about summary refs log tree commit diff
path: root/docs
diff options
context:
space:
mode:
Diffstat (limited to 'docs')
-rw-r--r--docs/www/ai-instruction.md27
1 files changed, 25 insertions, 2 deletions
diff --git a/docs/www/ai-instruction.md b/docs/www/ai-instruction.md
index d38e207..794167b 100644
--- a/docs/www/ai-instruction.md
+++ b/docs/www/ai-instruction.md
@@ -123,9 +123,19 @@ Two routes are safe to call first and need no token:
 
 ```sh
 curl -s "$API/openapi.json"   # every endpoint, parameter, schema
-curl -s "$API/version"        # build + git revision
+curl -s "$API/version"        # build, git revision, and whether the database answers
 ```
 
+`/version` is worth reading when something fails everywhere at once. It
+returns `git` (the commit the running binary was built from; a `-dirty`
+suffix means that build had uncommitted changes), `dirty`,
+`source_digest` (which identifies the build when a commit hash cannot),
+`database` (`reachable` or `unreachable`) and `date` (the *commit's* date,
+not the build time). If `database` is `unreachable`, every data endpoint
+will fail with `503 db_unavailable` or `500 database_error` and there is
+nothing you can do but report it — do not retry in a loop and do not
+conclude the resource does not exist.
+
 Every response is one envelope:
 
 | member | meaning |
@@ -409,7 +419,11 @@ A mapping's response ends with what to do next:
 
 **Covariates.** Add `covariates` to fit the study's own metadata — sex,
 diet, batch — alongside the trait. The easiest form names a case
-attribute (§3), and the server codes it:
+attribute (§3), and the server codes it. It is also the smallest: nothing
+is posted, where an inline covariate sends a value per individual — for a
+panel of thousands that is the difference between a few hundred bytes and
+a multi-megabyte body, and bodies are capped at 64 MiB (`413`, in plain
+text, if you pass it):
 
 ```json
 "covariates": [{"case_attribute": "Sex"}]
@@ -522,6 +536,7 @@ curl -s -H "$AUTH" "$API/groups/BXD/phewas?chr=9&mb=29.940961&window=10000"
 | `window` | **half-width in base pairs, centred**: `10000` is 10 kb up *and* downstream (the default) |
 | `genotype_version` | whose marker table defines the window; omit for the group's current one |
 
+
 ### What comes back
 
 `data` has `group`, `genotype_version`, `centre` (`locus`, `chr`, `mb`),
@@ -594,9 +609,17 @@ Errors are RFC 7807-style JSON (`type`, `title`, `status`, `code`,
 | 403 | `forbidden` | the data is not public: do not retry |
 | 404 | `not_found` | check the name; `next` often lists the valid values |
 | 422 | `invalid_query`, `invalid_lmm` | fix the query or parameter; do not retry |
+| 409 | `run_id_in_use` | the `?id=` you chose is already in use: pick another, or poll `/compute/{rqtl,gemma}/status/{id}` |
+| 413 | — (plain text, **not** JSON) | the request body is over the 64 MiB limit: send fewer values, or none at all (§4a's `case_attribute`) |
 | 429 | — | back off, honour `Retry-After` (not currently emitted, but written into the API's contract) |
+| 500 | `database_error` | if the `detail` mentions a connection (`Broken pipe`, `Connection reset`, `server has gone away`) it is transient: retry **once**, then report; otherwise treat it as a bug and report it with the `request_id` |
 | 502 / 503 | `backend_unreachable`, `db_unavailable` | transient: retry with exponential backoff, max ~5 tries |
 
+Two things about the error shape itself: a **413 comes back as plain
+text**, not the JSON error object (`Failed to buffer the request body:
+length limit exceeded`) — the body limit is enforced before any handler
+runs — and errors carry `took_ms` as well, like a success's `meta`.
+
 Quote `request_id` when reporting a problem — it is in the body, in the
 `x-request-id` header, and in the server's access log.