nidx-binding 6.3.5.post220__tar.gz → 6.3.5.post232__tar.gz

{nidx_binding-6.3.5.post220 → nidx_binding-6.3.5.post232}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nidx_binding
-Version: 6.3.5.post220
+Version: 6.3.5.post232
 Classifier: Programming Language :: Rust
 Classifier: Programming Language :: Python :: Implementation :: CPython
 Classifier: Programming Language :: Python :: Implementation :: PyPy

{nidx_binding-6.3.5.post220 → nidx_binding-6.3.5.post232}/nidx_protos/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "pdm.backend"
 
 [project]
 name = "nidx_protos"
-version = "6.3.5.post220"
+version = "6.3.5.post232"
 license = { text = "AGPL" }
 description = "Protobuf definitions for nucliadb/nidx"
 authors = [ { name = "Nuclia", email = "nucliadb@nuclia.com" }]

{nidx_binding-6.3.5.post220 → nidx_binding-6.3.5.post232}/nidx_text/src/reader.rs

@@ -35,12 +35,14 @@ use nidx_protos::{
 };
 use nidx_types::prefilter::{FieldId, PrefilterResult};
 use tantivy::collector::{Collector, Count, FacetCollector, FacetCounts, SegmentCollector, TopDocs};
-use tantivy::columnar::Column;
+use tantivy::columnar::{BytesColumn, Column};
+use tantivy::fastfield::FacetReader;
 use tantivy::query::{AllQuery, BooleanQuery, Query, QueryParser, TermQuery};
 use tantivy::schema::Value;
 use tantivy::schema::*;
 use tantivy::{DocAddress, Index, IndexReader, Searcher};
 use tracing::*;
+use uuid::Uuid;
 
 fn facet_count(facet: &str, facets_count: &FacetCounts) -> Vec<FacetResult> {
     facets_count
@@ -89,13 +91,85 @@ impl Debug for TextReaderService {
 }
 
 struct FieldUuidSegmentCollector {
-    encoded_field_id_reader: Column,
+    uuid_reader: BytesColumn,
+    field_reader: FacetReader,
     results: Vec<FieldId>,
 }
 
 impl SegmentCollector for FieldUuidSegmentCollector {
     type Fruit = Vec<FieldId>;
 
+    fn collect(&mut self, doc: tantivy::DocId, _score: tantivy::Score) {
+        let uuid_ord = self.uuid_reader.term_ords(doc).next().unwrap();
+        let mut uuid_bytes = Vec::new();
+        self.uuid_reader.ord_to_bytes(uuid_ord, &mut uuid_bytes).unwrap();
+
+        let mut facet_ords = self.field_reader.facet_ords(doc);
+        let mut facet = Facet::root();
+        self.field_reader
+            .facet_from_ord(facet_ords.next().unwrap(), &mut facet)
+            .expect("field facet not found");
+
+        if let Ok(resource_id) = Uuid::parse_str(std::str::from_utf8(&uuid_bytes).unwrap()) {
+            self.results.push(FieldId {
+                resource_id,
+                field_id: facet.to_path_string(),
+            });
+        }
+    }
+
+    fn harvest(self) -> Self::Fruit {
+        self.results
+    }
+}
+
+struct FieldUuidCollector;
+
+impl Collector for FieldUuidCollector {
+    type Fruit = Vec<FieldId>;
+
+    type Child = FieldUuidSegmentCollector;
+
+    fn for_segment(
+        &self,
+        _segment_local_id: tantivy::SegmentOrdinal,
+        segment: &tantivy::SegmentReader,
+    ) -> tantivy::Result<Self::Child> {
+        let uuid_reader = segment.fast_fields().bytes("uuid")?.unwrap();
+        let field_reader = segment.facet_reader("field")?;
+        Ok(FieldUuidSegmentCollector {
+            uuid_reader,
+            field_reader,
+            results: vec![],
+        })
+    }
+
+    fn requires_scoring(&self) -> bool {
+        false
+    }
+
+    fn merge_fruits(
+        &self,
+        segment_fruits: Vec<<Self::Child as tantivy::collector::SegmentCollector>::Fruit>,
+    ) -> tantivy::Result<Self::Fruit> {
+        Ok(segment_fruits
+            .into_iter()
+            .reduce(|mut a, mut b| {
+                a.append(&mut b);
+                a
+            })
+            .unwrap_or_default())
+    }
+}
+
+struct FieldUuidSegmentCollectorV2 {
+    encoded_field_id_reader: Column,
+    results: Vec<FieldId>,
+}
+
+impl SegmentCollector for FieldUuidSegmentCollectorV2 {
+    type Fruit = Vec<FieldId>;
+
     fn collect(&mut self, doc: tantivy::DocId, _score: tantivy::Score) {
         let mut data = Vec::new();
         self.encoded_field_id_reader.fill_vals(doc, &mut data);
@@ -112,12 +186,12 @@ impl SegmentCollector for FieldUuidSegmentCollector {
     }
 }
 
-struct FieldUuidCollector;
+struct FieldUuidCollectorV2;
 
-impl Collector for FieldUuidCollector {
+impl Collector for FieldUuidCollectorV2 {
     type Fruit = Vec<FieldId>;
 
-    type Child = FieldUuidSegmentCollector;
+    type Child = FieldUuidSegmentCollectorV2;
 
     fn for_segment(
         &self,
@@ -125,7 +199,7 @@ impl Collector for FieldUuidCollector {
         segment: &tantivy::SegmentReader,
     ) -> tantivy::Result<Self::Child> {
         let encoded_field_id_reader = segment.fast_fields().u64("encoded_field_id")?;
-        Ok(FieldUuidSegmentCollector {
+        Ok(FieldUuidSegmentCollectorV2 {
             encoded_field_id_reader,
             results: vec![],
         })
@@ -189,7 +263,11 @@ impl TextReaderService {
 
         let prefilter_query: Box<dyn Query> = Box::new(BooleanQuery::intersection(subqueries));
         let searcher = self.reader.searcher();
-        let docs_fulfilled = searcher.search(&prefilter_query, &FieldUuidCollector)?;
+        let docs_fulfilled = if schema.encoded_field_id.is_some() {
+            searcher.search(&prefilter_query, &FieldUuidCollectorV2)?
+        } else {
+            searcher.search(&prefilter_query, &FieldUuidCollector)?
+        };
         // If none of the fields match the pre-filter, thats all the query planner needs to know.
         if docs_fulfilled.is_empty() {
             return Ok(PrefilterResult::None);

{nidx_binding-6.3.5.post220 → nidx_binding-6.3.5.post232}/nidx_text/src/resource_indexer.rs

@@ -80,8 +80,10 @@ pub fn index_document(
         field_doc.add_facet(schema.field, facet_field);
         field_doc.add_text(schema.text, &text_info.text);
 
-        for d in encode_field_id(resource_uuid, &format!("/{field}")) {
-            field_doc.add_u64(schema.encoded_field_id, d);
+        if let Some(encoded_field_id) = schema.encoded_field_id {
+            for d in encode_field_id(resource_uuid, &format!("/{field}")) {
+                field_doc.add_u64(encoded_field_id, d);
+            }
         }
 
         if let Some(encoded_field_id_bytes) = schema.encoded_field_id_bytes {

{nidx_binding-6.3.5.post220 → nidx_binding-6.3.5.post232}/nidx_text/src/schema.rs

@@ -45,7 +45,7 @@ pub struct TextSchema {
     pub groups_public: Field,
     pub groups_with_access: Field,
 
-    pub encoded_field_id: Field,
+    pub encoded_field_id: Option<Field>,
     pub encoded_field_id_bytes: Option<Field>,
 }
 
@@ -85,13 +85,18 @@ impl TextSchema {
             // v4: Field ID encoded as array of u64 for faster retrieval during prefilter
             // Using a bytes field is slow due to tantivy's implementation being slow with many unique values.
             // A better implementation is tracked in https://github.com/quickwit-oss/tantivy/issues/2090
-            sb.add_u64_field("encoded_field_id", FAST)
-        } else {
+            Some(sb.add_u64_field("encoded_field_id", FAST))
+        } else if version == 3 {
             // v3: Field ID encoded as array of u64 for faster retrieval during prefilter
             // Using a bytes field is slow due to tantivy's implementation being slow with many unique values.
             // A better implementation is tracked in https://github.com/quickwit-oss/tantivy/issues/2090
             // The INDEXED flag is unnecessary here, and this is backwards compatible code until we remove it in v4.
-            sb.add_u64_field("encoded_field_id", FAST | INDEXED)
+            Some(sb.add_u64_field("encoded_field_id", FAST | INDEXED))
+        } else if version == 2 {
+            // v2 did not have the field indexed
+            Some(sb.add_u64_field("encoded_field_id", FAST))
+        } else {
+            None
         };
 
         let encoded_field_id_bytes = if version >= 4 {

{nidx_binding-6.3.5.post220 → nidx_binding-6.3.5.post232}/pyproject.toml

@@ -13,7 +13,7 @@ classifiers = [
     "Programming Language :: Python :: Implementation :: CPython",
     "Programming Language :: Python :: Implementation :: PyPy",
 ]
-version = "6.3.5.post220"
+version = "6.3.5.post232"
 
 [project.urls]
 Homepage = "https://nuclia.com"

{nidx_binding-6.3.5.post220 → nidx_binding-6.3.5.post232}/src/indexer.rs

@@ -42,10 +42,9 @@ use crate::errors::NidxError;
 use crate::grpc_server::GrpcServer;
 use crate::metrics::IndexKindLabels;
 use crate::metrics::OperationStatusLabels;
-use crate::metrics::indexer::INDEXING_COUNTER;
-use crate::metrics::indexer::PER_INDEX_INDEXING_TIME;
-use crate::metrics::indexer::TOTAL_INDEXING_TIME;
+use crate::metrics::indexer::*;
 use crate::segment_store::pack_and_upload;
+use crate::utilization_tracker::UtilizationTracker;
 use crate::{Settings, metadata::*};
 
 #[cfg(feature = "telemetry")]
@@ -138,7 +137,12 @@ pub async fn run_nats(settings: Settings, shutdown: CancellationToken) -> anyhow
         None => tempfile::env::temp_dir(),
     };
 
+    let utilization = UtilizationTracker::new(|busy, duration| {
+        INDEXING_BUSY.get_or_create(&busy.into()).inc_by(duration.as_secs_f64());
+    });
+
     while !shutdown.is_cancelled() {
+        utilization.idle().await;
         let sub_msg = tokio::select! {
             sub_msg = subscription.next() => sub_msg,
             _ = shutdown.cancelled() => { return Ok(()) }
@@ -164,6 +168,7 @@ pub async fn run_nats(settings: Settings, shutdown: CancellationToken) -> anyhow
             continue;
         }
 
+        utilization.busy().await;
         let span = info_span!("indexer_message", ?seq);
         let (msg, acker) = msg.split();
 

{nidx_binding-6.3.5.post220 → nidx_binding-6.3.5.post232}/src/lib.rs

@@ -29,6 +29,7 @@ pub mod searcher;
 mod segment_store;
 pub mod settings;
 pub mod tool;
+mod utilization_tracker;
 pub mod worker;
 
 #[cfg(feature = "telemetry")]

{nidx_binding-6.3.5.post220 → nidx_binding-6.3.5.post232}/src/metrics.rs

@@ -127,6 +127,31 @@ impl OperationStatusLabels {
     }
 }
 
+#[derive(Clone, Debug, PartialEq, Eq, Hash, EncodeLabelValue)]
+enum Utilization {
+    Free,
+    Busy,
+}
+
+#[derive(Clone, Debug, EncodeLabelSet, PartialEq, Eq, Hash)]
+pub struct UtilizationLabels {
+    status: Utilization,
+}
+
+impl From<bool> for UtilizationLabels {
+    fn from(busy: bool) -> Self {
+        if busy {
+            Self {
+                status: Utilization::Busy,
+            }
+        } else {
+            Self {
+                status: Utilization::Free,
+            }
+        }
+    }
+}
+
 pub mod common {
     metrics! {
         SPAN_DURATION: Family<Vec<(String, String)>, Histogram>{exponential_buckets(0.001, 2.0, 20)} as "span_duration_seconds" ("Duration of a tracing span"),
@@ -169,6 +194,7 @@ pub mod indexer {
         INDEXING_COUNTER: Family<OperationStatusLabels, Counter> as "indexer_message_count" ("Number of indexing operations per status"),
         TOTAL_INDEXING_TIME: Histogram[exponential_buckets(0.01, 2.0, 12)] as "total_indexing_time_seconds" ("Time it took to process an entire index message"),
         PER_INDEX_INDEXING_TIME: Family<IndexKindLabels, Histogram>{exponential_buckets(0.01, 2.0, 12)} as "indexing_time_seconds" ("Time it took to index a resource to an index"),
+        INDEXING_BUSY: Family<UtilizationLabels, Counter::<f64, AtomicU64>> as "indexer_utilization_seconds" ("Time spent in free/busy status"),
     }
 }
 
@@ -176,5 +202,6 @@ pub mod worker {
     metrics! {
         MERGE_COUNTER: Family<OperationStatusLabels, Counter> as "merge_job_count" ("Number of merge jobs per status"),
         PER_INDEX_MERGE_TIME: Family<IndexKindLabels, Histogram>{exponential_buckets(1.0, 2.0, 12)} as "merge_time_seconds" ("Time it took to run a merge job"),
+        WORKER_BUSY: Family<UtilizationLabels, Counter::<f64, AtomicU64>> as "merge_worker_utilization_seconds" ("Time spent in free/busy status"),
     }
 }

nidx_binding-6.3.5.post232/src/utilization_tracker.rs

@@ -0,0 +1,62 @@
+// Copyright (C) 2021 Bosutech XXI S.L.
+//
+// nucliadb is offered under the AGPL v3.0 and as commercial software.
+// For commercial licensing, contact us at info@nuclia.com.
+//
+// AGPL:
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program. If not, see <http://www.gnu.org/licenses/>.
+//
+
+use std::time::{Duration, Instant};
+
+use tokio::{sync::mpsc::*, time::sleep};
+
+pub struct UtilizationTracker {
+    channel: Sender<bool>,
+}
+
+impl UtilizationTracker {
+    pub fn new(publish: impl Fn(bool, Duration) + Send + 'static) -> Self {
+        let (tx, rx) = channel(10);
+        tokio::spawn(run(publish, rx));
+        Self { channel: tx }
+    }
+
+    pub async fn busy(&self) {
+        let _ = self.channel.send(true).await;
+    }
+
+    pub async fn idle(&self) {
+        let _ = self.channel.send(false).await;
+    }
+}
+
+async fn run(publish: impl Fn(bool, Duration), mut channel: Receiver<bool>) {
+    let mut busy = false;
+    let mut since = Instant::now();
+    loop {
+        tokio::select! {
+            rxed = channel.recv() => {
+                let Some(rxed) = rxed else { break };
+                publish(busy, since.elapsed());
+                busy = rxed;
+                since = Instant::now();
+            },
+            _ = sleep(Duration::from_secs(30)) => {
+                publish(busy, since.elapsed());
+                since = Instant::now();
+            }
+        }
+    }
+}

{nidx_binding-6.3.5.post220 → nidx_binding-6.3.5.post232}/src/worker.rs

@@ -39,11 +39,9 @@ use tracing::*;
 use crate::{
     NidxMetadata, Settings,
     metadata::{Deletion, Index, IndexKind, MergeJob, NewSegment, Segment},
-    metrics::{
-        IndexKindLabels, OperationStatusLabels,
-        worker::{MERGE_COUNTER, PER_INDEX_MERGE_TIME},
-    },
+    metrics::{IndexKindLabels, OperationStatusLabels, worker::*},
     segment_store::{download_segment, pack_and_upload},
+    utilization_tracker::UtilizationTracker,
 };
 
 pub async fn run(settings: Settings, shutdown: CancellationToken) -> anyhow::Result<()> {
@@ -55,9 +53,14 @@ pub async fn run(settings: Settings, shutdown: CancellationToken) -> anyhow::Res
         None => tempfile::env::temp_dir(),
     };
 
+    let utilization = UtilizationTracker::new(|busy, duration| {
+        WORKER_BUSY.get_or_create(&busy.into()).inc_by(duration.as_secs_f64());
+    });
+
     while !shutdown.is_cancelled() {
         let job = MergeJob::take(&meta.pool).await?;
         if let Some(job) = job {
+            utilization.busy().await;
             let span = info_span!("worker_job", ?job.id);
             info!(job.id, "Running job");
 
@@ -85,6 +88,7 @@ pub async fn run(settings: Settings, shutdown: CancellationToken) -> anyhow::Res
             // Stop keep alives
             keepalive.abort();
         } else {
+            utilization.idle().await;
             debug!("No jobs, waiting for more");
             tokio::select! {
                 _ = tokio::time::sleep(Duration::from_secs(5)) => {},