Use X-Databricks-Workspace-Id for workspace routing#1436
Merged
Conversation
Divyansh-db
temporarily deployed
to
test-trigger-is
GitHub Actions
Inactive
Divyansh-db
temporarily deployed
to
test-trigger-is
GitHub Actions
Inactive
Mechanical literal swap matching the updated pysdkv0/pysdkv1 generator templates that emit the workspace routing header on workspace-scoped operations. 35 generated files, 969 occurrences (symmetric +/-): - 30 generated service modules under databricks/sdk/service/ - databricks/sdk/__init__.py (WorkspaceClient.get_workspace_id()) - 4 generated test fixtures under tests/databricks/sdk/service/ No behavioral change beyond the header name; emission remains gated on the existing `cfg.workspace_id` check. Signed-off-by: Divyansh Vijayvergia <divyansh.vijayvergia@databricks.com>
Generator can't touch these (no DO NOT EDIT marker), or need post-generation reformatting because the new literal is longer: - config.py: SQL HTTP path resolver header read + widen workspace_id field comment (now accepts non-classic workspace identifier formats) - mixins/workspace.py: upload/download header injection (2 sites) - mixins/sharing.py: shares-list header injection - mixins/files.py: storage-proxy workspace probe (2 sites) - tests/test_config.py: renamed three header tests (test_workspace_org_id_header_* -> test_workspace_id_header_*, test_no_org_id_header_* -> test_no_workspace_id_header_*) and swapped asserted literals/docstrings - tests/test_files.py: updated mock response header fixtures - tests/integration/test_unified_profile.py: comment refresh - NEXT_CHANGELOG.md: internal-changes entry - __init__.py: black-wrap of the get_workspace_id() probe line that exceeded 120 chars after the longer literal substitution Signed-off-by: Divyansh Vijayvergia <divyansh.vijayvergia@databricks.com>
Mechanical literal swap on services that picked up the old header literal after the most recent SDK regen (#1438): - new file: databricks/sdk/service/bundle.py (14 sites) - new methods in databricks/sdk/service/dashboards.py (1 site) - new methods in databricks/sdk/service/postgres.py (1 site) Symmetric 16/16 diff; no logic change. Signed-off-by: Divyansh Vijayvergia <divyansh.vijayvergia@databricks.com>
Divyansh-db
force-pushed
the
new-workspace-parameter
branch
from
b2d0839 to
370975a
Compare
Divyansh-db
temporarily deployed
to
test-trigger-is
GitHub Actions
Inactive
Divyansh-db
temporarily deployed
to
test-trigger-is
GitHub Actions
Inactive
The server still emits the workspace ID in the legacy `X-Databricks-Org-Id` response header on `/api/2.0/preview/scim/v2/Me`, so reverts the response-side reads back to the old name. The request-side migration to `X-Databricks-Workspace-Id` is unchanged. 8 sites reverted on the same API (GET /api/2.0/preview/scim/v2/Me): - __init__.py: WorkspaceClient.get_workspace_id() docstring + the two response reads (response_headers arg + dict-key read) - mixins/files.py: FilesExt._resolve_workspace_id() storage-proxy probe - config.py: Config._workspace_id_from_jdbc_path() SQL HTTP path resolver - tests/test_files.py: two mock-response fixtures simulating the server Signed-off-by: Divyansh Vijayvergia <divyansh.vijayvergia@databricks.com>
Divyansh-db
temporarily deployed
to
test-trigger-is
GitHub Actions
Inactive
Divyansh-db
temporarily deployed
to
test-trigger-is
GitHub Actions
Inactive
Divyansh-db
temporarily deployed
to
test-trigger-is
GitHub Actions
Inactive
Divyansh-db
temporarily deployed
to
test-trigger-is
GitHub Actions
Inactive
Signed-off-by: Divyansh Vijayvergia <171924202+Divyansh-db@users.noreply.github.com>
Divyansh-db
temporarily deployed
to
test-trigger-is
GitHub Actions
Inactive
Divyansh-db
temporarily deployed
to
test-trigger-is
GitHub Actions
Inactive
Pure mechanical reformatting from running `make fmt` (ruff format) against files that still carried the prior black wrapping convention. Two patterns swapped: - `assert (cond), msg` with parens around the condition collapses to `assert cond, (msg)` with parens around the message - short `self._api.do(...)` calls that black had wrapped now fit on one line under ruff's 120-char limit 7 files touched, no behavioral changes. Signed-off-by: Divyansh Vijayvergia <divyansh.vijayvergia@databricks.com>
Divyansh-db
temporarily deployed
to
test-trigger-is
GitHub Actions
Inactive
Divyansh-db
temporarily deployed
to
test-trigger-is
GitHub Actions
Inactive
The prior reformat commit (db2a26f) was made with ruff 0.15.x, which applies a different `assert cond, (msg)` parenthesization than the pinned ruff 0.5.6 (which keeps `assert (cond), msg`). This commit reverts those five test files back to the layout that the pinned formatter produces, so `make lint` stays green. Signed-off-by: Divyansh Vijayvergia <divyansh.vijayvergia@databricks.com>
Divyansh-db
temporarily deployed
to
test-trigger-is
GitHub Actions
Inactive
Divyansh-db
temporarily deployed
to
test-trigger-is
GitHub Actions
Inactive
Rewrites the previous `test_no_workspace_id_header_on_regular_workspace` test (which probed an unrelated `/api/2.0/test` path via a raw ApiClient purely to assert header absence) into a direct test of `WorkspaceClient.get_workspace_id()` that exercises both sides of the header migration on the SCIM /Me probe: - Request side: when `Config.workspace_id` is empty, the SDK must NOT send the new `X-Databricks-Workspace-Id` header. - Response side: the SDK must read the workspace ID from the legacy `X-Databricks-Org-Id` response header, since the server-side migration hasn't happened yet. Mirrors the Go SDK's `TestCurrentWorkspaceIDOmitsWorkspaceIdHeaderWhenConfigMissingWorkspaceID` in `workspace_functions_test.go`. Signed-off-by: Divyansh Vijayvergia <divyansh.vijayvergia@databricks.com>
Divyansh-db
temporarily deployed
to
test-trigger-is
GitHub Actions
Inactive
Divyansh-db
temporarily deployed
to
test-trigger-is
GitHub Actions
Inactive
Mirrors the Go SDK's TestAccUnifiedHostSendsWorkspaceIdHeader in internal/unified_host_test.go. Probes /api/2.0/preview/scim/v2/Me against a real unified (SPOG) host with a transport-level capture adapter mounted on the SDK's session and asserts three things at the wire: 1. SDK sends X-Databricks-Workspace-Id on workspace-scoped requests 2. SDK does not send the legacy X-Databricks-Org-Id request header 3. Server still echoes X-Databricks-Org-Id on the response (server side has not migrated, which is why get_workspace_id() still reads the legacy name) Requires UNIFIED_HOST, THIS_WORKSPACE_ID, TEST_ACCOUNT_ID, TEST_DATABRICKS_CLIENT_ID, TEST_DATABRICKS_CLIENT_SECRET in the ucws isolated env (consistent with the other test_spog_workspace_* tests in this file). Signed-off-by: Divyansh Vijayvergia <divyansh.vijayvergia@databricks.com>
Divyansh-db
temporarily deployed
to
test-trigger-is
GitHub Actions
Inactive
Divyansh-db
temporarily deployed
to
test-trigger-is
GitHub Actions
Inactive
Signed-off-by: Divyansh Vijayvergia <171924202+Divyansh-db@users.noreply.github.com>
Divyansh-db
temporarily deployed
to
test-trigger-is
GitHub Actions
Inactive
Divyansh-db
temporarily deployed
to
test-trigger-is
GitHub Actions
Inactive
`_CaptureHeadersAdapter` was copying request/response headers into a plain `dict()`, which preserves the original wire casing as keys. The Azure unified host (HTTP/2) emits headers in lowercase, so the assertion `"X-Databricks-Org-Id" in transport.resp_headers` failed even though the server was sending the header — its key was `x-databricks-org-id`. Switch to `requests.structures.CaseInsensitiveDict` for both bags so assertions match what the production SDK code does (e.g. `_base_client.py` reads `response.headers.get(...)` on a `CaseInsensitiveDict`, which is why the SDK itself was never affected). Signed-off-by: Divyansh Vijayvergia <divyansh.vijayvergia@databricks.com>
Signed-off-by: Divyansh Vijayvergia <171924202+Divyansh-db@users.noreply.github.com>
Divyansh-db
temporarily deployed
to
test-trigger-is
GitHub Actions
Inactive
|
If integration tests don't run automatically, an authorized user can run them manually by following the instructions below: Trigger: Inputs:
Checks will be approved automatically on success. |
Divyansh-db
temporarily deployed
to
test-trigger-is
GitHub Actions
Inactive
hectorcast-db
approved these changes
May 28, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Workspace-scoped API requests now identify the target workspace via the
X-Databricks-Workspace-Idheader instead ofX-Databricks-Org-Id. The value source is unchanged: it still comes fromConfig.workspace_id(or theDATABRICKS_WORKSPACE_IDenvironment variable). Response-side reads of/api/2.0/preview/scim/v2/Mecontinue to read the workspace ID from the legacyX-Databricks-Org-Idresponse header — this PR is outbound-only.Why
On unified Databricks hosts that serve multiple workspaces, the SDK sends a routing header on every workspace-scoped HTTP request so the gateway can dispatch it to the correct workspace. Until now that header was
X-Databricks-Org-Id.The Databricks platform is consolidating workspace addressing onto a single header,
X-Databricks-Workspace-Id, which accepts a broader range of workspace identifier formats and replaces the older single-purpose channels going forward. The previous header continues to be accepted by the platform, so older SDK versions keep working, but new SDK versions should use the new name.This PR is the Python SDK's part of that migration.
What changed
Interface changes
None.
Config.workspace_idkeeps the same field name, type (str), and environment variable (DATABRICKS_WORKSPACE_ID). The accompanying comment is widened to note that the value may now be either a classic numeric workspace ID or another workspace identifier format that the server understands.Behavioral changes
X-Databricks-Org-Id. They now sendX-Databricks-Workspace-Id(with the same value, gated onConfig.workspace_idbeing non-empty). Account-scoped requests are unaffected.X-Databricks-Org-Idresponse header on/api/2.0/preview/scim/v2/Me, soWorkspaceClient.get_workspace_id(), the storage-proxy workspace probe inFilesExt._resolve_workspace_id(), and the SQL HTTP path resolver inConfig._workspace_id_from_jdbc_path()all continue to read the old header name.Internal changes
databricks/sdk/service/*.py(including the newly-addedbundle.py),databricks/sdk/__init__.py, and the generated test fixtures undertests/databricks/sdk/service/from the corresponding generator template change. Pure mechanical literal swap on workspace-scoped request emission.databricks/sdk/mixins/workspace.py(upload/download),databricks/sdk/mixins/sharing.py(list shares).Config.workspace_idto note the new accepted formats.Test changes
tests/test_config.pyso the names reflect the new request-side header (test_*_org_id_header_*→test_*_workspace_id_header_*).WorkspaceClient.get_workspace_id():test_get_workspace_id_reads_org_id_response_header_when_config_missing_workspace_idnow mocks/api/2.0/preview/scim/v2/Mereturning anX-Databricks-Org-Idresponse header and asserts both (a) the SDK does not send the new request header whenConfig.workspace_idis empty, and (b) the SDK correctly parses the legacy response header.test_spog_unified_host_sends_workspace_id_headerintests/integration/test_unified_profile.pythat mounts a transport-level capture adapter on the SDK'srequests.Sessionand probes/api/2.0/preview/scim/v2/Meagainst a real unified host, asserting (1) the request carriesX-Databricks-Workspace-Id, (2) the request does not carry the legacyX-Databricks-Org-Id, and (3) the response still echoesX-Databricks-Org-Id. Mirrors the Go SDK's equivalent acceptance test.How is this tested?
make testpasses locally (2073 passed, 3 skipped).tests/test_config.pycover request-header emission on workspace-scoped vs account-scoped calls on unified hosts.test_get_workspace_id_reads_org_id_response_header_when_config_missing_workspace_idis a direct unit test onWorkspaceClient.get_workspace_id()covering both directions (no new request header + legacy response header parsed correctly) on the SCIM /Me probe./Meresponse-parsing paths inmixins/files.pyare exercised bytests/test_files.py(mock fixtures continue to emitX-Databricks-Org-Id, matching the unchanged response-side behavior).tests/integration/test_unified_profile.py::test_spog_unified_host_sends_workspace_id_headerverifies wire-level header behavior against a real unified host (runs in the OAuth M2M-equipped matrices).databricks/sdk/service/*.pyis exclusively the header literal swap; no collateral changes.