debug1: userauth-request for user plex service ssh-connection method none [preauth]
debug1: attempt 0 failures 0 [preauth]
debug1: user plex matched 'User plex' at line 90
debug1: PAM: initializing for "plex"
debug1: PAM: setting PAM_RHOST to "192.168.1.219"
debug1: PAM: setting PAM_TTY to "ssh"
debug1: PAM: password authentication failed for plex: Authentication failure
Failed none for plex from 192.168.1.219 port 47476 ssh2
debug1: userauth-request for user plex service ssh-connection method publickey [preauth]
debug1: attempt 1 failures 0 [preauth]
debug1: test whether pkalg/pkblob are acceptable [preauth]
debug1: temporarily_use_uid: 108/117 (e=0/0)
debug1: trying public key file /var/lib/plexmediaserver/.ssh/authorized_keys
debug1: fd 7 clearing O_NONBLOCK
Authentication refused: bad ownership or modes for file /var/lib/plexmediaserver/.ssh/authorized_keys
debug1: restore_uid: 0/0
Failed publickey for plex from 192.168.1.219 port 47476 ssh2: RSA
I did do that actually based on a guide I found online.
So it seems my slave plex user cannot access that .ssh folder in the share for whatever reason. I set the user up exactly according to the docs, same gid/uid. Still no access to that stuff. Any idea?
This looks extremely interesting. I just have one question before I try it out. We have a big server farm with 48 xenon 6-core processors and I’m thinking about using that power for plex to transcode and generate thumbnails. I run the plex server on a separate i7 equipped server that holds the raidsystem/storage and it’s on it’s own external IP but on the same switch system as the farm.
Whats the best way to use all the power from the farm? I’m no server technician but have full access to the system with support from our “server guy” He asked if i needed ONE VM with acces to all the processors and cores or to a couple of VMs with access to one processor per VM?
Can I tell it to keep using the same node and all its power even when the next user request a transcode or does it sees it as occupied and sends the user to next node in list (the main plex server)?
@Nordicpipeline said:
This looks extremely interesting. I just have one question before I try it out. We have a big server farm with 48 xenon 6-core processors and I’m thinking about using that power for plex to transcode and generate thumbnails. I run the plex server on a separate i7 equipped server that holds the raidsystem/storage and it’s on it’s own external IP but on the same switch system as the farm.
Whats the best way to use all the power from the farm? I’m no server technician but have full access to the system with support from our “server guy” He asked if i needed ONE VM with acces to all the processors and cores or to a couple of VMs with access to one processor per VM?
Can I tell it to keep using the same node and all its power even when the next user request a transcode or does it sees it as occupied and sends the user to next node in list (the main plex server)?
Thumbnail generation currently doesn’t work (see here). We’re are tying to work on a fix though.
As for your configuration, I would say that one VM with all of the cores would be the way to go (less configuration on your end). Load balancing in PRT is currently done based on load, so the machine with the lowest load will get prioritized.
On a side note, I’m curious as to how your server farm is configured. The only platform I’ve seen that does what you’ve described is ScaleMP.
@Nordicpipeline said:
This looks extremely interesting. I just have one question before I try it out. We have a big server farm with 48 xenon 6-core processors and I’m thinking about using that power for plex to transcode and generate thumbnails. I run the plex server on a separate i7 equipped server that holds the raidsystem/storage and it’s on it’s own external IP but on the same switch system as the farm.
Whats the best way to use all the power from the farm? I’m no server technician but have full access to the system with support from our “server guy” He asked if i needed ONE VM with acces to all the processors and cores or to a couple of VMs with access to one processor per VM?
Can I tell it to keep using the same node and all its power even when the next user request a transcode or does it sees it as occupied and sends the user to next node in list (the main plex server)?
Thumbnail generation currently doesn’t work (see here). We’re are tying to work on a fix though.
As for your configuration, I would say that one VM with all of the cores would be the way to go (less configuration on your end). Load balancing in PRT is currently done based on load, so the machine with the lowest load will get prioritized.
On a side note, I’m curious as to how your server farm is configured. The only platform I’ve seen that does what you’ve described is ScaleMP.
Did you try running PMS on scalemp ? Im asking because I got 2 Dell 2950 servers with 2 quad core cpu each. Im looking for a good cluster application solution.
@igowas said:
Did you try running PMS on scalemp ? Im asking because I got 2 Dell 2950 servers with 2 quad core cpu each. Im looking for a good cluster application solution.
I haven’t tried ScaleMP, no. It sounds like a great idea, in theory, but you’d need to shell out for vSMP Foundation in order to use more than one processor at a time and the licensing fees are a bit steep.
@igowas said:
Did you try running PMS on scalemp ? Im asking because I got 2 Dell 2950 servers with 2 quad core cpu each. Im looking for a good cluster application solution.
I haven’t tried ScaleMP, no. It sounds like a great idea, in theory, but you’d need to shell out for vSMP Foundation in order to use more than one processor at a time and the licensing fees are a bit steep.
ScaleMP sounds really interesting, I haven’t heard of it before. I wonder if there is any open source alternative.
I was planning on building a small rpi3 cluster and this looks like a perfect project for that.
Does anyone know if this will run on a Raspberry pi 3? As it is right now my old Linux server basically dies when three or more people are transcoding at the same time.
Edit: After googling around a bit it seems that the rpi3 isnt even good enough to transcode one single stream so I’m probably gonna test to put up some VMs on my gaming rig instead.
I Got it ! It’s now working for a week with 3 computers 2x Dell Poweredge 2950 and 1x Dell desktop. Thx I really like your application. Are you planning to add more features ?
Thumbnail generation has been tested to work. See here for the solution.
You did the change on the Master right ? Im asking because I dont want to brick my actuall installation Lol.
I am guessing you didn’t even bother to read the link? If you did, you would have seen it was just symlinking some directories for the fix. Nothing with the code. I did an alternative fix and just allowed read write access on the other shares and I haven’t had any issues with index file creation.
Thumbnail generation has been tested to work. See here for the solution.
You did the change on the Master right ? Im asking because I dont want to brick my actuall installation Lol.
I am guessing you didn’t even bother to read the link? If you did, you would have seen it was just symlinking some directories for the fix. Nothing with the code. I did an alternative fix and just allowed read write access on the other shares and I haven’t had any issues with index file creation.
Yes, I read all the link. Thx I’ll go with your solution.
question: Has it been proposed yet to use wake on lan combined with load checking/balancing? Usually a lot of these transcode systems would be pretty high power consumption?
So for example I have a 12k passmark master server - after load goes above x, wake slave transcoder 1 - watch load, if load continues to increase after both are at load x, wake next transcoder if available and it continues to scale n.
Once load drops back down, new sessions go back to highest system in chain (or even current sessions moved back up chain). After x minutes of inactivity, slaves go back into suspend mode
He asked if i needed ONE VM with acces to all the processors and cores or to a couple of VMs with access to one processor per VM?
He asked if i needed ONE VM with acces to all the processors and cores or to a couple of VMs with access to one processor per VM?
thanks.