Try above, but with the endpoint /identity since there’s no auth on that
And only test with post 32400, since the rest is irrelevant!
And what happens, if you browse towards the IP V6 loopback, on port 32400 and with the /identity endpoint?
Actually, going to port 32401 I get.
root@ithaki:~# curl http://127.0.0.1:32401/identity
<?xml version="1.0" encoding="UTF-8"?>
<MediaContainer size="0" claimed="0" machineIdentifier="44deff4c2239a15d3c9aec1bbb0fa4cf04dc6769" version="1.19.3.2740-add6f438d">
</MediaContainer>
root@ithaki:~#
That simply doesn’t compute
Are you sure, you don’t have some port remapping stuff running?
Yup. Positive on port forwarding.
If you look at the netstat output it is clearly listening to 32401. That’s the only reason I tried it.
Problem here is, that port 32400 is hardcoded in PMS, so simply has to be something on your end!
And the curl result you got, was indeed what was expected for port 32400 !
Ok. I’ll dig to see if I can find something. But you see what I’m seeing in the netstat (repeated below) right?
Since I can access Server on port 32401. And it shows unclaimed, is there something I can do to claim it?
Unsure here, as well as not sure if claiming it will help, due to above…
But try this little gem of mine:
And note, that all internal stuff in PMS will look for port 32400 !
To provide information here:
tcp6 0 0 :::32400 :::* LISTEN 19306/Plex Media Se
Is because PMS is LISTENing on both IPv4 and IPv6 on the same port.
PMS only has partial IPv6 support at this time.
The only results which work reliably use IPv4.
Unless absolutely required turn OFF IPv6 in the host’s network settings.
1 Like
Or in PMS Preferences.xml file
EnableIPv6="0"
Thanks for all of your help today. I’ll disable IPv6 tomorrow. And will try your claimit code (Thanks) if I still have problems
I don’t need IPv6, its just been there. So that won’t be a loss - and I hope it works. I found the change to be kernel via grub.
Will let you know.
1 Like
So, after many more hours digging, I have it working.
BTW - this is a great tool - https://github.com/iovisor/bcc
While testing,
- ran ‘iptables -F’ (as root)
- ran ‘iptables -L -n’ they were empty
- ran ‘ufw status’ showed as disabled
- stopped plex
still didn’t work. ran a bunch of tests with some of the bcc tools and also tcpdump. And used ‘netcat’ to send/receive packets on port 32400. Clearly packets were getting blocked in the kernel.
in a state of desperation :
ufw show raw
and saw two entries for port 32400 - both for kubeflow’s ‘kfserving’. But the firewall was off and iptables was clear. ?? !! ??
so, I turned off docker and rebooted. and voila - its working.
moral of the story is to also look at the output of ‘ufw show raw’ just in case…
now I need to figure out what to do about kubeflow as plex is running on my only machine with a GPU…
Thanks again for all of your help - greatly appreciated. And sorry that it was a - very non obvious - kernel issue…
Pax.
so Docker blocked it? that’s interesting.
I didn’t know this was a Docker-Plex installation.
Docker on the same machine. Plex installed natively. On Saturday I had stopped docker and flushed iptables… verified the firewall was off. and ‘iptables -L -n’ was clear. so I thought that would be enough.
Now that it is working I’d consider getting Plex to run in a container under Kubernetes with Multus-CNI networking… So the Plex container would get its own IP address on the physical LAN.
Ah. thanks.
Would you consider writing them and alerting them to the conflict?
The Host ports plex cares most about are:
This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.
