diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 3b7ad05696d..fcf6af9382e 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -111,7 +111,7 @@ jobs:
         working-directory: ./quickwit
       - name: cargo nextest
         if: always() && steps.modified.outputs.rust_src == 'true'
-        run: cargo nextest run --features=postgres --retries 1
+        run: cargo nextest run --features=postgres --profile ci
         working-directory: ./quickwit
       - name: Install python packages
         if: always() && steps.modified.outputs.rust_src == 'true'
diff --git a/.github/workflows/coverage.yml b/.github/workflows/coverage.yml
index 3b1be41fa16..dd45a377897 100644
--- a/.github/workflows/coverage.yml
+++ b/.github/workflows/coverage.yml
@@ -190,9 +190,9 @@ jobs:
       - name: Generate code coverage
         run: |
           cargo llvm-cov clean --workspace
-          cargo llvm-cov nextest --no-report --test failpoints --features fail/failpoints --retries 4
+          cargo llvm-cov nextest --no-report --profile ci --test failpoints --features fail/failpoints
           # increase stack size for test_all_with_s3_localstack_cli, see quickwit#4963
-          RUST_MIN_STACK=67108864 CARGO_BUILD_JOBS=4 cargo llvm-cov nextest --no-report --all-features --retries 4
+          RUST_MIN_STACK=67108864 CARGO_BUILD_JOBS=4 cargo llvm-cov nextest --no-report --all-features --profile ci
           cargo llvm-cov report --lcov --output-path lcov.info
         working-directory: ./quickwit
 
diff --git a/AGENTS.md b/AGENTS.md
index f9d8206fd6b..a92db39e62b 100644
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -140,7 +140,7 @@ Run `cargo check` after editing `Cargo.toml` to update `Cargo.lock`.
 ### Testing
 
 - Single crate test: `cargo nextest run -p quickwit-search my_test_name`
-- `make test-all` — starts Docker services (LocalStack S3, PostgreSQL, Pub/Sub emulator) and runs the full test suite with `cargo nextest run --all-features --retries 5`.
+- `make test-all` — starts Docker services (LocalStack S3, PostgreSQL, Pub/Sub emulator) and runs the full test suite with `cargo nextest run --all-features --profile ci` (the `ci` profile in `quickwit/.config/nextest.toml` retries failing tests twice).
 - `make test-failpoints` — runs failpoint tests only: `cargo nextest run --test failpoints --features fail/failpoints`.
 - Docker services: `make docker-compose-up` / `make docker-compose-down` (subset: `DOCKER_SERVICES=kafka,postgres`).
 - Integration tests are under `rest-api-tests`; compile and run the `quickwit-cli` binary to have an instance available to test against.
diff --git a/quickwit/.config/nextest.toml b/quickwit/.config/nextest.toml
index 3c8dd80439a..8e88cf9e541 100644
--- a/quickwit/.config/nextest.toml
+++ b/quickwit/.config/nextest.toml
@@ -6,4 +6,6 @@ slow-timeout = "10s"
 # of the run (for easy scrollability).
 failure-output = "immediate-final"
 # Do not cancel the test run on the first failure.
-fail-fast = false
\ No newline at end of file
+fail-fast = false
+slow-timeout = { period = "60s", terminate-after = 10 }
+retries = { backoff = "exponential", count = 2, delay = "1s", max-delay = "4s", jitter = true }
diff --git a/quickwit/Makefile b/quickwit/Makefile
index 5bd42980a24..8b9222736ff 100644
--- a/quickwit/Makefile
+++ b/quickwit/Makefile
@@ -55,8 +55,8 @@ test-all:
 	QW_S3_FORCE_PATH_STYLE_ACCESS=1 \
 	QW_TEST_DATABASE_URL=postgres://quickwit-dev:quickwit-dev@localhost:5432/quickwit-metastore-dev \
 	RUST_MIN_STACK=67108864 \
-	cargo nextest run --all-features --retries 5
-	cargo nextest run --test failpoints --features fail/failpoints
+	cargo nextest run --all-features --profile ci
+	cargo nextest run --profile ci --test failpoints --features fail/failpoints
 
 test-failpoints:
 	cargo nextest run --test failpoints --features fail/failpoints
diff --git a/quickwit/quickwit-common/src/test_utils.rs b/quickwit/quickwit-common/src/test_utils.rs
index 85d50b87a3e..a1854a425eb 100644
--- a/quickwit/quickwit-common/src/test_utils.rs
+++ b/quickwit/quickwit-common/src/test_utils.rs
@@ -39,36 +39,26 @@ where
     .await
 }
 
-/// Tries to connect at most 3 times to `SocketAddr`.
+/// Tries to connect to `SocketAddr` for up to 30 seconds.
 /// If not successful, returns an error.
 /// This is a convenient function to wait before sending gRPC requests
 /// to this `SocketAddr`.
 pub async fn wait_for_server_ready(socket_addr: SocketAddr) -> anyhow::Result<()> {
-    let mut num_attempts = 0;
-    let max_num_attempts = 10;
     let uri = Uri::builder()
         .scheme("http")
         .authority(socket_addr.to_string().as_str())
         .path_and_query("/")
         .build()?;
-
-    while num_attempts < max_num_attempts {
-        tokio::time::sleep(Duration::from_millis(50 * (num_attempts + 1))).await;
-        let mut http = hyper_util::client::legacy::connect::HttpConnector::new();
-        match http.call(uri.clone()).await {
-            Ok(_) => break,
-            Err(_) => {
-                println!(
-                    "Failed to connect to `{}` failed, retrying {}/{}",
-                    socket_addr,
-                    num_attempts + 1,
-                    max_num_attempts
-                );
-                num_attempts += 1;
-            }
-        }
-    }
-    if num_attempts == max_num_attempts {
+    let wait_result = wait_until_predicate(
+        || async {
+            let mut http = hyper_util::client::legacy::connect::HttpConnector::new();
+            http.call(uri.clone()).await.is_ok()
+        },
+        Duration::from_secs(30),
+        Duration::from_millis(100),
+    )
+    .await;
+    if wait_result.is_err() {
         anyhow::bail!("too many attempts to connect to `{}`", socket_addr);
     }
     Ok(())
diff --git a/quickwit/quickwit-indexing/src/actors/indexer.rs b/quickwit/quickwit-indexing/src/actors/indexer.rs
index 742c94c25c1..6cf97ab8fbf 100644
--- a/quickwit/quickwit-indexing/src/actors/indexer.rs
+++ b/quickwit/quickwit-indexing/src/actors/indexer.rs
@@ -1475,7 +1475,7 @@ mod tests {
 
     #[tokio::test]
     async fn test_indexer_exceeding_max_num_partitions() {
-        let universe = Universe::with_accelerated_time();
+        let universe = Universe::new();
         let pipeline_id = IndexingPipelineId {
             index_uid: IndexUid::new_with_random_ulid("test-index"),
             source_id: "test-source".to_string(),
diff --git a/quickwit/quickwit-indexing/src/source/kafka_source.rs b/quickwit/quickwit-indexing/src/source/kafka_source.rs
index a6769fc169d..1fe54abf4da 100644
--- a/quickwit/quickwit-indexing/src/source/kafka_source.rs
+++ b/quickwit/quickwit-indexing/src/source/kafka_source.rs
@@ -759,9 +759,11 @@ fn message_payload_to_doc(message: &BorrowedMessage) -> Option<Bytes> {
 #[cfg(all(test, feature = "kafka-broker-tests"))]
 mod kafka_broker_tests {
     use std::num::NonZeroUsize;
+    use std::sync::Arc;
 
     use quickwit_actors::{ActorContext, Universe};
     use quickwit_common::rand::append_random_suffix;
+    use quickwit_common::test_utils::wait_until_predicate;
     use quickwit_config::{SourceConfig, SourceInputFormat, SourceParams};
     use quickwit_metastore::checkpoint::SourceCheckpointDelta;
     use quickwit_metastore::metastore_for_test;
@@ -817,7 +819,45 @@ mod kafka_broker_tests {
                     err_code
                 )
             })?;
-        Ok(())
+        wait_for_topic_ready(topic, num_partitions).await
+    }
+
+    async fn wait_for_topic_ready(topic: &str, num_partitions: i32) -> anyhow::Result<()> {
+        let consumer = Arc::new(create_base_consumer("quickwit-test-wait-for-topic-ready"));
+        let is_topic_ready = || {
+            let consumer = consumer.clone();
+            let topic = topic.to_string();
+            async move {
+                spawn_blocking(move || {
+                    let Ok(cluster_metadata) =
+                        consumer.fetch_metadata(Some(&topic), Duration::from_secs(1))
+                    else {
+                        return false;
+                    };
+                    cluster_metadata.topics().iter().any(|topic_metadata| {
+                        topic_metadata.name() == topic
+                            && topic_metadata.error().is_none()
+                            && topic_metadata.partitions().len() == num_partitions as usize
+                            && topic_metadata
+                                .partitions()
+                                .iter()
+                                .all(|partition_metadata| {
+                                    partition_metadata.error().is_none()
+                                        && partition_metadata.leader() >= 0
+                                })
+                    })
+                })
+                .await
+                .unwrap_or(false)
+            }
+        };
+        wait_until_predicate(
+            is_topic_ready,
+            Duration::from_secs(30),
+            Duration::from_millis(100),
+        )
+        .await
+        .with_context(|| format!("topic `{topic}` is not ready after 30 seconds"))
     }
 
     async fn populate_topic<K, M, J, Q>(
@@ -1280,7 +1320,10 @@ mod kafka_broker_tests {
             let source_actor = SourceActor::new(source, doc_processor_mailbox);
             let (_source_mailbox, source_handle) = universe.spawn_builder().spawn(source_actor);
             let (exit_status, exit_state) = source_handle.join().await;
-            assert!(exit_status.is_success());
+            assert!(
+                exit_status.is_success(),
+                "unexpected exit status: {exit_status:?}"
+            );
 
             let messages: Vec<RawDocBatch> = doc_processor_inbox.drain_for_test_typed();
             assert!(messages.is_empty());
@@ -1330,7 +1373,10 @@ mod kafka_broker_tests {
             let source_actor = SourceActor::new(source, doc_processor_mailbox);
             let (_source_mailbox, source_handle) = universe.spawn_builder().spawn(source_actor);
             let (exit_status, exit_state) = source_handle.join().await;
-            assert!(exit_status.is_success());
+            assert!(
+                exit_status.is_success(),
+                "unexpected exit status: {exit_status:?}"
+            );
 
             let messages: Vec<RawDocBatch> = doc_processor_inbox.drain_for_test_typed();
             assert!(!messages.is_empty());
@@ -1402,7 +1448,10 @@ mod kafka_broker_tests {
             let source_actor = SourceActor::new(source, doc_processor_mailbox);
             let (_source_mailbox, source_handle) = universe.spawn_builder().spawn(source_actor);
             let (exit_status, exit_state) = source_handle.join().await;
-            assert!(exit_status.is_success());
+            assert!(
+                exit_status.is_success(),
+                "unexpected exit status: {exit_status:?}"
+            );
 
             let messages: Vec<RawDocBatch> = doc_processor_inbox.drain_for_test_typed();
             assert!(!messages.is_empty());
@@ -1453,7 +1502,10 @@ mod kafka_broker_tests {
             let source_actor = SourceActor::new(source, doc_processor_mailbox);
             let (_source_mailbox, source_handle) = universe.spawn_builder().spawn(source_actor);
             let (exit_status, exit_state) = source_handle.join().await;
-            assert!(exit_status.is_success());
+            assert!(
+                exit_status.is_success(),
+                "unexpected exit status: {exit_status:?}"
+            );
 
             let messages: Vec<RawDocBatch> = doc_processor_inbox.drain_for_test_typed();
             assert!(messages.is_empty());
diff --git a/quickwit/quickwit-ingest/src/ingest_v2/ingester.rs b/quickwit/quickwit-ingest/src/ingest_v2/ingester.rs
index bb02bef59a8..95633826939 100644
--- a/quickwit/quickwit-ingest/src/ingest_v2/ingester.rs
+++ b/quickwit/quickwit-ingest/src/ingest_v2/ingester.rs
@@ -3065,6 +3065,7 @@ mod tests {
         let state_guard = ingester.state.lock_partially("test").await.unwrap();
         let shard = state_guard.shards.get(&queue_id).unwrap();
         shard.assert_is_closed();
+        drop(state_guard);
 
         let fetch_response = timeout(Duration::from_millis(100), fetch_stream.next())
             .await
diff --git a/quickwit/quickwit-integration-tests/src/test_utils/cluster_sandbox.rs b/quickwit/quickwit-integration-tests/src/test_utils/cluster_sandbox.rs
index 7b4980f428d..ced09060bed 100644
--- a/quickwit/quickwit-integration-tests/src/test_utils/cluster_sandbox.rs
+++ b/quickwit/quickwit-integration-tests/src/test_utils/cluster_sandbox.rs
@@ -500,7 +500,7 @@ impl ClusterSandbox {
                     }
                 }
             },
-            Duration::from_secs(10),
+            Duration::from_secs(30),
             Duration::from_millis(100),
         )
         .await?;
diff --git a/quickwit/quickwit-integration-tests/src/tests/ingest_v1_tests.rs b/quickwit/quickwit-integration-tests/src/tests/ingest_v1_tests.rs
index af28f5b21a1..0b40b423601 100644
--- a/quickwit/quickwit-integration-tests/src/tests/ingest_v1_tests.rs
+++ b/quickwit/quickwit-integration-tests/src/tests/ingest_v1_tests.rs
@@ -12,10 +12,15 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+use std::time::Duration;
+
+use quickwit_common::test_utils::wait_until_predicate;
 use quickwit_config::ConfigFormat;
 use quickwit_config::service::QuickwitService;
 use quickwit_metastore::SplitState;
+use quickwit_rest_client::error::Error as RestClientError;
 use quickwit_rest_client::rest_client::CommitType;
+use reqwest::StatusCode;
 use serde_json::json;
 
 use crate::ingest_json;
@@ -58,14 +63,34 @@ async fn test_ingest_v1_happy_path() {
         .await
         .unwrap();
 
-    ingest(
-        &indexer_client,
-        index_id,
-        ingest_json!({"body": "my-doc"}),
-        CommitType::Auto,
+    let indexer_client_ref = &indexer_client;
+    wait_until_predicate(
+        || async move {
+            let ingest_result = ingest(
+                indexer_client_ref,
+                index_id,
+                ingest_json!({"body": "my-doc"}),
+                CommitType::Auto,
+            )
+            .await;
+            let Err(error) = ingest_result else {
+                return true;
+            };
+            let status_code_opt = error
+                .downcast_ref::<RestClientError>()
+                .and_then(RestClientError::status_code);
+            assert_eq!(
+                status_code_opt,
+                Some(StatusCode::NOT_FOUND),
+                "unexpected ingest error: {error:?}"
+            );
+            false
+        },
+        Duration::from_secs(10),
+        Duration::from_millis(100),
     )
     .await
-    .unwrap();
+    .expect("the indexer should create the ingest v1 queue of the index");
 
     sandbox
         .wait_for_splits(index_id, Some(vec![SplitState::Published]), 1)
diff --git a/quickwit/quickwit-integration-tests/src/tests/ingest_v2_tests.rs b/quickwit/quickwit-integration-tests/src/tests/ingest_v2_tests.rs
index 4036e47bc36..64557eb5733 100644
--- a/quickwit/quickwit-integration-tests/src/tests/ingest_v2_tests.rs
+++ b/quickwit/quickwit-integration-tests/src/tests/ingest_v2_tests.rs
@@ -29,7 +29,36 @@ use quickwit_serve::{ListSplitsQueryParams, RestIngestResponse, RestParseFailure
 use serde_json::json;
 
 use crate::ingest_json;
-use crate::test_utils::{ClusterSandboxBuilder, ingest};
+use crate::test_utils::{ClusterSandbox, ClusterSandboxBuilder, ingest};
+
+async fn wait_for_each_query_to_match(
+    sandbox: &ClusterSandbox,
+    index_id: &str,
+    queries: &[&str],
+    timeout: Duration,
+) -> anyhow::Result<()> {
+    let searcher_client = &sandbox.rest_client(QuickwitService::Searcher);
+    wait_until_predicate(
+        || async move {
+            for query in queries {
+                let search_request = quickwit_serve::SearchRequestQueryString {
+                    query: query.to_string(),
+                    max_hits: 1,
+                    ..Default::default()
+                };
+                match searcher_client.search(index_id, search_request).await {
+                    Ok(search_response) if search_response.num_hits > 0 => {}
+                    _ => return false,
+                }
+            }
+            true
+        },
+        timeout,
+        Duration::from_millis(200),
+    )
+    .await
+    .map_err(|_| anyhow::anyhow!("some queries still match no document after {timeout:?}"))
+}
 
 /// Ingesting on a freshly re-created index sometimes fails, see #5430
 #[tokio::test]
@@ -1017,26 +1046,11 @@ async fn test_graceful_shutdown_no_data_loss() {
     // All 3 documents should eventually be searchable. Documents 1 & 2 were
     // in-flight on the decommissioning indexer and should have been committed during
     // the decommission step. Document 3 was ingested to the surviving indexer.
-    wait_until_predicate(
-        || async {
-            match sandbox
-                .rest_client(QuickwitService::Searcher)
-                .search(
-                    index_id,
-                    quickwit_serve::SearchRequestQueryString {
-                        query: "*".to_string(),
-                        max_hits: 10,
-                        ..Default::default()
-                    },
-                )
-                .await
-            {
-                Ok(resp) => resp.num_hits == 3,
-                Err(_) => false,
-            }
-        },
+    wait_for_each_query_to_match(
+        &sandbox,
+        index_id,
+        &["body:1", "body:2", "body:during"],
         Duration::from_secs(30),
-        Duration::from_millis(500),
     )
     .await
     .expect("expected 3 documents after decommission shutdown, some data may have been lost");
@@ -1143,7 +1157,7 @@ async fn test_retiring_indexer_receives_empty_plan() {
                 Err(_) => false,
             }
         },
-        Duration::from_secs(1),
+        Duration::from_secs(10),
         Duration::from_millis(25),
     )
     .await
@@ -1155,18 +1169,18 @@ async fn test_retiring_indexer_receives_empty_plan() {
                 Err(_) => false,
             }
         },
-        Duration::from_secs(1),
+        Duration::from_secs(10),
         Duration::from_millis(25),
     )
     .await
     .expect("the surviving indexer should also be indexing a shard");
 
-    // Trigger the retiring node's decommission in the background. We only assert that it is told
-    // to shed its plan; we deliberately do not await full decommission.
+    // Trigger the retiring node's decommission in the background. We assert that it is told to
+    // shed its plan while it is still decommissioning, then await the end of the decommission.
     let shutdown_handle = sandbox
         .remove_node(&retiring_node_id)
         .expect("the retiring node should be in the sandbox");
-    tokio::spawn(shutdown_handle.shutdown());
+    let retiring_shutdown_join_handle = tokio::spawn(shutdown_handle.shutdown());
 
     // The fix: the new plan excludes the retiring node but is still sent to it, so it shuts down
     // its indexing pipelines (`num_running_pipelines` excludes merge pipelines) while still alive.
@@ -1178,11 +1192,21 @@ async fn test_retiring_indexer_receives_empty_plan() {
                 Err(_) => false,
             }
         },
-        Duration::from_secs(1),
+        Duration::from_secs(10),
         Duration::from_millis(25),
     )
     .await
     .expect("retiring indexer should be sent an empty plan and shut down its indexing pipelines");
+
+    tokio::time::timeout(Duration::from_secs(30), retiring_shutdown_join_handle)
+        .await
+        .expect("decommission of the retiring indexer timed out")
+        .expect("retiring indexer shutdown task panicked")
+        .expect("retiring indexer shutdown returned an error");
+    tokio::time::timeout(Duration::from_secs(30), sandbox.shutdown())
+        .await
+        .expect("sandbox shutdown timed out")
+        .unwrap();
 }
 
 #[tokio::test]
@@ -1264,7 +1288,7 @@ async fn test_retiring_indexer_decommissions_gracefully() {
                 Err(_) => false,
             }
         },
-        Duration::from_secs(1),
+        Duration::from_secs(10),
         Duration::from_millis(25),
     )
     .await
@@ -1276,7 +1300,7 @@ async fn test_retiring_indexer_decommissions_gracefully() {
                 Err(_) => false,
             }
         },
-        Duration::from_secs(1),
+        Duration::from_secs(10),
         Duration::from_millis(25),
     )
     .await
@@ -1291,26 +1315,11 @@ async fn test_retiring_indexer_decommissions_gracefully() {
         .expect("retiring indexer shutdown returned an error");
 
     // No data lost: all 6 docs remain searchable after the decommission.
-    wait_until_predicate(
-        || async {
-            match sandbox
-                .rest_client(QuickwitService::Searcher)
-                .search(
-                    index_id,
-                    quickwit_serve::SearchRequestQueryString {
-                        query: "*".to_string(),
-                        max_hits: 10,
-                        ..Default::default()
-                    },
-                )
-                .await
-            {
-                Ok(resp) => resp.num_hits == 6,
-                Err(_) => false,
-            }
-        },
-        Duration::from_secs(3),
-        Duration::from_millis(200),
+    wait_for_each_query_to_match(
+        &sandbox,
+        index_id,
+        &["body:0", "body:1", "body:2", "body:3", "body:4", "body:5"],
+        Duration::from_secs(10),
     )
     .await
     .expect("all 6 documents should be searchable after decommission");
diff --git a/quickwit/quickwit-integration-tests/src/tests/sqs_tests.rs b/quickwit/quickwit-integration-tests/src/tests/sqs_tests.rs
index e9b01834791..68c06f65c81 100644
--- a/quickwit/quickwit-integration-tests/src/tests/sqs_tests.rs
+++ b/quickwit/quickwit-integration-tests/src/tests/sqs_tests.rs
@@ -23,10 +23,32 @@ use quickwit_common::uri::Uri;
 use quickwit_config::ConfigFormat;
 use quickwit_config::service::QuickwitService;
 use quickwit_indexing::source::sqs_queue::test_helpers as sqs_test_helpers;
-use quickwit_metastore::SplitState;
 use tempfile::NamedTempFile;
 
-use crate::test_utils::ClusterSandboxBuilder;
+use crate::test_utils::{ClusterSandbox, ClusterSandboxBuilder};
+
+async fn wait_for_num_hits_at_least(sandbox: &ClusterSandbox, index_id: &str, num_hits: u64) {
+    wait_until_predicate(
+        || async {
+            sandbox
+                .rest_client(QuickwitService::Searcher)
+                .search(
+                    index_id,
+                    quickwit_serve::SearchRequestQueryString {
+                        query: "".to_string(),
+                        max_hits: 0,
+                        ..Default::default()
+                    },
+                )
+                .await
+                .is_ok_and(|search_response| search_response.num_hits >= num_hits)
+        },
+        Duration::from_secs(30),
+        Duration::from_millis(200),
+    )
+    .await
+    .unwrap_or_else(|_| panic!("index `{index_id}` should have at least {num_hits} documents"));
+}
 
 fn create_mock_data_file(num_lines: usize) -> (NamedTempFile, Uri) {
     let mut temp_file = tempfile::NamedTempFile::new().unwrap();
@@ -103,11 +125,7 @@ async fn test_sqs_with_duplicates() {
     sqs_test_helpers::send_message(&sqs_client, &queue_url, tmp_mock_data_files[5].1.as_str())
         .await;
 
-    sandbox
-        .wait_for_splits(index_id, Some(vec![SplitState::Published]), 1)
-        .await
-        .unwrap();
-
+    wait_for_num_hits_at_least(&sandbox, index_id, 10 * 1000).await;
     sandbox.assert_hit_count(index_id, "", 10 * 1000).await;
 
     // The two duplicates could not be acknowledged when the were received
@@ -203,11 +221,7 @@ async fn test_sqs_garbage_collect() {
         sqs_test_helpers::send_message(&sqs_client, &queue_url, uri.as_str()).await;
     }
 
-    sandbox
-        .wait_for_splits(index_id, Some(vec![SplitState::Published]), 1)
-        .await
-        .unwrap();
-
+    wait_for_num_hits_at_least(&sandbox, index_id, 10 * 1000).await;
     sandbox.assert_hit_count(index_id, "", 10 * 1000).await;
 
     wait_until_predicate(
diff --git a/quickwit/quickwit-janitor/src/actors/delete_task_pipeline.rs b/quickwit/quickwit-janitor/src/actors/delete_task_pipeline.rs
index 3ae458bdfa8..4abcb5a63ba 100644
--- a/quickwit/quickwit-janitor/src/actors/delete_task_pipeline.rs
+++ b/quickwit/quickwit-janitor/src/actors/delete_task_pipeline.rs
@@ -282,10 +282,13 @@ impl Handler<Observe> for DeleteTaskPipeline {
 
 #[cfg(test)]
 mod tests {
+    use std::time::Duration;
+
     use async_trait::async_trait;
     use quickwit_actors::{Handler, Universe};
     use quickwit_common::pubsub::EventBroker;
     use quickwit_common::temp_dir::TempDirectory;
+    use quickwit_common::test_utils::wait_until_predicate;
     use quickwit_indexing::TestSandbox;
     use quickwit_indexing::actors::MergeSchedulerService;
     use quickwit_metastore::{ListSplitsRequestExt, MetastoreServiceStreamSplitsExt, SplitState};
@@ -400,9 +403,35 @@ mod tests {
         let (pipeline_mailbox, pipeline_handler) = universe.spawn_builder().spawn(pipeline);
         // Ensure that the message sent by initialize method is processed.
         let _ = pipeline_handler.process_pending_and_observe().await.state;
-        // Pipeline will first fail and we need to wait a OBSERVE_PIPELINE_INTERVAL * some number
-        // for the pipeline state to be updated.
-        universe.sleep(OBSERVE_PIPELINE_INTERVAL * 5).await;
+        // Pipeline will first fail, then publish the split with the delete applied. We wait for
+        // that split, then a OBSERVE_PIPELINE_INTERVAL * some number for the pipeline state to be
+        // updated.
+        wait_until_predicate(
+            || {
+                let metastore = metastore.clone();
+                let index_uid = index_uid.clone();
+                async move {
+                    let Ok(splits) = metastore
+                        .list_splits(ListSplitsRequest::try_from_index_uid(index_uid).unwrap())
+                        .await
+                        .unwrap()
+                        .collect_splits()
+                        .await
+                    else {
+                        return false;
+                    };
+                    splits.iter().any(|split| {
+                        split.split_state == SplitState::Published
+                            && split.split_metadata.delete_opstamp == 1
+                    })
+                }
+            },
+            Duration::from_secs(30),
+            Duration::from_millis(50),
+        )
+        .await
+        .expect("the split with the delete applied should be published");
+        universe.sleep(OBSERVE_PIPELINE_INTERVAL * 2).await;
         let pipeline_state = pipeline_handler.process_pending_and_observe().await.state;
         assert_eq!(pipeline_state.delete_task_planner.metrics.num_errors, 1);
         assert_eq!(pipeline_state.downloader.metrics.num_errors, 0);
diff --git a/quickwit/quickwit-janitor/src/actors/garbage_collector.rs b/quickwit/quickwit-janitor/src/actors/garbage_collector.rs
index 239c9914bd4..d40526d96a2 100644
--- a/quickwit/quickwit-janitor/src/actors/garbage_collector.rs
+++ b/quickwit/quickwit-janitor/src/actors/garbage_collector.rs
@@ -247,7 +247,7 @@ mod tests {
     use std::path::Path;
     use std::sync::Arc;
 
-    use quickwit_actors::Universe;
+    use quickwit_actors::{ActorHandle, ObservationType, Universe};
     use quickwit_common::ServiceStream;
     use quickwit_common::shared_consts::split_deletion_grace_period;
     use quickwit_metastore::{
@@ -264,6 +264,19 @@ mod tests {
 
     use super::*;
 
+    async fn observe_after_pass(
+        handle: &ActorHandle<GarbageCollector>,
+    ) -> GarbageCollectorCounters {
+        loop {
+            let observation = handle.process_pending_and_observe().await;
+            match observation.obs_type {
+                ObservationType::Alive => return observation.state,
+                ObservationType::Timeout => continue,
+                ObservationType::PostMortem => panic!("the garbage collector exited"),
+            }
+        }
+    }
+
     fn hashmap<K: Eq + std::hash::Hash, V>(key: K, value: V) -> HashMap<K, V> {
         let mut map = HashMap::new();
         map.insert(key, value);
@@ -763,7 +776,7 @@ mod tests {
         let universe = Universe::with_accelerated_time();
         let (_mailbox, handle) = universe.spawn_builder().spawn(garbage_collect_actor);
 
-        let counters = handle.process_pending_and_observe().await.state;
+        let counters = observe_after_pass(&handle).await;
         assert_eq!(counters.num_passes, 1);
         assert_eq!(counters.num_deleted_files, 14000);
         assert_eq!(counters.num_deleted_bytes, 20 * 14000);
diff --git a/quickwit/quickwit-serve/src/health_check_api/handler.rs b/quickwit/quickwit-serve/src/health_check_api/handler.rs
index 2ed474284a1..01be1d28533 100644
--- a/quickwit/quickwit-serve/src/health_check_api/handler.rs
+++ b/quickwit/quickwit-serve/src/health_check_api/handler.rs
@@ -31,7 +31,7 @@ use crate::rest::recover_fn;
 use crate::with_arg;
 
 const HEALTH_CHECK_ASK_TIMEOUT: Duration = if cfg!(any(test, feature = "testsuite")) {
-    Duration::from_millis(100)
+    Duration::from_secs(1)
 } else {
     Duration::from_secs(5)
 };