## Changes
This PR adds support for UC volumes to DABs.
### Can I use a UC volume managed by DABs in `artifact_path`?
Yes, but we require the volume to exist before being referenced in
`artifact_path`. Otherwise you'll see an error that the volume does not
exist. For this case, this PR also adds a warning if we detect that the
UC volume is defined in the DAB itself, which informs the user to deploy
the UC volume in a separate deployment first before using it in
`artifact_path`.
We cannot create the UC volume and then upload the artifacts to it in
the same `bundle deploy` because `bundle deploy` always uploads the
artifacts to `artifact_path` before materializing any resources defined
in the bundle. Supporting this in a single deployment requires us to
migrate away from our dependency on the Databricks Terraform provider to
manage the CRUD lifecycle of DABs resources.
### Why do we not support `preset.name_prefix` for UC volumes?
UC volumes will not have a `dev_shreyas_goenka` prefix added in `mode:
development`. Configuring `presets.name_prefix` will be a no-op for UC
volumes. We have decided not to support prefixing for UC resources. This
is because:
1. UC provides its own namespace hierarchy that is independent of DABs.
2. Users can always manually use `${workspace.current_user.short_name}`
to configure the prefixes manually.
Customers often manually set up a UC hierarchy for dev and prod,
including a schema or catalog per developer. Thus, it's often
unnecessary for us to add prefixing in `mode: development` by default
for UC resources.
In retrospect, supporting prefixing for UC schemas and registered models
was a mistake and will be removed in a future release of DABs.
## Tests
Unit, integration test, and manually.
### Manual Testing cases:
1. UC volume does not exist:
```
➜ bundle-playground git:(master) ✗ cli bundle deploy
Error: failed to fetch metadata for the UC volume /Volumes/main/caps/my_volume that is configured in the artifact_path: Not Found
```
2. UC Volume does not exist, but is defined in the DAB
```
➜ bundle-playground git:(master) ✗ cli bundle deploy
Error: failed to fetch metadata for the UC volume /Volumes/main/caps/managed_by_dab that is configured in the artifact_path: Not Found
Warning: You might be using a UC volume in your artifact_path that is managed by this bundle but which has not been deployed yet. Please deploy the UC volume in a separate bundle deploy before using it in the artifact_path.
at resources.volumes.bar
in databricks.yml:24:7
```
---------
Co-authored-by: Pieter Noordhuis <pieter.noordhuis@databricks.com>
## Changes
After introducing the `SyncRootPath` field on the bundle (#1694), the
previous `RootPath` became ambiguous. Does it mean the bundle root path
or the sync root path? This PR renames to field to `BundleRootPath` to
remove the ambiguity.
## Tests
n/a
---------
Co-authored-by: shreyas-goenka <88374338+shreyas-goenka@users.noreply.github.com>
## Changes
Previously for all the libraries referenced in configuration DABs made
sure that there is corresponding artifact section.
But this is not really necessary and flexible, because local libraries
might be built outside of dabs context.
It also created difficult to follow logic in code where we back
referenced libraries to artifacts which was difficult to fllow
This PR does 3 things:
1. Allows all local libraries referenced in DABs config to be uploaded
to remote
2. Simplifies upload and glob references expand logic by doing this in
single place
3. Speed things up by uploading library only once and doing this in
parallel
## Tests
Added unit + integration tests + made sure that change is backward
compatible (no changes in existing tests)
---------
Co-authored-by: Pieter Noordhuis <pieter.noordhuis@databricks.com>
## Changes
This change allows to specify UC volumes path as an artifact paths so
all artifacts (JARs, wheels) are uploaded to UC Volumes.
Example configuration is here:
```
bundle:
name: jar-bundle
workspace:
host: https://foo.com
artifact_path: /Volumes/main/default/foobar
artifacts:
my_java_code:
path: ./sample-java
build: "javac PrintArgs.java && jar cvfm PrintArgs.jar META-INF/MANIFEST.MF PrintArgs.class"
files:
- source: ./sample-java/PrintArgs.jar
resources:
jobs:
jar_job:
name: "Test Spark Jar Job"
tasks:
- task_key: TestSparkJarTask
new_cluster:
num_workers: 1
spark_version: "14.3.x-scala2.12"
node_type_id: "i3.xlarge"
spark_jar_task:
main_class_name: PrintArgs
libraries:
- jar: ./sample-java/PrintArgs.jar
```
## Tests
Manually + added E2E test for Java jobs
E2E test is temporarily skipped until auth related issues for UC for
tests are resolved
## Changes
The main changes are:
1. Don't link artifacts to libraries anymore and instead just iterate
over all jobs and tasks when uploading artifacts and update local path
to remote
2. Iterating over `jobs.environments` to check if there are any local
libraries and checking that they exist locally
3. Added tests to check environments are handled correctly
End-to-end test will follow up
## Tests
Added regression test, existing tests (including integration one) pass
## Changes
The bundle path was previously stored on the `config.Root` type under
the assumption that the first configuration file being loaded would set
it. This is slightly counterintuitive and we know what the path is upon
construction of the bundle. The new location for this property reflects
this.
## Tests
Unit tests pass.
## Changes
This diagnostics type allows us to capture multiple warnings as well as
errors in the return value. This is a preparation for returning
additional warnings from mutators in case we detect non-fatal problems.
* All return statements that previously returned an error now return
`diag.FromErr`
* All return statements that previously returned `fmt.Errorf` now return
`diag.Errorf`
* All `err != nil` checks now use `diags.HasError()` or `diags.Error()`
## Tests
* Existing tests pass.
* I confirmed no call site under `./bundle` or `./cmd/bundle` uses
`errors.Is` on the return value from mutators. This is relevant because
we cannot wrap errors with `%w` when calling `diag.Errorf` (like
`fmt.Errorf`; context in https://github.com/golang/go/issues/47641).
## Changes
This is a fundamental change to how we load and process bundle
configuration. We now depend on the configuration being represented as a
`dyn.Value`. This representation is functionally equivalent to Go's
`any` (it is variadic) and allows us to capture metadata associated with
a value, such as where it was defined (e.g. file, line, and column). It
also allows us to represent Go's zero values properly (e.g. empty
string, integer equal to 0, or boolean false).
Using this representation allows us to let the configuration model
deviate from the typed structure we have been relying on so far
(`config.Root`). We need to deviate from these types when using
variables for fields that are not a string themselves. For example,
using `${var.num_workers}` for an integer `workers` field was impossible
until now (though not implemented in this change).
The loader for a `dyn.Value` includes functionality to capture any and
all type mismatches between the user-defined configuration and the
expected types. These mismatches can be surfaced as validation errors in
future PRs.
Given that many mutators expect the typed struct to be the source of
truth, this change converts between the dynamic representation and the
typed representation on mutator entry and exit. Existing mutators can
continue to modify the typed representation and these modifications are
reflected in the dynamic representation (see `MarkMutatorEntry` and
`MarkMutatorExit` in `bundle/config/root.go`).
Required changes included in this change:
* The existing interpolation package is removed in favor of
`libs/dyn/dynvar`.
* Functionality to merge job clusters, job tasks, and pipeline clusters
are now all broken out into their own mutators.
To be implemented later:
* Allow variable references for non-string types.
* Surface diagnostics about the configuration provided by the user in
the validation output.
* Some mutators use a resource's configuration file path to resolve
related relative paths. These depend on `bundle/config/paths.Path` being
set and populated through `ConfigureConfigFilePath`. Instead, they
should interact with the dynamically typed configuration directly. Doing
this also unlocks being able to differentiate different base paths used
within a job (e.g. a task override with a relative path defined in a
directory other than the base job).
## Tests
* Existing unit tests pass (some have been modified to accommodate)
* Integration tests pass
## Changes
The approach to do this was:
1. Iterate over all libraries in all job tasks
2. Find references to local libraries
3. Store pointer to `compute.Library` in the matching artifact file to
signal it should be uploaded
This breaks down when introducing #1098 because we can no longer track
unexported state across mutators. The approach in this PR performs the
path matching twice; once in the matching mutator where we check if each
referenced file has an artifacts section, and once during artifact
upload to rewrite the library path from a local file reference to an
absolute Databricks path.
## Tests
Integration tests pass.
## Changes
Removed hash from the upload path since it's not useful anyway.
The main reason for that change was to make it work on all-purpose
clusters. But in order to make it work, wheel version needs to be
increased anyway. So having only hash in path is useless.
Note: using --build-number (build tag) flag does not help with
re-installing libraries on all-purpose clusters. The reason is that
`pip` ignoring build tag when upgrading the library and only look at
wheel version.
Build tag is only used for sorting the versions and the one with higher
build tag takes priority when installed. It only works if no library is
installed.
See
a15dd75d98/src/pip/_internal/index/package_finder.py (L522-L556)https://github.com/pypa/pip/issues/4781
Thus, the only way to reinstall the library on all-purpose cluster is to
increase wheel version manually or use automatic version generation,
f.e.
```
setup(
version=datetime.datetime.utcnow().strftime("%Y%m%d.%H%M%S"),
...
)
```
## Tests
Integration tests passed.
## Changes
This PR:
1. Renames `FilesPath` -> `FilePath` and `ArtifactsPath` ->
`ArtifactPath` in the bundle and metadata configuration to make them
consistant with the json tags.
2. Fixes development / production mode error messages to point to
`file_path` and `artifact_path`
## Tests
Existing unit tests. This is a strightforward renaming of the fields.
## Changes
Previously we only supported uploading Python wheels smaller than 10mb
due to using Workspace.Import API and `content ` field
https://docs.databricks.com/api/workspace/workspace/import
By switching to use `WorkspaceFilesClient` we overcome the limit because
it uses POST body for the API instead.
## Tests
`TestAccUploadArtifactFileToCorrectRemotePath` integration test passes
```
=== RUN TestAccUploadArtifactFileToCorrectRemotePath
artifacts_test.go:28: gcp
2023/10/17 15:24:04 INFO Using Google Credentials sdk=true
helpers.go:356: Creating /Users/.../integration-test-wsfs-ekggbkcfdkid
artifacts.Upload(test.whl): Uploading...
2023/10/17 15:24:06 INFO Using Google Credentials mutator=artifacts.Upload(test) sdk=true
artifacts.Upload(test.whl): Upload succeeded
helpers.go:362: Removing /Users/.../integration-test-wsfs-ekggbkcfdkid
--- PASS: TestAccUploadArtifactFileToCorrectRemotePath (5.66s)
PASS
coverage: 14.9% of statements in ./...
ok github.com/databricks/cli/internal 6.109s coverage: 14.9% of statements in ./...
```
shreyas-goenka
Pieter Noordhuis
Andrew Nester