: 64.67kB 0.0s => CACHED [2/6] COPY ./app/pyproject.toml ./app/poetry.lock* /app/ 0.0s => CACHED [3/6] WORKDIR /app/ 0.0s => ERROR [4/6] RUN bash -c "if [ true == 'true' ] ; then poetry install --no-root ; else poetr 14.4s ------ > [4/6] RUN bash -c "if [ true == 'true' ] ; then poetry install --no-root ; else poetry install --no-root --no-dev ; fi": Skipping virtualenv creation, as specified in config file. Installing dependencies from lock file Warning: The lock file is not up to date with the latest changes in pyproject.toml. You may be getting dependencies. Run update to update them. SolverProblemError Because app depends on sqlalchemy-stubs (^0.3) which doesn't match any versions, version solving failed. at /opt/poetry/lib/poetry/puzzle/solver.py:241 in _solve 237│ packages = result.packages 238│ except OverrideNeeded as e: 239│ return self.solve_in_compatibility_mode(e.overrides, use_latest=use_latest) 240│ except SolveFailure as e: → 241│ raise SolverProblemError(e) 242│ 243│ results = dict( 244│ depth_first_search( 245│ PackageNode(self._package, packages), aggregate_package_nodes ------ executor failed running [/bin/sh -c bash -c "if [ $INSTALL_DEV == 'true' ] ; then poetry install --no-root ; else poetry install --no-root --no-dev ; fi"]: exit code: 1 ERROR: Service 'backend' failed to build : Build failed ```Pin mher/flower Docker image to fix startup error
3d2e093
tiangolo/full-stack-fastapi-postgresql#398 tiangolo/full-stack-fastapi-postgresql#399
Error:
Update Celery Docker image to Python 3.9
bbd84b8
Note that, while the Celery Docker image has been upgraded to Python 3.9, the application is still running Celery 4. Further updates to the source code will be needed before upgrading from Celery 4.
Fix Travis CI Docker
ARGerror2cbf37f
Travis CI may be running an old version of Docker.