Merge branch 'main' into TEM-3229-2

.github/workflows/extension_ci.yml

@@ -135,6 +135,9 @@ jobs:
     if: github.event_name == 'release'
     name: trunk publish
     runs-on: ubuntu-22.04
+    strategy:
+      matrix:
+        pg-version: [14, 15, 16]
     steps:
       - uses: actions/checkout@v2
       - name: Install Rust stable toolchain
@@ -158,7 +161,7 @@ jobs:
          cargo install pg-trunk
       - name: trunk build
         working-directory: ./
-        run: ~/.cargo/bin/trunk build
+        run: ~/.cargo/bin/trunk build --pg-version ${{ matrix.pg-version }}
       - name: trunk publish
         working-directory: ./
         env:

src/guc.rs

@@ -9,6 +9,7 @@ pub static BATCH_SIZE: GucSetting<i32> = GucSetting::<i32>::new(10000);
 pub static NUM_BGW_PROC: GucSetting<i32> = GucSetting::<i32>::new(1);
 pub static EMBEDDING_SERVICE_HOST: GucSetting<Option<&CStr>> =
     GucSetting::<Option<&CStr>>::new(None);
+pub static EMBEDDING_REQ_TIMEOUT_SEC: GucSetting<i32> = GucSetting::<i32>::new(120);
 
 // initialize GUCs
 pub fn init_guc() {
@@ -56,6 +57,17 @@ pub fn init_guc() {
         GucContext::Suset,
         GucFlags::default(),
     );
+
+    GucRegistry::define_int_guc(
+        "vectorize.embedding_req_timeout_sec",
+        "Timeout, in seconds, for embedding transform requests",
+        "Number of seconds to wait for an embedding http request to complete. Default is 120 seconds.",
+        &EMBEDDING_REQ_TIMEOUT_SEC,
+        1,
+        1800,
+        GucContext::Suset,
+        GucFlags::default(),
+    );
 }
 
 // for handling of GUCs that can be error prone

src/search.rs

@@ -230,7 +230,14 @@ pub fn search(
     let schema = proj_params.schema;
     let table = proj_params.table;
 
-    let embeddings = transform(query, &project_meta.transformer, api_key);
+    let proj_api_key = match api_key {
+        // if api passed in the function call, use that
+        Some(k) => Some(k),
+        // if not, use the one from the project metadata
+        None => proj_params.api_key,
+    };
+
+    let embeddings = transform(query, &project_meta.transformer, proj_api_key);
 
     match project_meta.search_alg {
         types::SimilarityAlg::pgv_cosine_similarity => cosine_similarity_search(

src/transformers/generic.rs

@@ -6,6 +6,7 @@ use crate::{
     executor::VectorizeMeta,
     guc,
     transformers::types::{EmbeddingPayload, EmbeddingRequest, Inputs},
+    types,
 };
 
 use super::openai::trim_inputs;
@@ -63,12 +64,14 @@ pub fn prepare_generic_embedding_request(
         model: job_meta.transformer.to_string(),
     };
 
+    let job_params: types::JobParams = serde_json::from_value(job_meta.params)?;
+
     let svc_host = get_generic_svc_url().context("failed to get embedding service url from GUC")?;
 
     Ok(EmbeddingRequest {
         url: svc_host,
         payload,
-        api_key: None,
+        api_key: job_params.api_key,
     })
 }
 

src/transformers/http_handler.rs

@@ -2,6 +2,7 @@ use super::generic::get_generic_svc_url;
 use super::types::TransformerMetadata;
 use anyhow::Result;
 
+use crate::guc::EMBEDDING_REQ_TIMEOUT_SEC;
 use crate::transformers::types::{
     EmbeddingPayload, EmbeddingRequest, EmbeddingResponse, Inputs, PairedEmbeddings,
 };
@@ -32,8 +33,10 @@ pub async fn openai_embedding_request(request: EmbeddingRequest) -> Result<Vec<V
         request.payload.input.len()
     );
     let client = reqwest::Client::new();
+    let timeout = EMBEDDING_REQ_TIMEOUT_SEC.get();
     let mut req = client
         .post(request.url)
+        .timeout(std::time::Duration::from_secs(timeout as u64))
         .json::<EmbeddingPayload>(&request.payload)
         .header("Content-Type", "application/json");
     if let Some(key) = request.api_key {
@@ -92,7 +95,10 @@ pub async fn get_model_info(
     let svc_url = get_generic_svc_url()?;
     let info_url = svc_url.replace("/embeddings", "/info");
     let client = reqwest::Client::new();
-    let mut req = client.get(info_url).query(&[("model_name", model_name)]);
+    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
     if let Some(key) = api_key {
         req = req.header("Authorization", format!("Bearer {}", key));
     }

src/transformers/openai.rs

@@ -82,6 +82,7 @@ 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
             .send()
             .await
             .unwrap_or_else(|e| error!("failed to make Open AI key validation call: {}", e));

src/workers/mod.rs

@@ -35,10 +35,15 @@ pub async fn run_worker(
         msg.message.job_name
     );
     let job_success = execute_job(conn.clone(), msg).await;
-    let delete_it = if job_success.is_ok() {
-        true
-    } else {
-        read_ct > 2
+    let delete_it = match job_success {
+        Ok(_) => {
+            info!("pg-vectorize: job success");
+            true
+        }
+        Err(e) => {
+            warning!("pg-vectorize: job failed: {:?}", e);
+            read_ct > 2
+        }
     };
 
     // delete message from queue