gRPC Client/Server Setup

DR_EVT can run as a network service, enabling remote simulation control from any language that supports gRPC. This page is a quick-start summary; see the gRPC Client/Server Guide for the full picture (service definition, dependency resolution, the MPI multi-client/multi-server harness, testing).

Architecture

        graph LR
    Client[dr_evt_client<br/>or your own gRPC client] -->|gRPC Stream| Server[dr_evt_server<br/>C++]
    Server -->|Simulation| Engine[Scheduler<br/>Engine]
    Client -->|AppendJobRequest| Server
    Client -->|SubmitJobRequest| Server
    Client -->|AdvanceToRequest| Server
    Server -->|Events, statistics| Client

Use cases:

Run the simulator on an HPC cluster, control it from a laptop
Multi-language integration (a client in any language gRPC supports, talking to the C++ server)
Modeling multiple independent job-arrival streams that need to stay wall-clock coordinated (see the MPI harness in the full guide)

Building with gRPC Support

The gRPC client/server is optional and requires two CMake flags:

cmake .. \
  -DDR_EVT_ENABLE_PROTOBUF=ON \
  -DDR_EVT_ENABLE_GRPC=ON
make dr_evt_server-bin dr_evt_client-bin
make install

You do not need to install gRPC/Protobuf system packages first: if they aren’t found via find_package, the build automatically falls back to fetching and building gRPC’s own source tree (slower the first time, but works without root/sudo access - the common case on shared HPC/cluster environments). See the full guide’s Dependency resolution section for exactly what’s tried and in what order.

Running the Server

${CMAKE_INSTALL_PREFIX}/bin/dr_evt_server 0.0.0.0:50051

The address argument is optional (defaults to 0.0.0.0:50051 if omitted). There are no other command-line options - no --port, --max_connections, --timeout, --tls_cert, or --verbose flags; the server prints one line on startup:

DR_EVT simulation server listening on 0.0.0.0:50051

and otherwise runs silently, one Simulation instance per connected session, until stopped (e.g. Ctrl-C, or however your process supervisor manages it).

Bind 0.0.0.0 (shown above) so the server is reachable from other machines; bind 127.0.0.1 instead if you only need same-machine access.

Running the Client

${CMAKE_INSTALL_PREFIX}/bin/dr_evt_client <server_host>:50051 /path/to/trace.csv

Both arguments are required (server address, then a job data file). There are no other command-line options. dr_evt_client is a minimal example client demonstrating genuine streaming: it reads job data (submit_time, num_nodes, queue, time_limit) from the file client-side only - the server never loads this file itself - appends each job to the server via AppendJobRequest, submits each with SubmitJobRequest, advances the simulation, and prints final statistics. It’s a reference for writing your own client against the same .proto service, not a general-purpose tool with its own configuration surface. See the full guide for what each RPC corresponds to in the in-process streaming API.

There is no Python client or Python gRPC bindings shipped with this project today - the Python API is a separate, in-process (no network) binding, unrelated to the gRPC service. Writing a Python gRPC client is possible (gRPC supports Python), but would mean generating your own stubs from src/proto/dr_evt_service.proto with grpc_tools.protoc - not something this project provides out of the box.

Network Configuration

Firewall rules

Allow gRPC traffic on whatever port the server is bound to (50051 in the examples above):

Linux (ufw):

sudo ufw allow 50051/tcp
sudo ufw reload

Linux (iptables):

sudo iptables -A INPUT -p tcp --dport 50051 -j ACCEPT
sudo iptables-save

SSH tunneling

The server has no built-in TLS/encryption support today - traffic is plaintext gRPC. For access over an untrusted network, tunnel it over SSH instead:

# On your local machine: forward local port 50051 to the remote server
ssh -L 50051:localhost:50051 user@remote-server

# In another terminal: connect the client as if the server were local
${CMAKE_INSTALL_PREFIX}/bin/dr_evt_client localhost:50051 /path/to/trace.csv

Troubleshooting

Connection refused

Problem: the client reports it failed to connect.

Solutions:

Check the server is actually running: ps aux | grep dr_evt_server
Check it’s listening on the port you expect: netstat -an | grep 50051
Check your firewall: sudo ufw status
Try 127.0.0.1:50051 explicitly instead of localhost:50051, in case of IPv4/IPv6 resolution differences

Large trace, high memory use

If the job data file is large enough that loading it client-side (or, separately, running the simulator in single-file batch mode) uses more memory than you’d like, see --infile_list (progressive/ multi-file loading) for the CLI/batch case - it’s unrelated to the gRPC client/server specifically, but addresses the same underlying concern.