Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[eve]

Is somewhat poor performance on the docker ( https://gitlab.com/shiomax/jrivermc-docker ) to be expected? The server I'm running it on is rarely loaded heavily and runs 12c 24t at 3 ghz, not a champ but nothing to sneeze at, I have a couple threadripper servers sitting around that are moderately faster I can try but like, I picked this one because it's rarely loaded and mainly just handles media serving.
Everything but the media itself is on SSDs (an SSD for JRiver and a dedicated metadata SSD that lays OVER my actual media, presenting as a single directory structure alongside the media files). While my JRiver install on my PC is connected to the library hosted on this server, I've noticed that right click operations in JRiver's library lag significantly, and by hopping onto the VNC client of the Docker container, those slowdowns are present as well meaning it's not just a network issue
I dunno, it barely hits more than 1 core at a time and I don't remember JRiver being that single thread limited. I guess 3Ghz from maybe 7 years ago doesn't compare to something recent that's more like 4.5 and has major IPC improvements.
I'll probably spin up a VM later and see if performance improves. I don't mind running a VM for this but if it's just a matter of tweaking the settings for this specific docker container, or changing a setting in JRiver, I'd rather stick with docker.

[JimH]

Could be related to initial import, analysis, thumbnailing, etc. 

There's a Docker thread here somewhere.

[eve]

Could be related to initial import, analysis, thumbnailing, etc. 

There's a Docker thread here somewhere.

None of that is running at the moment. It took a half a day or so to pull in and perf issues have persisted.
Thumbnailing is off in JRiver as I have thumbs, posters and more for each file already.
The library isn't massive, like 25K media files (all video). It's primarily remuxed MKVs. I haven't gotten around to making the full discs available to JRiver yet as I need to do another pass on the metadata and try to clarify pressings / releases.

Jim, unrelated question. Does audio analysis work on bitstream tracks like DTS-HD and TrueHD?
Would be cool to have a quick reference for differences between releases and pressings.
Not all my MKVs have FLAC conversions in them. I usually take all the original audio, and as of late, I've also been including a FLAC conversion for relevant multichannel and stereo mixes which you should do for stereo anyways since they often compress much smaller than the bitstream track with it's core.

[Hendrik]

Audio Analysis works for all audio. However, during playback, volume adjustments etc can only be done when actually decoding the audio.

[max096]

Generally, performance should not be noticeably worse in docker unless you need GPU acceleration (but kinda the same is true for a VM without passing threw a GPU).

Would be interesting if you could try with a VM on the same PC if there is any difference. Really, you should not see a significant performance difference between VM and docker. If you do see a difference try with docker inside the VM as well. The image uses debian 11, maybe there is something going on when you run docker on an older kernel that was released before it as docker uses the kernel of the host, which then would be the VMs kernel.

[eve]

Generally, performance should not be noticeably worse in docker unless you need GPU acceleration (but kinda the same is true for a VM without passing threw a GPU).

Would be interesting if you could try with a VM on the same PC if there is any difference. Really, you should not see a significant performance difference between VM and docker. If you do see a difference try with docker inside the VM as well. The image uses debian 11, maybe there is something going on when you run docker on an older kernel that was released before it as docker uses the kernel of the host, which then would be the VMs kernel.

Identical, or near identical perf Docker vs VM has always been my experience as well (and frankly GPU acceleration in dockers isn't that bad nowadays, I have one passed to emby and then I've spun up others for rendering in the past without much hassle or issue)
Hence why I was curious if there was something limiting this specific docker container. Or I was doing something stupid.

I'm just running Unraid as the host right now. 5.15.46-Unraid is my Kernel version.