use same timeout on all calls

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "vectorize"
-version = "0.11.0"
+version = "0.11.1"
 edition = "2021"
 publish = false
 

Trunk.toml

@@ -6,7 +6,7 @@ description = "The simplest way to orchestrate vector search on Postgres."
 homepage = "https://github.com/tembo-io/pg_vectorize"
 documentation = "https://github.com/tembo-io/pg_vectorize"
 categories = ["orchestration", "machine_learning"]
-version = "0.11.0"
+version = "0.11.1"
 
 [build]
 postgres_version = "15"

src/transformers/http_handler.rs

@@ -94,11 +94,12 @@ pub async fn get_model_info(
 ) -> Result<TransformerMetadata> {
     let svc_url = get_generic_svc_url()?;
     let info_url = svc_url.replace("/embeddings", "/info");
+    let timeout = EMBEDDING_REQ_TIMEOUT_SEC.get();
     let client = reqwest::Client::new();
     let mut req = client
         .get(info_url)
         .query(&[("model_name", model_name)])
-        .timeout(std::time::Duration::from_secs(5)); // model info must always be fast
+        .timeout(std::time::Duration::from_secs(timeout as u64));
     if let Some(key) = api_key {
         req = req.header("Authorization", format!("Bearer {}", key));
     }

src/transformers/openai.rs

@@ -4,7 +4,7 @@ use anyhow::Result;
 
 use crate::{
     executor::VectorizeMeta,
-    guc::OPENAI_KEY,
+    guc::{EMBEDDING_REQ_TIMEOUT_SEC, OPENAI_KEY},
     transformers::{
         http_handler::handle_response,
         types::{EmbeddingPayload, EmbeddingRequest, Inputs},
@@ -72,6 +72,7 @@ pub fn trim_inputs(inputs: &[Inputs]) -> Vec<String> {
 
 pub fn validate_api_key(key: &str) -> Result<()> {
     let client = reqwest::Client::new();
+    let timeout = EMBEDDING_REQ_TIMEOUT_SEC.get();
     let runtime = tokio::runtime::Builder::new_current_thread()
         .enable_io()
         .enable_time()
@@ -82,7 +83,7 @@ pub fn validate_api_key(key: &str) -> Result<()> {
             .get("https://api.openai.com/v1/models")
             .header("Content-Type", "application/json")
             .header("Authorization", format!("Bearer {}", key))
-            .timeout(std::time::Duration::from_secs(5)) // api validation should be fast
+            .timeout(std::time::Duration::from_secs(timeout as u64))
             .send()
             .await
             .unwrap_or_else(|e| error!("failed to make Open AI key validation call: {}", e));