diff --git a/Cargo.lock b/Cargo.lock
index f1c109a79..26c53663a 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -1,5 +1,7 @@
 # This file is automatically @generated by Cargo.
 # It is not intended for manual editing.
+version = 3
+
 [[package]]
 name = "actix-codec"
 version = "0.4.0"
@@ -1840,8 +1842,8 @@ dependencies = [
 
 [[package]]
 name = "milli"
-version = "0.2.0"
-source = "git+https://github.com/meilisearch/milli.git?tag=v0.2.0#792225eaffce6b3682f9b30b7370b6a547c4757e"
+version = "0.2.1"
+source = "git+https://github.com/meilisearch/milli.git?tag=v0.2.1#25f75d4d03732131e6edcf20f4d126210b159d43"
 dependencies = [
  "anyhow",
  "bstr",
diff --git a/meilisearch-http/Cargo.toml b/meilisearch-http/Cargo.toml
index 7ac3ecb38..c9f8d63b7 100644
--- a/meilisearch-http/Cargo.toml
+++ b/meilisearch-http/Cargo.toml
@@ -51,7 +51,7 @@ main_error = "0.1.0"
 meilisearch-error = { path = "../meilisearch-error" }
 meilisearch-tokenizer = { git = "https://github.com/meilisearch/Tokenizer.git", tag = "v0.2.2" }
 memmap = "0.7.0"
-milli = { git = "https://github.com/meilisearch/milli.git", tag = "v0.2.0" }
+milli = { git = "https://github.com/meilisearch/milli.git", tag = "v0.2.1" }
 mime = "0.3.16"
 once_cell = "1.5.2"
 oxidized-json-checker = "0.3.2"
diff --git a/meilisearch-http/src/index_controller/index_actor/actor.rs b/meilisearch-http/src/index_controller/index_actor/actor.rs
index ca23663b7..1f0091265 100644
--- a/meilisearch-http/src/index_controller/index_actor/actor.rs
+++ b/meilisearch-http/src/index_controller/index_actor/actor.rs
@@ -333,16 +333,12 @@ impl<S: IndexStore + Sync + Send> IndexActor<S> {
 
                 // Get write txn to wait for ongoing write transaction before dump.
                 let txn = index.write_txn()?;
-                let documents_ids = index.documents_ids(&txn)?;
-                // TODO: TAMO: calling this function here can consume **a lot** of RAM, we should
-                // use some kind of iterators -> waiting for a milli release
-                let documents = index.documents(&txn, documents_ids)?;
-
                 let fields_ids_map = index.fields_ids_map(&txn)?;
                 // we want to save **all** the fields in the dump.
                 let fields_to_dump: Vec<u8> = fields_ids_map.iter().map(|(id, _)| id).collect();
 
-                for (_doc_id, document) in documents {
+                for document in index.all_documents(&txn)? {
+                    let (_doc_id, document) = document?;
                     let json = milli::obkv_to_json(&fields_to_dump, &fields_ids_map, document)?;
                     file.write_all(serde_json::to_string(&json)?.as_bytes())?;
                     file.write_all(b"\n")?;