docs(deploy,configuration): fix commands that fail on paste, wrong parameter names, and missing reference - #497
Merged
Conversation
…rameter names, and missing reference Three things break when copy-pasted: the EKS helm upgrade ended on a dangling line continuation so the shell hangs, a second one was found in the 1.25 migration guide, the backup module was named gcs-backup when the module id is backup-gcs, and the MUVERA parameter was documented as repetition when the real name is repetitions, contradicted by the page's own tested sample. Undocumented reference content operators can hit: six environment variables (the hybrid result floor, the three CORS settings, and the three backup chunking overrides) with their defaults and the releases they arrived in, and the six reserved property-name suffixes that cause a hard validation error with nothing in the docs to explain it. The SQ compression page imported TypeScript code and rendered none of it. Two sections now have TypeScript tabs; the third has no such snippet, and since all three tab sets share a group id, a reader who picked TypeScript would silently have been served Python there, so that section says so explicitly. The SQ options source turned out to be a Binary Quantization copy-paste, and both SQ sources passed a cache option that the TypeScript client's sq helper does not accept, so neither would have compiled. Also: HNSW_STARTUP_WAIT_FOR_VECTOR_CACHE is deprecated and was not marked so while its neighbour was; the combined module example used an environment variable core no longer reads; embedded XDG paths are used verbatim rather than having a subdirectory appended; two Go tabs were tagged as bash and a curl tab as Python; language-agnostic prose used Python-only parameter spellings; the tenant status vocabulary was the deprecated HOT/COLD; the MCP hybrid tool's filters argument was undocumented; the spellcheck module linked to the NER page; client versions moved out of prose into install commands; and HFresh is now mentioned on the compression guide, since it is the default for free clusters.
![orca-security-eu[bot]](https://avatars.githubusercontent.com/in/276244?s=60&v=4){kind=small}
There was a problem hiding this comment.
Orca Security Scan Summary
| Status | Check | Issues by priority | |
|---|---|---|---|
 Passed |
Secrets |  0  0  0  0 |
View in Orca |
Ivan asked for these two files to be left as they were. Restores docs/deploy/migration/weaviate-1-30.md and weaviate-1-25.md to their pre-review state. The findings behind them stay tracked: the removed REST endpoints (deploy-rest-004, deploy-rest-005) and the observation that the REINDEX_MAP_TO_BLOCKMAX workflow is inert on v1.38.0+ (board df1f6185).
Restores the two migration guides to the base branch's state so this PR does not touch them. They are reverted to their pre-review state on the base branch instead.
Contributor
There was a problem hiding this comment.
Pull request overview
Corrects deployment commands, configuration references, API terminology, and code samples across Weaviate documentation.
Changes:
- Fixes invalid commands, module names, parameter names, and syntax highlighting.
- Expands environment-variable, MCP, compression, and property-name references.
- Repairs TypeScript SQ examples and clarifies client requirements.
Reviewed changes
Copilot reviewed 23 out of 23 changed files in this pull request and generated 6 comments.
Show a summary per file
| File | Description |
|---|---|
docs/weaviate/modules/index.md |
Fixes spellcheck link. |
docs/weaviate/manage-collections/multi-tenancy.mdx |
Corrects code languages. |
docs/weaviate/manage-collections/inverted-index.mdx |
Clarifies parameter spellings. |
docs/weaviate/configuration/modules.md |
Updates module enablement guidance. |
docs/weaviate/configuration/mcp-server.mdx |
Expands MCP tool documentation. |
docs/weaviate/configuration/compression/sq-compression.md |
Adds TypeScript SQ tabs. |
docs/weaviate/configuration/compression/multi-vectors.md |
Corrects MUVERA parameters. |
docs/weaviate/configuration/authz-authn.md |
Revises MCP reference wording. |
docs/weaviate/config-refs/collections.mdx |
Documents reserved suffixes. |
docs/weaviate/concepts/modules.md |
Corrects GCS module ID. |
docs/deploy/production/aws/hardening-eks.md |
Improves wording. |
docs/deploy/installation-guides/spcs-integration.mdx |
Repairs Python example. |
docs/deploy/installation-guides/embedded.md |
Clarifies XDG path behavior. |
docs/deploy/installation-guides/eks.md |
Fixes Helm command termination. |
docs/deploy/configuration/monitoring.md |
Refines monitoring prose. |
docs/deploy/configuration/export.md |
Refines preview and tenant wording. |
docs/deploy/configuration/env-vars/runtime-config.md |
Clarifies runtime MCP settings. |
docs/deploy/configuration/env-vars/index.md |
Adds and updates environment variables. |
docs/deploy/configuration/backups.md |
Cross-links chunking settings. |
docs/cloud/manage-collections/enable-compression.mdx |
Documents HFresh compression behavior. |
docs/cloud/embeddings/quickstart.mdx |
Adds client installation guidance. |
_includes/code/howto/configure-sq/sq-compression.options-v3.ts |
Replaces incorrect BQ example with SQ. |
_includes/code/howto/configure-sq/sq-compression-v3.ts |
Corrects SQ options and assertions. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
…requirements Remove the backup chunking work from this PR. The BACKUP_CHUNK_TARGET_SIZE, BACKUP_MIN_CHUNK_SIZE and BACKUP_SPLIT_FILE_SIZE rows are deleted from the environment variables page, and the ChunkSize cell in backups.md is restored to its original text, since the cross-link it gained pointed at those now removed rows. backups.md therefore drops out of this PR entirely. The other additions to the environment variables page (QUERY_HYBRID_MAXIMUM_RESULTS, the three CORS_ALLOW_* rows and the HNSW_STARTUP_WAIT_FOR_VECTOR_CACHE deprecation note) are unchanged. Rewrite the Requirements section of the Weaviate Embeddings quickstart so it lists requirements only. Weaviate Cloud no longer runs anything below 1.32, so the version qualifier on the free cluster bullet is dead weight and is dropped. The hand rolled client install tabs are replaced with the shared clients.install.new.mdx include that the official quickstart uses, which also removes two pinned client version numbers from prose. The page already imported CodeClientInstall for step 1.2, so that duplicate import is removed and both sections now render the same shared include. The Go tab being deleted carried a caveat that appears nowhere else in the docs, so it is kept as a sentence below the install tabs: the Go client does not support Weaviate Embeddings directly and needs the headers passed manually. The old wording promised the headers were shown below, which they never were, so that false pointer is gone.
There was a problem hiding this comment.
Orca Security Scan Summary
| Status | Check | Issues by priority | |
|---|---|---|---|
| Passed |
Infrastructure as Code | 0 0 0 0 |
View in Orca |
| Passed |
SAST | 0 0 0 0 |
View in Orca |
| Passed |
Secrets | 0 0 0 0 |
View in Orca |
| Passed |
Vulnerabilities | 0 0 0 0 |
View in Orca |
… SQ on an existing collection The "Enable compression for existing collection" section carried a note saying a TypeScript example was not available for this operation. That is not true: the client exposes the reconfigure quantizer on the HNSW update type, so the operation is fully supported. - Add an `UpdateSchema` block to the SQ TypeScript include that enables SQ on an existing collection via `collection.config.update()` with `reconfigure.vectors.update()`, mirroring the Python `UpdateSchema` block on the same page (same collection, same rescore limit). - Wire a `ts` tab into the section, matching the ordering, labels and `groupId` of the other two tab sets on the page. - Drop the note. Verified: the rendered tab type-checks against the installed client (`tsc --noEmit`, exit 0) and runs against a local Weaviate, turning an uncompressed HNSW index into an SQ-compressed one. `yarn build-dev` exits 0 with no change to the broken link or anchor counts. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01DzkwXqkMC6agu8oGNswh2b
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
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.
Medium tier of the docs deep review, group 4 of 4. 17 findings, 24 files.
Stacked on #493. Base retargets to
mainwhen that merges.Four things break when copy-pasted. The EKS
helm upgradeended on a dangling line continuation, so the shell hangs waiting for input; a sweep ofdocs/deploy/**found a second one in the 1.25 migration guide. The backup module was namedgcs-backupwhen the id isbackup-gcs, soENABLE_MODULESfails with module-not-found. The MUVERA parameter was documented asrepetitionwhen the real name isrepetitions, contradicted by the page's own tested sample.Reference content operators can hit but could not look up. Six environment variables with their defaults and the releases they arrived in: the hybrid result floor, the three CORS settings, and the three backup chunking overrides (which are distinct from the per-request
ChunkSizealready documented, now cross-linked). Plus the reserved property-name suffixes that cause a hard validation error with nothing in the docs to explain it. The review listed five suffixes; core defines six —__meta_countwas missing.A page that imported TypeScript and rendered none of it. Two SQ compression sections now have TypeScript tabs. The third has no such snippet, so that section now says so explicitly.
Correction to an earlier version of this description: it claimed a missing tab in a shared group id causes Docusaurus to silently serve the first tab's language. That is true of stock Docusaurus, but this repo routes
className="code"to its ownCodeDropdownTabs(src/theme/Tabs/index.js:466), which keeps the reader's language and renders a visible "not yet available here" placeholder. No reader was being silently served the wrong language. The added note is still accurate, just redundant with that placeholder. Two real bugs surfaced while doing it: the SQ options source was an entire Binary Quantization copy-paste (quantizer.bq(...), asserting"bq"), and both SQ sources passed acacheoption that the TypeScript client'ssqhelper does not accept, so neither would have compiled.Also:
HNSW_STARTUP_WAIT_FOR_VECTOR_CACHEis deprecated and was unmarked while its neighbour was marked; the combined module example used a variable core no longer reads; embedded XDG paths are used verbatim rather than having a subdirectory appended; two Go tabs were taggedbashand a curl tabpy; language-agnostic prose used Python-only parameter spellings; tenant status used the deprecated HOT/COLD vocabulary; the MCP hybrid tool'sfiltersargument was entirely undocumented; the spellcheck module linked to the NER page; client versions moved out of prose into install commands; and HFresh is now mentioned on the compression guide, since it is the default index for free clusters.Every version claim was verified by reading the value at the boundary release tags, checking the patch before and after each claimed release, rather than by
git tag --contains, which under-reports cherry-picked backports.Verified:
yarn build-devexit 0; broken links and anchors identical to the base; the corrected TypeScript now type-checks where the previous code did not.