Monitor bandwidth utilization on a per-stream basis.
Packets are captured via one of the following routes:
pcap, if it is installed, orAF_PACKET
AF_PACKET is only available on Linux (as far as I know), so if you are running on a different platform, try installing pcap. On Ubuntu, the following should do it:
sudo apt-get install python-pypcap libpcap0.8
You also need to open plex_bwmon.py and set your PLEX_TOKEN,
which should be a valid X-Plex-Token.
Now you can run python plex_bwmon.py and view real-time bandwidth
data about your streams.
Does not recognize transcoded streams yet (only direct play/stream works at the moment). This is a simple fix that will be addressed.- If two clients from the same network (i.e. with the same public IP
address) request the same file at the same time, then they will not
be identified as separate streams. This is due to the limited data
available in the
PMSlogs. I'm still working on a solution to this. Occasionally the stream metadata lookup fails to grab the metadata due to a race condition between when the stream is detected and when Plex reports the stream metadata (at the/status/sessionsendpoint). This should be pretty easy to fix.- Only tested on Linux. However,
libpcapis available on most platforms (including Windows viaWinPcap), so cross-platform support should be fairly easily doable.
This works by listening to all outbound TCP traffic. When a new stream request is observed on the Plex server the destination IP address is used to create a packet filter. As packets are sent to the client at that IP address, the bytes are counted and used to calculate the effective bandwidth.
Example output:
[xxx.xxx.xxx.xxx] [directplay] [username] [ 3.31, 4.11, 4.34 Mbps] [movie: Movie Name]
The three sets of numbers before the Mbps are the average bandwidths for the last minute, 5 minutes and 10 minutes, respectively.