Forward-merge branch-25.02 into branch-25.06 (#2132)

Forward-merge triggered by push to branch-25.02 that creates a PR to
keep branch-25.06 up-to-date. If this PR is unable to be immediately
merged due to conflicts, it will remain open for the team to manually
merge. See [forward-merger
docs](https://docs.rapids.ai/maintainers/forward-merger/) for more info.

docker/build_container.sh

@@ -24,7 +24,6 @@ popd &> /dev/null
 DOCKER_IMAGE_NAME=${DOCKER_IMAGE_NAME:?"Must set \$DOCKER_IMAGE_NAME to build. Use the dev/release scripts to set these automatically"}
 DOCKER_IMAGE_TAG=${DOCKER_IMAGE_TAG:?"Must set \DOCKER_IMAGE_TAG to build. Use the dev/release scripts to set these automatically"}
 DOCKER_TARGET=${DOCKER_TARGET:-"runtime"}
-DOCKER_TARGET_ARCH=${DOCKER_TARGET_ARCH:-$(dpkg --print-architecture)}
 
 if [ "${DOCKER_TARGET_ARCH}" == "amd64" ]; then
     REAL_ARCH="x86_64"
@@ -58,7 +57,7 @@ PYTHON_VER=${PYTHON_VER:-3.10}
 MORPHEUS_ROOT_HOST=${MORPHEUS_ROOT_HOST:-"$(realpath --relative-to=${PWD} ${MORPHEUS_ROOT})"}
 
 # Build the docker arguments
-DOCKER_ARGS="-t ${DOCKER_IMAGE_NAME}:${DOCKER_IMAGE_TAG}-${DOCKER_TARGET_ARCH}"
+DOCKER_ARGS="-t ${DOCKER_IMAGE_NAME}:${DOCKER_IMAGE_TAG}"
 DOCKER_ARGS="${DOCKER_ARGS} --target ${DOCKER_TARGET}"
 DOCKER_ARGS="${DOCKER_ARGS} --build-arg CUDA_MAJOR_VER=${CUDA_MAJOR_VER}"
 DOCKER_ARGS="${DOCKER_ARGS} --build-arg CUDA_MINOR_VER=${CUDA_MINOR_VER}"

docker/build_container_dev.sh

@@ -18,7 +18,8 @@
 SCRIPT_DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" &> /dev/null && pwd )"
 
 export DOCKER_IMAGE_NAME=${DOCKER_IMAGE_NAME:-"morpheus"}
-export DOCKER_IMAGE_TAG=${DOCKER_IMAGE_TAG:-"dev-$(date +'%y%m%d')"}
+export DOCKER_TARGET_ARCH=${DOCKER_TARGET_ARCH:-$(dpkg --print-architecture)}
+export DOCKER_IMAGE_TAG=${DOCKER_IMAGE_TAG:-"dev-$(date +'%y%m%d')-${DOCKER_TARGET_ARCH}"}
 export DOCKER_TARGET=${DOCKER_TARGET:-"development"}
 
 # Call the general build script

docker/build_container_release.sh

@@ -21,7 +21,8 @@ SCRIPT_DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" &> /dev/null && pwd )"
 pushd ${SCRIPT_DIR} &> /dev/null
 
 export DOCKER_IMAGE_NAME=${DOCKER_IMAGE_NAME:-"nvcr.io/nvidia/morpheus/morpheus"}
-export DOCKER_IMAGE_TAG=${DOCKER_IMAGE_TAG:-"$(git describe --tags --abbrev=0)-runtime"}
+export DOCKER_TARGET_ARCH=${DOCKER_TARGET_ARCH:-$(dpkg --print-architecture)}
+export DOCKER_IMAGE_TAG=${DOCKER_IMAGE_TAG:-"$(git describe --tags --abbrev=0)-runtime-${DOCKER_TARGET_ARCH}"}
 export DOCKER_TARGET=${DOCKER_TARGET:-"runtime"}
 
 popd &> /dev/null

docs/source/extra_info/known_issues.md

@@ -19,5 +19,10 @@ limitations under the License.
 
 - `vdb_upload` example pipeline triggers an internal error in Triton ([#1649](https://github.com/nv-morpheus/Morpheus/issues/1649))
 - `ransomware_detection` example pipeline occasionally logs a `distributed.comm.core.CommClosedError` error during shutdown ([#2026](https://github.com/nv-morpheus/Morpheus/issues/2026)).
+- `abp_pcap_detection` pipeline running slowly on AArch64 ([#2120](https://github.com/nv-morpheus/Morpheus/issues/2120))
+- LLM `vdb_upload` and `rag` pipelines not supported on AArch64 ([#2122](https://github.com/nv-morpheus/Morpheus/issues/2122))
+- `gnn_fraud_detection_pipeline` not working on AArch64 ([#2123](https://github.com/nv-morpheus/Morpheus/issues/2123))
+- `ransomware_detection` pipeline running slowly on AArch64 ([#2124](https://github.com/nv-morpheus/Morpheus/issues/2124))
+- DFP visualization fails to install on AArch64 ([#2125](https://github.com/nv-morpheus/Morpheus/issues/2125))
 
 Refer to [open issues in the Morpheus project](https://github.com/nv-morpheus/Morpheus/issues)

examples/abp_pcap_detection/abp_pcap_preprocessing.py

@@ -81,19 +81,22 @@ def supports_cpp_node(self):
     def pre_process_batch(msg: ControlMessage, fea_len: int, fea_cols: typing.List[str],
                           req_cols: typing.List[str]) -> ControlMessage:
         meta = msg.payload()
+        assert meta is not None, "Payload is None"
+        orig_df = meta.get_data()
+
         # Converts the int flags field into a binary string
-        flags_bin_series = meta.get_data("flags").to_pandas().apply(lambda x: format(int(x), "05b"))
+        flags = orig_df["flags"].astype("int8")
+        flags_bin_series = (flags // 16 % 2).astype('O') + (flags // 8 % 2).astype('O') + (
+            flags // 4 % 2).astype('O') + (flags // 2 % 2).astype('O') + (flags % 2).astype('O')
 
-        # Expand binary string into an array
-        df = cudf.DataFrame(np.vstack(flags_bin_series.str.findall("[0-1]")).astype("int8"),
-                            index=meta.get_data().index)
+        flag_array = flags_bin_series.str.findall("[0-1]").list.astype("int8")
 
-        # adding [ack, psh, rst, syn, fin] details from the binary flag
+        # Expand binary string into an array adding [ack, psh, rst, syn, fin] details from the binary flag
         rename_cols_dct = {0: "ack", 1: "psh", 2: "rst", 3: "syn", 4: "fin"}
-        df = df.rename(columns=rename_cols_dct)
+        df = cudf.DataFrame({rename_cols_dct[i]: flag_array.list.get(i) for i in range(5)}, index=orig_df.index)
 
         df["flags_bin"] = flags_bin_series
-        df["timestamp"] = meta.get_data("timestamp").astype("int64")
+        df["timestamp"] = orig_df["timestamp"].astype("int64")
 
         def round_time_kernel(timestamp, rollup_time, secs):
             for i, time in enumerate(timestamp):
@@ -112,8 +115,8 @@ def round_time_kernel(timestamp, rollup_time, secs):
         df["rollup_time"] = cudf.to_datetime(df["rollup_time"], unit="us").dt.strftime("%Y-%m-%d %H:%M")
 
         # creating flow_id "src_ip:src_port=dst_ip:dst_port"
-        df["flow_id"] = (meta.get_data("src_ip") + ":" + meta.get_data("src_port").astype("str") + "=" +
-                         meta.get_data("dest_ip") + ":" + meta.get_data("dest_port").astype("str"))
+        df["flow_id"] = (orig_df["src_ip"] + ":" + orig_df["src_port"].astype("str") + "=" + orig_df["dest_ip"] + ":" +
+                         orig_df["dest_port"].astype("str"))
         agg_dict = {
             "ack": "sum",
             "psh": "sum",
@@ -124,7 +127,7 @@ def round_time_kernel(timestamp, rollup_time, secs):
             "flow_id": "count",
         }
 
-        df["data_len"] = meta.get_data("data_len").astype("int16")
+        df["data_len"] = orig_df["data_len"].astype("int16")
 
         # group by operation
         grouped_df = df.groupby(["rollup_time", "flow_id"]).agg(agg_dict)

examples/digital_fingerprinting/production/grafana/README.md

@@ -63,20 +63,6 @@ configure_logging(loki_handler, log_level=log_level)
 
 More information about Loki Python logging can be found [here](https://pypi.org/project/python-logging-loki/).
 
-## Build the Morpheus container:
-From the root of the Morpheus repo:
-```bash
-./docker/build_container_release.sh
-```
-
-Build `docker compose` services:
-
-```
-cd examples/digital_fingerprinting/production
-export MORPHEUS_CONTAINER_VERSION="$(git describe --tags --abbrev=0)-runtime"
-docker compose build
-```
-
 ## Start Grafana and Loki services:
 
 To start Grafana and Loki, run the following command on host in `examples/digital_fingerprinting/production`:
@@ -86,7 +72,7 @@ docker compose up grafana
 
 ## Run Azure DFP Training
 
-Create `bash` shell in `morpheus_pipeline` container:
+Start `bash` shell in `morpheus_pipeline` container, run the following command on host in `examples/digital_fingerprinting/production`:
 
 ```bash
 docker compose run --rm morpheus_pipeline bash

examples/digital_fingerprinting/visualization/README.md

@@ -18,6 +18,13 @@
 
 We show here how to set up and run the Production DFP pipeline on Azure and Duo log data to generate input files for the DFP visualization UI. You can find more information about the Production DFP pipeline in this [README](../production/README.md) and the [DFP Developer Guide](../../../docs/source/developer_guide/guides/5_digital_fingerprinting.md).
 
+## Supported Architectures
+| Architecture | Supported | Issue |
+|--------------|-----------|-------|
+| x86_64 | ✔ | |
+| aarch64 | ✘ | [#2125](https://github.com/nv-morpheus/Morpheus/issues/2125) |
+
+
 ## Prerequisites
 
 To run the demo you will need the following:
@@ -30,17 +37,10 @@ To run the demo you will need the following:
 git submodule update --init --recursive
 ```
 
-## Build the Morpheus container
-This is necessary to get the latest changes needed for DFP. From the root of the Morpheus repo:
-```bash
-./docker/build_container_release.sh
-```
-
 ## Building Services via `docker compose`
 
 ```bash
 cd examples/digital_fingerprinting/production
-export MORPHEUS_CONTAINER_VERSION="$(git describe --tags --abbrev=0)-runtime"
 docker compose build
 ```
 

examples/gnn_fraud_detection_pipeline/README.md

@@ -25,6 +25,12 @@ All environments require additional Conda packages which can be installed with e
 | Morpheus Release Container | ✔ |  |
 | Dev Container | ✔ |  |
 
+### Supported Architectures
+| Architecture | Supported | Issue |
+|--------------|-----------|-------|
+| x86_64 | ✔ | |
+| aarch64 | ✘ | [#2123](https://github.com/nv-morpheus/Morpheus/issues/2123) |
+
 ## Requirements
 
 Prior to running the GNN fraud detection pipeline, additional requirements must be installed in to your Conda environment.

examples/llm/rag/README.md

@@ -26,6 +26,12 @@ All environments require additional Conda packages which can be installed with e
 | Morpheus Release Container | ✔ | Requires launching Milvus on the host |
 | Dev Container | ✘ |  |
 
+### Supported Architectures
+| Architecture | Supported | Issue |
+|--------------|-----------|-------|
+| x86_64 | ✔ | |
+| aarch64 | ✘ | [#2122](https://github.com/nv-morpheus/Morpheus/issues/2122) |
+
 ## Table of Contents
 
 ## Background Information

examples/llm/vdb_upload/README.md

@@ -42,6 +42,11 @@ All environments require additional Conda packages which can be installed with e
 | Morpheus Release Container | ✔ | Requires launching Triton and Milvus on the host |
 | Dev Container | ✘ |  |
 
+### Supported Architectures
+| Architecture | Supported | Issue |
+|--------------|-----------|-------|
+| x86_64 | ✔ | |
+| aarch64 | ✘ | [#2122](https://github.com/nv-morpheus/Morpheus/issues/2122) |
 
 ## Background Information
 

examples/llm/vdb_upload/vdb_utils.py

@@ -15,12 +15,10 @@
 import logging
 import typing
 
-import pymilvus
 import yaml
 
 from morpheus.config import Config
 from morpheus.config import PipelineModes
-from morpheus_llm.service.vdb.milvus_client import DATA_TYPE_MAP
 
 logger = logging.getLogger(__name__)
 
@@ -147,6 +145,10 @@
 
 
 def build_milvus_config(resource_schema_config: dict):
+    import pymilvus
+
+    from morpheus_llm.service.vdb.milvus_client import DATA_TYPE_MAP
+
     schema_fields = []
     for field_data in resource_schema_config["schema_conf"]["schema_fields"]:
         field_data["dtype"] = DATA_TYPE_MAP.get(field_data["dtype"])

examples/ransomware_detection/README.md

@@ -88,22 +88,24 @@ Usage: run.py [OPTIONS]
 
 Options:
   --debug BOOLEAN
-  --num_threads INTEGER RANGE     Number of internal pipeline threads to use
+  --num_threads INTEGER RANGE     Number of internal pipeline threads to use.
                                   [x>=1]
-  --n_dask_workers INTEGER RANGE  Number of dask workers  [x>=2]
+  --n_dask_workers INTEGER RANGE  Number of dask workers.  [x>=1]
   --threads_per_dask_worker INTEGER RANGE
-                                  Number of threads per each dask worker
-                                  [x>=2]
+                                  Number of threads per each dask worker.
+                                  [x>=1]
   --model_max_batch_size INTEGER RANGE
-                                  Max batch size to use for the model  [x>=1]
-  --model_fea_length INTEGER RANGE
-                                  Features length to use for the model  [x>=1]
-  --features_file TEXT            File path for ransomware detection features
+                                  Max batch size to use for the model.  [x>=1]
+  --pipeline_batch_size INTEGER RANGE
+                                  Internal batch size for the pipeline. Can be
+                                  much larger than the model batch size.
+                                  [x>=1]
+  --conf_file TEXT                Ransomware detection configuration filepath.
   --model_name TEXT               The name of the model that is deployed on
-                                  Tritonserver
-  --server_url TEXT               Tritonserver url  [required]
+                                  Tritonserver.
+  --server_url TEXT               Tritonserver url.  [required]
   --sliding_window INTEGER RANGE  Sliding window to be used for model input
-                                  request  [x>=1]
+                                  request.  [x>=3]
   --input_glob TEXT               Input glob pattern to match files to read.
                                   For example,
                                   './input_dir/*/snapshot-*/*.json' would read
@@ -120,6 +122,6 @@ Options:
   --output_file TEXT              The path to the file where the inference
                                   output will be saved.
   --help                          Show this message and exit.
-  ```
+```
 
 > **Note**: There is a known race condition in `dask.distributed` which occasionally causes `tornado.iostream.StreamClosedError` to be raised during shutdown, but does not affect the output of the pipeline. If you see this exception during shutdown, it is typically safe to ignore unless it corresponds to other undesirable behavior. For more information see ([#2026](https://github.com/nv-morpheus/Morpheus/issues/2026)).