From fb685665e2c75cd16490257cf55df2104d3f190a Mon Sep 17 00:00:00 2001 From: Robert Schmidt Date: Wed, 18 Mar 2026 08:59:05 +0100 Subject: [PATCH] vrtsim doc: clarify container requirements --- radio/vrtsim/README.md | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) diff --git a/radio/vrtsim/README.md b/radio/vrtsim/README.md index 2cd90e7886..02366c99fc 100644 --- a/radio/vrtsim/README.md +++ b/radio/vrtsim/README.md @@ -8,7 +8,8 @@ interface. The `vrtsim` architecture follows a server-client model: -- **Server (gNB)**: On startup, the server writes a file, default location is +- **Server (gNB)**: On startup, the server writes connection information to a + file as provided in `--vrtsim.descriptor`, default location is `/tmp/vrtsim_connection`. This file contains information required by the client to establish communication. - **Client (UE)**: The client reads this file from, verifies its contents, and uses @@ -31,6 +32,13 @@ Additionally on gNB use `vrtsim.role server` and optionally `vrtsim.timescale ` to set the timescale. Timescale 1.0 is the default and means realtime. +To use `vrtsim` in containers, you have to enable inter-process communication, +and mount the directory of the connection information in a shared volume +between server and clients. For instance, in docker and default +`vrtsim.connection_descriptor`, set `ipc: host` and share the server's and +client's `/tmp` directory through a volume. For an example, refer to +[this docker-compose file](../../ci-scripts/yaml_files/5g_vrtsim_cirdb/docker-compose.yaml). + Channel modelling can be enabled by adding `vrtsim.chanmod 1` to the command line and should work the same as channel modelling in rfsimulator, see rfsimulator [documentation](../rfsimulator/README.md) with a slight difference: