Those jsut thermal limit the GPU - it is similar to wattage capping, but instead caps based on the temp of the cpu instead of setting a hard limit of the wattage. So if your cooler is good enough, it’d still use the same amount of power. Similar in theory, but slightly different than what I was talking about.
Either way - don’t worry about it too much, your temps are just fine & require no additional hardware. Changing wattage caps though could lower your power consumption though & might be worth looking into further.
this is the results. nas powered on, with docker containers running, and with vm fedora on with a video playback active.
Not sure about power consumption, i will have to check that when i am able to power off stuff from rack
and this is temps when fedora vm stop video playback (vm is still running). the temps flatten out near the tail end, so most of the time is will be under 50c most part unless i do stuff with fedora vm, or jellyfin is doing heavy indexing or some similar activity and even then it won’t be for a long duration
That’s unlikely because “X670” is physically two B650 chips, daisy-chained. More power consumption is to be expected—as well as extra latency, extra contention and increased oversubsribing of the x4 link between the (pair of) chipset(s) and the CPU. All AMD5 server boards I know of use B650, or no chipset at all; none goes for X670.
The solution would be to use a server-style GPU, with power connectors at the end of the card rather than at the top.
still figuring out how to fix the minor graphics and audio delay… its minor but it’s perceptible enough to make the experience not as smooth as i would like.
i added 2 cores, 4 threads and 8gb ram. was that not enough?
tested the youtube reco, but the rdp didn’t work well. Video playback smoothness was a mix bag. It played without the lag, but i couldnt pause, or exit the video player. unresponsive.
also bluetooth was not working either.
so remote viewer though not perfect, worked more reliably despite the micro stutters every few seconds for a milisecond pause/stutter effect for video playback and audio.
I second this. Rsync makes sense in some applications but it will be way slower than a replication because replication makes use of snapshots (as should you). Snapshots automatically gather everything that changed, was added, etc. while rsync literally has to traverse every single directory and manually check every single file to see if something was changed.
Worse, unlike replication, there is no rolling back to a previous snapshot with rsync. Just stop using rsync and start using replication. Once replication works with standard SSH, upgrade to replication with netcat enabled.
was fixing dockers, npm, fedora etc. so many things
to do this i would first have to setup my 2nd truenas wouldnt i? problem is that will take a while. ordered part and moving stuff. will take time
important thing is i got a backup already done and tested. so i can try replication when i get a chance.
plz keep in mind, my backup isnt on 24/7 … after i manual backup, i turn it off… so im not constantly snapshot syncing all the time. so even rsync isnt something i do often. usually it doesnt take long to backup, so was never really an issue for me. though i see that snapshot is more efficient for sure.
atm im mostly testing fedora vm on truenas.
*update
actually i may decommission one of these other qnap box i have. and put the drives in a prepped spare truenas (653a) i got and use that for the ts-877 to test replications so i dont have to worry if my stuff gets messed up or not. ill do that later.
also for firefox, apparently you needed to use the one by flathub which has the codecs come with it. so i removed fedora and added that, now the youtube on firefox is a bit more smoother. at least compared to before the video lipsyncs with audio when before it didn’t. big improvement.
so either remoteviewer isn’t good enough though it comes close, and maybe i need to connect monitor directly to nas to fix the issue for good
also i may have made a mistake. when setting up the vm i think truenas used some default amd driver? but may have accidentally removed it to replace with the ones i downloaded from shell fedora. not sure.
i checked truenas vm, there is no rollback or reset for this.
*update
added VRR support. apparently its not enabled by default in fedora 41 workstation. i couldnt enable it
there i added snapshots. i use owls youtube, but he set daily retention, but i left mine the default keep 2 weeks for lifetime, and take daily snapshots.
the 2nd part of his snapshot explanation was the snapshot replication. ill set that up later. cauz i have to configure the other 2 nas when i’m able to
resilver priority i left default disabled. not sure what to do there or if needed
the reason why the img looked bad, was when u upload to imgur they probably over compress it till it becomes awful. also the same with the memegenerator
ooo according to your youtube found out how to create the snapshots for the vm. before i was using timeshift within the fedora os. guess i should using the truenas method instead.
u also answered a question i was wondering about.
i made the snapshots for each dataset, whereas you snapshot the whole pool but used exclusions for datasets not to snapshot for
some confusion here
i got 2 pools
tank and tank2
tank 1 has 5 datasets. i only want to backup 4 out of the 5. But i still am okay making snapshot locally for one of the excluded snapshots.
tank2, has 2 datasets. 1 for docker, the other for jailmaker. i only want to backup docker, but not jailmaker. But i also want to still create local snapshots for jailmaker.
But based on your youtube, it said create snapshot based on pool, and exclude the datasets you dont want. But as u can see i have some further requirements for this so not sure doing that would fit with what i need.
So does that mean for my situation im better off just create the snapshots for each dataset separately rather than just only the pool with exclusion?
or maybe im over complicating it.
just data snapshots the 2 pools minus any i dont want to.
then replication, here i can exclude what not to backup up via replication. i will try that later i was just worried it can only backup the entire pool which is not what i wanted.
jailmaker wouldnt start on reboot. Apparently, when i recovered from backup, i didnt create the dataset i previously had within docker.
this was already created /mnt/tank/docker/
but the problem was within docker, there were other sub datasets like data and stacks e.g.
/mnt/tank/docker/data
/mnt/tank/docker/stacks
so copying the folders to another dataset temporarily (i used winscp but u can do the same in shell if u know the commands), then creating those dataset names, then moving them back to there from the copies. Then delete the temporary dataset (use truenas ui to delete the no longer needed temp dataset to hold the copies) now auto start for jailmaker works on reboot.
just in case anyone had this problem but couldnt figure it out
tested backing up windows 11 system to the new truenas. I used aomei backupper since i got a professional license for it. this particular software supports pointing the backup location to a nas (tested for truenas)
i’m pulling 584 MB/s from desktop to the new nas.
It says about 35minutes to backup 1.2tb
So yeah not quite 10g but at this speed, it was worth it for me but it’s not often i move lots of data like this, just that when you need it, you will be glad you do.
i’m backing up before i update windows just in case
cpu, hdd/ssd temps everything is fine after using the new build for a week or so. though i had to tinker in motherboard to make it such, no big deal.
during the backup (pushing as much throughput as it can) cpu usage is about under 5-20% or less. temps remain below 60c. During idle its usually 40-50c.
