OrdinaryButt

Adding 8125 (Trendnet) adds exactly 4 watts to the kill-a-watt measurement at the tap. The problem is, the card prevents Packaged C-states - with it, I am stuck at C3; with card removed, it goes to C8. 4 watts may not be alot, but put it another way, it is 25% of this system usage (no HDD connected yet). 13 vs 17 watt at idle. If you share how to enable ASPM in the driver (I might find it by the time your respond, still reading), I can test to meassure the actual difference with ASPM on vs off. Thanks

I can confirm that ODD do not count towards the license limit on 6.12. ODD will not show up under list of devices, nor will be discovered by Unassigned device (at least for SATA, didn' try USB). You can check if ODD is detected by sending following console command: lsscsi -g Also, in order to pass ODD to container, you will likely have to adjust permissions. This comment explains how to do it for beginners:

Thanks, this is still required in 6.12 and does work. What is/where is "go file" ? EDIT: Found it. For the inexperienced user: Terminal to server change directory to boot / config ('cd /boot/config') send 'ls' command to confirm 'go' is listed then edit the file, for example 'nano go' Add comment (begins with #) and then parameters as listed above. Exit and save with Ctrl+X Reboot server to test.

I felt like you might be lacking some understanding, so I will explain like I would my kids. Ignore if I felt you wrong. ASM1166 with PCIe x1 will have a max theoretical bandwidth (limited by PCIe) of 8gigaBIT/s or 1gigaBYTE/s (1000megaBYTE/s). 1000/6 and considering overhead, you will get a maximum speed of 150megaBYTE (this is the measurement unit you typically see when you do file transfers in Windows environment). Modern HDDs can go upwards of 250MB/s, so the x1 PCIe slot will be a bottleneck if you have all 6 drives read/write at max capacity at the same time (parity check for example). With SATA SSD, you will reach that limit with 2 units. If you would be getting the ASM1166, it better have x4 PCIe (that would be 1000MB/s x4). This will allow you to have up to 6 SATA SSD on that card with no bandwidth limitation at all (provided your motherboard x4 or larger PCIe is electrical x4 or larger; physical dimension of the slot doesn't always guarantee the speed)

I get that. Have you seen any tests showing that the ASM1164 is indeed faster than 1064? Why would they allow the 1164 to be also x1, if there is 1064 for that? Makes no sense.

Well, in the opening post you do say that you have seen 1164 with x1, and ASM says 1-, 2- (see screenshot) what ever it means. I am looking for actual data that shows - is 1164 faster than 1064 or any other differences. Their web pages shows minimal details as each is listed as 25mhz Crystal https://www.asmedia.com.tw/product/17fYQ85SPeqG8MT8/58dYQ8bxZ4UR9wG5 https://www.asmedia.com.tw/product/A58yQC9Sp5qg6TrF/58dYQ8bxZ4UR9wG5

Does anyone know which ASM supports ASPM? Having hard time finding concrete information except for ASM1166 (positive reports that it does). I only need a controller with 4 SATA, least amount of power draw. EDIT: on ASMedia site, they say all 1064/1164/1166 Support L0s/L1/L2.... This I guess answers my original questions. What I can't find is the difference between 1064 and 1164.

yes, that's the one I meant (forgot the link). I completely missed the fact that the m.2 heat sink/fan is separate from the main sink. In this case, shouldn't adding a M.2 to 5x Sata ports adapters (2) be decent option? The bottom ports can be populated with m.2 for Cache, and you got 10 sata ports from the top 2 m.2. It's enough for me. Everyone focuses on how AMD is more power efficient than intel, However, it seems like AMX AM5 chipsets are twice more power hungry than intel at idle/low CPU tasks (chiplets draw more power). My server is probably at near idle 20 hours a day, so the low load power is more important than full load. EDIT: Personally, the Minis forum approach with regular 20/24 pin power connector fits NAS server better - ATX power supply can be used for the entire machine. Many of the above mentioned boards have only 12/19v DC jack power input, which would require some funky stuff to do for HDD power.

Duh, forgot to add the link https://store.minisforum.com/products/minisforum-ar900i

Did you guys see this? Got exited with the 4x M.2 slots, then I realized 2 are on the bottom and 2 are covered with the heat sink. May be they can be "extended" with cables? A bit steep at $560 but the assumed to be cheaper version with 13650HX isn't available yet (price unknown).

Any updates on this topic?

I run into write permission issues while installing a grafana as part of compos bundle of containers. Was able to "resolve" by adding 'user:"0" ' (root) to compose file. I feel like this is a bad practice, but I don't know if I am actually compromising anything by doing that. Anyone knows if this is a potential week point and if there might a different way to solve it?

Can someone with E3C246D4U2-2L2T confirm that when M.2 is populated with PCIe SSD, all 8 SATA ports are available? The manual only states that when M.2 SATA SSD is populated, one of the SATA ports is disabled (shared with M.2 sata), but there is no chart to show that M.2 PCIe SSD + 8 SATA ports is supported (like the Gigabyte version has). Also, the manual of the smaller ASRock board, C246M WS, does say that even PCIe M.2 disabled one of the SATA ports. Do the 10G ports on the E3C246D4U2-2L2T support 2.5/5G speeds as well? Here is what Intel says for X550-AT2 NIC: Data Rate Per Port 10/5/2.5/1GbE (NBASE-T in Linux Only)

Thanks for the tips!

What if older version is removed from repo? I have trust issues with "the cloud"

I might be wrong here and probably would like a dumb ass, but Appdata doesn't seem to be containing the ...how should I say this... the "installation files" (excuse me for the dirty Windows comparison, still learning Linux), with the Appdata being the config folder. I am guessing those are in the "docker image". Not really sure, I see some of my containers have many files and folders, some such as GluetunVPN only has 2 files in it's appdata folder. So, to match the topic's subject line, how can I go about backing up a docker container entirely, say before update of container, and if the updates breaks something (often for me with my luck), restore the previous version? I am thinking about switching my cache drive to ZFS and adding another ZFS pool, so I could take such backups perhaps.

I don't think you got my point, I see now that it nothing to do with you containers either. The guy that came up with that name "noVNC - a VNC Clinet..." has no clue how much confusion his chosen name will cause. Could have labeled it hVNC, htmlVNC, h5VNC...anything but noVNC.

Thanks. I did search for "PORT" and similar but that yielded way too many results. Got it changed. I am eager to ask why is it labeled NOVNC_PORT (and default No VNC WEB UI) when clearly VNC is in use, but I am not going to. Got enough info today 😎

I think I figured out what is going on with Gluetun. The recent update seems to be doubling some of the parameters("VPN_SERVICE_PROVIDER: *" for example, added at the end) in the template and it was trying to open second connection (likely). I end up removing the duplicates and all is good, except there was also a non working server. Hope that helps.

Thanks. I ended up fixing the gluetun VPN as that does it automatically. Another question - anyway to change web port on your Chromium or Ungoogled-Chromium containers (or is that a question for someone else)? Not seeing it in the documentation. Note, not changing the port map but the actual web port of the application (e.g. instead of 8080, to something else). From what I understand, if I have another container with same port (also no way to change that I can find), I can't run both thru the same VPN container network as there will be port map conflict.

Thanks. I was able to change this application, as developer provided a way to do that via config variable. In case that is provided/documented (I am not seeing one for Chromium, 8080 default), I basically have to build from source, right?

Before I even go that far, I decided to test if it will recover outage. Rebooted my network router (not Unraid server) and when gateway/internet traffic was restored, OpenVPN-client did not recover with these messages in the log for over an hour (had to restart the container eventually). I could not ping 8.8.8.8 in this condition from the contained console, ping successful in normal conditions. It is a standard install with default --dns 8.8.8.8. Is this a good place to troubleshoot/request help?

Does the OpneVPN-Client/ICH777 have a "kill switch" ? I am using other containers routed thru it, with the --net=container:openvpnclient method. Just want to make sure if VPN connection drops (but not internet), sub containers will not be able to get out. Any way to test this?

I also run into this today. Strangely enough, works with a server location "Bulgaria", but not the "Netherlands". OpenVPNclient works fine with Netherlands. My provider is FastestVPN. Not sure what the deal is.

Hello, Say I have container A that uses container port 8080 and container B that also uses same port (or second instance of cA). I understand how to re-map the ports when using network=bridge; cB gets host port 8081, and the system maps (forwards) that to container on 8080. However, when I want to change both cA and cB thru openVPN container network, I am not able to achieve that. I've tried everything I could think of, either I am getting container init error - port duplicate, or somehow both applications reacting on same port, or on the Docker tab the Unraid spinner never stops and the VPN containers restarts forever. Is this possible, and if not, any lead on how can I modify the container A or B so internal port is different?