Hi all,
I just stumbled across this, and I have been searching for a solution like this for ages now. Thank you wnielson!
I was just wondering if in a future release if it is possible to implement a sleep mode for the slave when the slave is idle. This should be woken up when there is a need for it. Let me try to illustrate what I have in mind:
PMS-> wake up slave 1.
Slave 1 run with >90% CPU for 1 min. then wake up slave 2.
Slave 2 run with >90% CPU for 1 min. then wake up slave 3.
Etc until you have no more slaves.
When going to sleep:
Slave 3 run with <10% CPU for 10 min. then go to sleep.
Slave 2 run with <10% CPU for 10 min. then go to sleep.
Slave 1 run with <10% CPU for 10 min. then go to sleep.
Now this constellation requires that the cache/buffer is set to 3-4 mintues in order to have some time to wake another slave, so the user(s) do not experience any degradation nor lag.
Now let me make one thing 100% clear I am NOT a developer. I am just an āidiotā with strange ideas. (and too many computers at home)
is possible to change uid from plex user on master server because on slave when i create user the uid is already used on other as i see from nfs i am asking if is possible to change uid from master server to some else and then create plex user with the same uid on slaveā¦
If you install PRT with no clients, or if clients cannot be reached it will still transcode locally?
. If you are using /dev/shm for the transcoder temp directory, would you share this out, and does one think it would become a problem with multiple clients writing to it?
If you install PRT with no clients, or if clients cannot be reached it will still transcode locally?
Yes. The master is always a fallback transcoder.
. If you are using /dev/shm for the transcoder temp directory, would you share this out, and does one think it would become a problem with multiple clients writing to it?
Iāve never tried network sharing the shared memory device before. I suppose itāll work but it is untested.
Iāve been trying to get PRT to work master salveā¦ I keep getting
JobManager: child process returned: 1 (Operation not permitted)
Jobs: ā/usr/lib/plexmediaserver/Plex Transcoderā exit code for process 2725 is 1
no other errors and the web browser that I started the movie will wait forever till i stop it. If I put the original Plex Transcoder back it works fineā¦
If you install PRT with no clients, or if clients cannot be reached it will still transcode locally?
Yes. The master is always a fallback transcoder.
. If you are using /dev/shm for the transcoder temp directory, would you share this out, and does one think it would become a problem with multiple clients writing to it?
Iāve never tried network sharing the shared memory device before. I suppose itāll work but it is untested.
I just setup a spare server tonight, and installed PRT on the master. I need to change the transcoding directory and then setup a client transcoder and see how it works.
Theoretically the number of transcodings wont change anymore than what they are now. The processing load would just spread out, although if the number of transcodings were to go up I suppose I could run into an issue.
If you install PRT with no clients, or if clients cannot be reached it will still transcode locally?
Yes. The master is always a fallback transcoder.
. If you are using /dev/shm for the transcoder temp directory, would you share this out, and does one think it would become a problem with multiple clients writing to it?
Iāve never tried network sharing the shared memory device before. I suppose itāll work but it is untested.
If you install PRT with no clients, or if clients cannot be reached it will still transcode locally?
Yes. The master is always a fallback transcoder.
. If you are using /dev/shm for the transcoder temp directory, would you share this out, and does one think it would become a problem with multiple clients writing to it?
Iāve never tried network sharing the shared memory device before. I suppose itāll work but it is untested.
So I got it working without using /dev/shm as the transcoding directory, but the minute I switched to /dev/shm, and update my mounts, it quit working.
In the logs I see the transcode request sent, and a second later I see the Transcode stopped message.
So looks like I have to decide between writing to /dev/shm or using separate hosts for transcoding.
I have PRT working now. Iāve set the slaves to sleep if there is no activity. I can send a WOL command from the terminal and wake each one up. Would like to add that command into PRT, so when it needs the additional processing power it could wake up the node on its own, just not sure how to go about doing that. Any pointers would be greatly appreciated.
You can use it on CentOSā¦ but I donāt know of anyone that actually has gotten it to work per say on it. It was developed for ubuntu and works quite well on that.
Filesystem wise, it does not matter, you will just be sharing out the folders as exports, so the underlying FS does not really matter.
I tried to install it on Centos 7ā¦ PMS worked just fine but PRT would not work. The problem seemed to be that PRT would create the transcode job but Centos would report,
JobManager: child process returned: 1 (Operation not permitted)
Jobs: ā/usr/lib/plexmediaserver/Plex Transcoderā exit code for process 2725 is 1
With no other errorsā¦ And PMS would just keep acting as if everything was fine. The client that started the video would just keep waiting for it to start till it was manually stopped.
Hi, I dont know if this as been already anwsered but iām planning to move my instalation to Unraid OS and I was wondering if I can make PRT working through a docker ? Otherwise I think a Ubuntu server VM can be a work around.
(and too many computers at home)
