docs: describe the Postgres backend instead of Convex - #3150
Open
Israeltheminer wants to merge 1 commit into
Open
docs: describe the Postgres backend instead of Convex#3150Israeltheminer wants to merge 1 commit into
Israeltheminer wants to merge 1 commit into
Conversation
The port replaced the Convex backend and the tree retirement deleted it, but the docs still told operators about a tale-convex container, Convex storage, Convex tables, and paths under services/platform/convex/. The self-hosted pages are the sharpest of these because they are procedures. Backups and restore listed rag-data and crawler-data, which no longer exist, and did not say that uploaded file blobs sit OUTSIDE a snapshot — verified against SNAPSHOT_VOLUMES, which is db-data, convex-data, caddy-data and caddy-config, with object-store-data deliberately absent. An operator following the old page would have restored an instance with no files. The container overview said eight containers, then ten; compose declares eleven. bgutil-provider was the one nobody had documented, and it is the only image in the stack Tale does not build. The count is now eleven in all three locales, with that container described. convex-data keeps its name and now says why: the volume predates the retirement and renaming it would have forced every operator to migrate one. All three locales move together. Each replacement claim was read out of the code or the compose file first; nothing was inferred from the old text.
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The documentation still told operators to run a
tale-convexcontainer, store blobs in Convex storage, and look for code underservices/platform/convex/. None of that exists. 63 pages across all three locales.Refs #3142.
Why
#3107 replaced the Convex backend with Node, Hono and Postgres; #3125 deleted the tree. The docs were not swept.
The self-hosted pages matter most because they are procedures someone follows during an incident.
Backups and restore was the dangerous one. It listed
rag-dataandcrawler-dataas snapshot volumes — neither exists — and never said that uploaded file blobs sit outside a snapshot. An operator following it would restore an instance with no files and no warning that anything was missing.Verified against
SNAPSHOT_VOLUMESintools/cli/src/lib/backup/constants.ts:db-data,convex-data,caddy-data,caddy-config.object-store-datais deliberately absent. The page now says so, before the procedure rather than after it.The container overview was wrong twice. It said eight containers; a first pass said ten;
compose.ymldeclares eleven. The one nobody had documented isbgutil-provider— and it is the only image in the stack Tale does not build, which is exactly the kind of thing an operator reviewing their supply chain wants named. All three locales now say eleven and describe it.convex-datakeeps its name, and the page now explains why: the volume predates the retirement, and renaming it would have forced every operator to migrate a volume for cosmetics. The code comment says the same thing.Also corrected: the status page's liveness probe (it described probing
${CONVEX_URL}/version, a service that no longer exists), the WebDAV page's blob storage and lock table, the API reference's content endpoint, and the data-residency request path.What is not here
services/docs/app/content/frontmatter.jsonis a generated manifest whose key order is not deterministic — regenerating it churns 2,196 lines with zero net change, and a structural test requires it to match the pages. So its diff is hand-limited to the 6 entries that actually changed, keeping the committed key order, rather than 4,392 lines of noise that would conflict with every other docs branch.That non-determinism is a separate repo defect and is not fixed here.
Method
Every replacement claim was read out of the code, the compose file, or
backend/MIGRATION.mdbefore being written. No env var, container name, path or command is invented. English follows the repo's docs contract; German and French are written natively per the translation contract rather than rendered word-for-word from the English.Gate:
oxfmt --checkclean, docs lint exit 0, docs structural suite 194/194.