ConnerVT

Thanks for the post. I haven't gotten any better. I've meant to update this thread a few times - As you can see from my siggy, the number of servers I have has doubled, and most of what is inside them has changed over time. There is little that is left from my original server build. Just like the ship of Theseus, if all of the parts are replaced, is it still the same server? This week I got a call and a voicemail from ServerPartDeals. They must of been worried about me, as I hadn't ordered any drives in several months. I've had a number of cold calls from businesses over the years, but this was the first time I've had a computer parts store call...

The Plex browser client has always had issues. One of these is it seems to always force a transcode even when it should certainly be able to direct play (or at least direct stream). You wrote that you tried several different containers. Did you use a new appdata set for these, or were you pointing it to the same appdata folder? One thing that can fix funky Plex transcoding is to stop the container, then delete all of the contents of your codec folder (in the plex appdata folder, several levels down). When you restart the container, it will go out and repopulate the codec folder with fresh files. This is another one of the long time fixes for weird plex issues.

Yeah, but now he needs to update to Win11.

For me, if the listing for sale doesn't specifically answer the questions I have for what is important to me, I move on and look for something else.

Thanks so much for the input. I wish to nail this down, before I scale up adding the rest of the cameras and find I need to change a ton of configuration. I'll play with my FPS when I add the cameras, if usage starts to be too extreme. I figured the 8971 port change was correct, but I was eager to see frigate actually run and 5000 got me there. I'll work on getting it configured for 8971 the next time I get a chance to work with it. (Time again to do my real job.)

I have 0.14-rc2 fired up on my server. Things seem to be running well. Just wish to pass on an observation, and get feedback on one thing I'm seeing, and find out if I'm chasing ghosts. Setup is using AMD graphics hardware acceleration, USB Coral, and a Amcrest IP5M-T1179EW-28MM. I wish to get this configuration nailed down, as final setup will have 6 cameras. The observation is that I needed to update your template, changing the WebUI to 5000 from the 8554 in your template. Otherwise, the WebUI would not launch when clicking the frigate icon and selecting Webui on the Unraid Docker tab. Is this an oversight on your part? Or is there something I need to add to my config.yml? The question I have is that on the System Metrics/Stats page, I am seeing a warning that my primary feed "has high FFMPEG CPU usage" an the low-mid 20% range. Is this something I should be concerned about, or is this expected? I believe I have my camera settings and config.yml file set properly, but would really appreciate your feedback and suggestions. mqtt: host: 192.168.1.29 user: mqtt-user password: password #--------------------------------------------------------------------- ffmpeg: global_args: -hide_banner -loglevel warning -threads 2 hwaccel_args: preset-vaapi input_args: preset-rtsp-restream output_args: record: preset-record-generic-audio-copy #was = preset-record-generic-audio-aac retry_interval: 10 #-------------------------------------------------------------------- detectors: coral: type: edgetpu device: usb #---------------------------------------------------------------------------------------- record: enabled: true retain: days: 7 mode: motion events: retain: default: 30 mode: motion #---------------------------------------------------------------------- snapshots: enabled: true retain: default: 30 #-------------------------------------------------------------------------------- go2rtc: streams: basement: - rtsp://admin:[email protected]:554/cam/realmonitor?channel=1&subtype=0 # - ffmpeg:basement#audio=aac basement_sub: - rtsp://admin:[email protected]:554/cam/realmonitor?channel=1&subtype=1 # - ffmpeg:basement_sub#audio=aac #----------------------------------------------------------------------------------- cameras: basement: ffmpeg: output_args: record: preset-record-generic-audio-copy #was = preset-record-generic-audio-aac inputs: - path: rtsp://127.0.0.1:8554/basement input_args: preset-rtsp-restream roles: - record motion: mask: 0.732,0.023,0.732,0.081,0.969,0.088,0.966,0.025 basement_sub: ffmpeg: output_args: record: preset-record-generic-audio-copy # was = preset-record-generic-audio-aac inputs: - path: rtsp://127.0.0.1:8554/basement_sub input_args: preset-rtsp-restream roles: - detect # - audio # <- only necessary if audio detection is enabled motion: mask: 0.712,0.018,0.712,0.097,0.982,0.105,0.973,0.018 #----------------------------------------------------------------------------------- version: 0.14 camera_groups: Basement_Group: order: 1 icon: LuArrowBigDownDash cameras: basement Thanks in advance for your input, and for this great application.

Will it work? Probably. Will you be happy with the result, after the money, time and effort spent? Most likely not. I had to look to see if the 7800X3D even has an integrated GPU. It does, but it is as cut down, basic iGPU that AMD has ever offered. As Plex has only recently even supported AMD transcoding (and what some quick search of recent comments on the subject say) I see nothing but disappointment ahead.

Can you click on the "67 enabled" link, then select the unchecked stations? IIRC, if you do a rescan, Plex only selects the channels that were selected during a previous scan and does not automatically add any new ones.

A 3900X is more than powerful enough to CPU transcode almost any format to another, for a single stream on a local LAN.

Always a hard question to answer, as it depends on the details. "Can my car take me to the airport?" vs "Can my car take 20 people and their luggage to the airport?" Same car, same destination, completely different task. For a media server, it all depends on how many streams you are transcoding at one time, and the formats you are transcoding from > to. I used a 1500X (4c/8t) in my original Plex server with no GPU transcoding, and it worked fine on my home LAN. Your 3900X has much more horsepower. AMD GPUs aren't the best choice for media transcoding. They weren't supported by media server apps until just recently, and still are a work in progress. If you can sell it, you would be much better served getting a Quadro T400 (or even its predecessor, P400). It consumes much less power (TDP 30W vs 180W) and is well supported in all apps that perform transcoding. You hardware is very solid for the rest of your desired tasks. Unraid for NAS and Docker can run on most any x64 setup (if not pushed to extremes). It is when running VM that the hardware needs go up. Let's use a Win11 VM as an example: Visualize this as there are now two computers in your server's case - one running Unraid, one running Win11. You allocate CPU cores, DRAM, and storage to the VM. In simplest terms, Unraid no longer sees what is given to the VM, the VM no longer sees what is Unraid. With your 3900X being 12c/24t) you have plenty of threads to allocate to the VM (nobody wants to work on a 1c/2t 4GB Windows machine). As for power usage, your hardware would be about average for a typical server (especially if you ditch the AMD GPU). Most of the time the box sits idle, so not consuming much power at all. You will find that the spinning drives in the server account for the majority of your power usage.

I've always found QDirStat to be one of my best go-to utilities. It's available in CA.

I'm not much of a database guy. Probably a good place to start looking for answers would be in the support thread for the docker container you are using. On the Docker tab, click the docker's icon and the Support.

If I may ask, what do you have in appdata that is 800GB large? Mine is 1/10th of that (81GB) and I have 36 Docker containers installed, 27 which run 24/7/365. The conventional thinking for appdata is it contains configuration/logs and static resources needed for the container to run. Having a backup of this is most useful not only for a major failure (such as a dead drive) but to recover if a container gets messed up by either human error (a reconfiguration gone wrong) or by automation (a bad update, for example). Many times replacing a containers appdata folder puts things back to how they ran before. If what is making your appdata so large is working data (such as media files for Plex, or photos for Immich/PhotoPrism), those would likely be better off being mapped to another share, somewhere outside of the appdata share. It would then also allow you to use one of many backup options, where an incremental backup is done. In either case, the best methodology is that any docker container that modifies the data being backed up is stopped before backing up the data. Modifications done to the backup's source dataset during backup will make any verification check fail, as well as potentially corrupting the backed up data by it not being a true snapshot in time of that data.

I have a Kingston HyperX 3K 120GB SSD in my server now, which I bought in 2012. It has lived in many different systems over those years. Originally a Windows boot drive for my daily use PC, it now is used for transcoding space for Plex in my server. I just pulled up the SMART for it, now at 66971 hours (7.6 years). It has written a little more than 21.6 TB of data, which is its full capacity 180 times. No errors, no retired blocks. The little bastard will probably outlive all of us. 😆

Thanks for the quick response and your feedback. My plan is to get this installed this week, and your reply boosts my confidence.

The Appdata pluging backs up the appdata share and flash drive. It is a very configurable plugin, where you can select which docker appdata folders to back up, any additional folders, and exclude files/folders in any specific docker appdata folder. The Docker templates are stored in your flash drive, and (obviously) the Dockers are stored in their repositories (Github and such). Realize that the "cache" pool is just another pool. If you have data outside of your appdata share, you can use the same backup strategy you would use to back up any other data.

I'm preparing to do a fresh install Frigate on my backup server (so nothing to carry over or save). My thoughts are to start directly with v0.14.0-rc2, as to avoid any re-configuring once this release candidate is a stable release (and also enjoy the new features). I know I can set the Repository to ghcr.io/blakeblackshear/frigate:v0.14.0-rc2 to pull the 0.14 release candidate. The question(s) I have is this all I need to change in the Docker template currently in CA or are there other mappings/changes needed? Or is there another template that I should use? These are questions that came to me reading a post from you, from back in June:

search CA for "appdata". There are several versions of the backup plugin, due to changes in the maintainer of the plugin and Unraid updates.

I love reading how the sausage is made. Thanks for sharing, and for your outstanding work as well as your continued dedication. (Sometime a guy needs to hear that what they do is appreciated...)

You are stopping Plex, but do you possibly have some other program which may be updating the Plex database? If so, you can/should stop all of the related Dockers together.

Look at the Docker in the Apps tab. The comments from the author will be listed there.

Yes, I've seen the BIOS setting sometimes with 2 choices (Power Off, Power On) as well as with three (Off, On, Last State). Depends on the model of CPU/Motherboard chipset and probably MB manufacturer.

Looks as your UPS has something call Sleep Mode, which will automatically power back on the UPS AC outlets once the battery charges. Not every function of every UPS ever made is in any software. As for the server coming back up, check your BIOS settings. There usually is a section for what can power on a system (Keyboard, Ethernet, PCIe card, etc). Usually has a setting to either power the system up when power is applied or when power button is pressed.

Not an issue with SanDisk directly. There are a number of "questionable" look alike flash drives out on the market, many coming from the big internet retailers.