I’m trying to get the Simple-WHIP-Server (SWS) up and running dockerized, with the VideoRoom Demo (multistream).
None of the Docker images I found worked, so I created my own. This is the Dockerfile:
FROM ubuntu:20.04
RUN \
apt-get update && \
DEBIAN_FRONTEND=noninteractive apt-get -y install curl g++-10 git nano vim net-tools htop libmicrohttpd-dev libjansson-dev libssl-dev libsofia-sip-ua-dev libglib2.0-dev libopus-dev libogg-dev libcurl4-openssl-dev liblua5.3-dev libconfig-dev pkg-config gengetopt libtool automake python3 python3-pip python3-setuptools python3-dev python3-wheel ninja-build libavcodec-dev libwebsockets-dev && \
update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-10 100 --slave /usr/bin/g++ g++ /usr/bin/g++-10 --slave /usr/bin/gcov gcov /usr/bin/gcov-10
RUN pip3 install meson
WORKDIR /tmp
ENV LIBNICE_VERSION=0.1.18
ENV LIBSRTP_VERSION=v2.4.2
ENV JANUSGATEWAY_VERSION=v1.2.1
ENV CMAKE_VERSION=3.26.4
RUN curl -L -o /tmp/cmake-install.sh https://github.com/Kitware/CMake/releases/download/v${CMAKE_VERSION}/cmake-${CMAKE_VERSION}-linux-x86_64.sh \
&& chmod u+x /tmp/cmake-install.sh \
&& mkdir -p /usr \
&& /tmp/cmake-install.sh --skip-license --prefix=/usr \
&& rm /tmp/cmake-install.sh
RUN \
DIR=/tmp/libnice && \
mkdir -p ${DIR} && \
cd ${DIR} && \
curl -sLf https://github.com/libnice/libnice/archive/${LIBNICE_VERSION}.tar.gz | tar -zx --strip-components=1 && \
meson --prefix=/usr build/ && \
ninja -C build && \
ninja -C build install && \
rm -rf ${DIR}
RUN \
DIR=/tmp/libsrtp && \
mkdir -p ${DIR} && \
cd ${DIR} && \
curl -sLf https://github.com/cisco/libsrtp/archive/${LIBSRTP_VERSION}.tar.gz | tar -zx --strip-components=1 && \
./configure --prefix=/usr --enable-openssl && \
make shared_library && \
make install && \
rm -rf ${DIR}
RUN \
DIR=/tmp/janus-gateway && \
mkdir -p ${DIR} && \
cd ${DIR} && \
curl -sLf https://github.com/meetecho/janus-gateway/archive/${JANUSGATEWAY_VERSION}.tar.gz | tar -zx --strip-components=1 && \
sh autogen.sh && \
./configure --prefix=/opt/janus && \
make && \
make configs && \
make install && \
rm -rf ${DIR}
WORKDIR /app
COPY . /app
# Janus API HTTP
EXPOSE 8088/tcp
# Janus API HTTPS
EXPOSE 8089/tcp
# RTP Media
EXPOSE 20000-20100/udp
# NOTE: Usually we'd want a way larger Media/RTP port range
# but Docker is extremely slow at opening huge port ranges
# (see moby/moby#14288)
CMD exec /opt/janus/bin/janus --rtp-port-range=20000-20100 --nat-1-1=${DOCKER_IP} --configs-folder=/opt/janus/etc/janus
Maybe somebody wants to verify that it’s working and link it instead of the other ones.
Now I have a persistent issue where if more than one stream connects to the server, the connection becomes unstable and streams are disconnected and reconnected all the time.
[WARN] Video offered as 'sendonly', but we need 'inactive' for us: using 'inactive'
[janus] [Wed Feb 28 23:27:40 2024] [ERR] [sdp-utils.c:janus_sdp_get_codec_pt_full:780] Unsupported codec 'none'
[janus] [Wed Feb 28 23:27:40 2024] [WARN] Couldn't find codec we needed (none) in the offer, rejecting video
When I run whip-client -u https://api.janus.ourdomain.com/whip/endpoint/whip1 -V "filesrc location=/drone.mp4 ! qtdemux ! h264parse ! rtph264pay ! application/x-rtp,media=video,encoding-name=H264,payload=96" -S stun://stun.l.google.com:19302 -t "testwhip1" -l 7
I don’t get an error, I see a WHIP Publisher 1234 in Endpoints, but I get “no remote video available” and as soon as someone else joins the room, WHIP Publisher 1234 gets kicked.
Do we need a WHEP server + client to see the stream in the VideoRoom demo?
You’re doing something wrong with codecs. Probably you’re trying to publish to a room that doesn’t support the codec you configured in the WHIP client, and so video is rejected.
I sent you an email with a link to a video yesterday. I realize this is a bit intrusive, but the video contains identifiable data and I didn’t want to post it here.
In this video, I successfully connect one simple-whip-client with a VideoRoom (multistream). When I then connect a second whip-client with a different id to the room, the first one says “Bye!” and the second one claims it’s sending data, but nothing arrives.
I don’t think there’s anything I can do with a video. If you say users are being kicked, you should check what’s happening either at the signalling level (is that user getting a Janus API from the VideoRoom kicking them?) or WebRTC (DTLS alert? ICE disconnection? originated by who?). The latter you can check with tools like wireshark/tcpdump. Looking at the Janus logs may help too.
the video shows what happens in Simple Whip Client running 2 streams parrallel.
I followed all the steps you lined out on Github, but I can’t get it to work.
It’s two Simple Whip Client instances connecting to a Simple Whip Server connected to a Janus instance, compiled and run locally – no Docker image – with commands copied directly from the READ.ME.
The streams stop without any visible reason, logging: "[WHIP] GStreamer pipeline stopped, Bye!"
There is no error visible in the Janus log, no Ice error, no DTLS error.
The Janus Admin Api shows this:
"janus": "error", ~~ "error": {
"code": 459,
"reason": "No such handle 315845740987946 in session 7772435643855860"
}
Have you tested the Simple Whip Server with multiple Simple Whip Client streams?
@Squid I just pushed a fix. The problem was a dumb mistake on my side, as the label was being used as a unique ID in the publishers table, but the label could be the same (e.g., even when autogenerated to WHIP Publisher <room number> if two endpoints are pushing to the same room). With the fix I can’t replicate the problem anymore. Thanks for spotting this!
