💬 OT 💻 Post your computer setup
- Thread starter DanJ
- Start date
Everything arrived, except for the Torque Wrench (yes, this needs a torque wrench for the mounting screws), and the mounting kit for the HSF.
The torque wrench is 1.5NM, or about 14.something inch-lbs, aka 1.1ft-lbs.
As the HSF won't arrive till the end of this month, I chose to buy the torque wrench so I have it on hand and don't have to think about tightening each one that little bit at a time. I could probably start finger tight and guess work it. But overdo it and you'll either bend the pins, or crack the socket on the Mobo.
However, just to make sure I had HSF clearance, I lightly assembled everything. I was concerned that the HSF would possibly block the RAM, or be in the wrong orientation. But all is well.
The EPYC CPU is HEAVY. Like shockingly heavy.
The Mobo has 1x 24PIN (or 20+4), and 2x 8PIN adapters.
This leaves 2x 8pin adapters free if I ever decided to add a video card for hardware transcoding, or decide to install my RTX 3080 Ti in it and visualize my gaming box.
The torque wrench is 1.5NM, or about 14.something inch-lbs, aka 1.1ft-lbs.
As the HSF won't arrive till the end of this month, I chose to buy the torque wrench so I have it on hand and don't have to think about tightening each one that little bit at a time. I could probably start finger tight and guess work it. But overdo it and you'll either bend the pins, or crack the socket on the Mobo.
However, just to make sure I had HSF clearance, I lightly assembled everything. I was concerned that the HSF would possibly block the RAM, or be in the wrong orientation. But all is well.
The EPYC CPU is HEAVY. Like shockingly heavy.
The Mobo has 1x 24PIN (or 20+4), and 2x 8PIN adapters.
This leaves 2x 8pin adapters free if I ever decided to add a video card for hardware transcoding, or decide to install my RTX 3080 Ti in it and visualize my gaming box.
I wanted to do more with the existing server. It just can't do it. Not enough memory bandwidth or cores to handle hosting VMs. Specially not if I want to host things like an SCCM environment, or MBAM, while testing out OSD Deployments and all sorts of of other server things.
I was considering 128GB, but changed my mind at the last minute to 192GB so I could turn on DeDuplication in ZFS or other Memory Hungry things.
I've had the storage for ages. PLENTY of it. Just not enough Pah to do anything with that storage.
Pulled the trigger on a wd-sn850, got it cloned using their software and it shows up in bios no problem and looks right but when I try and boot it just shows a black screen but if I move the mouse the curser shows for a second then goes away. This is a new one for me 
Thirdgen89GTA when you're talking about virtualizing your gaming rig, would you then run your monitor and keyboard directly to the server or would you use less powerful PC in-between and use RDP or something of the sort?
No RDP or anything like that.
The VM would run on the host, and I'd be passing USB, GPU, and other hardware PCIe devices directly to the guest VM, giving them exclusive access to the hardware. So I'd just plug the DisplayPort cable directly into the GPU, same for USB. Other things would get handled by native VirtIO drivers which allow full speed access to system resources.
PCIe Pass-thru is a wonderful thing.
Well that was a basically 100% useless failure to migrate from TrueNAS Core to Scale. Just spent total of 6 hours today trying to convert from Core to Scale. It did not go well. Permissions all fucked up, could not get a NFS mount to work AT ALL. SMB Permissions super whack too.
Thankfully, smart me has snapshots hourly, daily, and weekly. So I just rolled back to the snapshots from 12AM this morning, and wiped the boot disks, installed Core, and re-uploaded my TrueNAS Core config file.
All back to the way it was at 9AM this morning.
With the way this went, I'll just snapshot, export the config, and do the upgrade to EPYC hardware while keeping Core. Then if that is all working well for a week or so, the following week I'll destroy it all and build a new TrueNAS Scale from scratch.
Thankfully, smart me has snapshots hourly, daily, and weekly. So I just rolled back to the snapshots from 12AM this morning, and wiped the boot disks, installed Core, and re-uploaded my TrueNAS Core config file.
All back to the way it was at 9AM this morning.
With the way this went, I'll just snapshot, export the config, and do the upgrade to EPYC hardware while keeping Core. Then if that is all working well for a week or so, the following week I'll destroy it all and build a new TrueNAS Scale from scratch.
I've been having this really weird issue of my computer not turning on when I press the power button. Or, it turns on for a split second and immediately turns back off. Almost like the PSU was discharging on boot attempt and inadvertently shutting the whole thing off. I was about ready to drive an hour to Micro Center and have them test the PSU and components. I have a Power PC and Cooling 750W that has gone through 4 iterations of builds and is still rock solid according to HWInfo (talking <1% fluctuations on any rail, 10 years and 4 builds later).
I built this Ryzen 5600X setup specifically because I thought my 4790k build was dead. Well, now I have the exact same issue I was experiencing except now with this 5600 build. I finally got it to a point it would turn on but it would sit at the Windows loading screen forever and just spin. I booted into BIOS and I noticed a drive missing, my "brand new" 860 Evo which is my gaming drive.
Turns out, the SATA data cable going to that drive was the cause of this whole fucking "no boot" issue. Including my 4790. I pulled the cable out of the drive and a bunch of gunk came out with it. Cleaned the drive-side of things as they looked fine, replaced EVERY SATA cable to all 3 of my drives, and the computer boots up before I barely even touch the power button now.
I was sitting here ready to drop another $200-250 on a new PSU and case (that Fractal Meshify C is calling me) all because of a cable I've used, for the same amount of time as my PSU, eventually started corroding.
I should go get my 4790k and see if that works. I bet a 3 dollar bill it's totally fine and will boot.
I built this Ryzen 5600X setup specifically because I thought my 4790k build was dead. Well, now I have the exact same issue I was experiencing except now with this 5600 build. I finally got it to a point it would turn on but it would sit at the Windows loading screen forever and just spin. I booted into BIOS and I noticed a drive missing, my "brand new" 860 Evo which is my gaming drive.
Turns out, the SATA data cable going to that drive was the cause of this whole fucking "no boot" issue. Including my 4790. I pulled the cable out of the drive and a bunch of gunk came out with it. Cleaned the drive-side of things as they looked fine, replaced EVERY SATA cable to all 3 of my drives, and the computer boots up before I barely even touch the power button now.
I was sitting here ready to drop another $200-250 on a new PSU and case (that Fractal Meshify C is calling me) all because of a cable I've used, for the same amount of time as my PSU, eventually started corroding.
I should go get my 4790k and see if that works. I bet a 3 dollar bill it's totally fine and will boot.
Making a mistake in diagnosing a problem, meaning you buy upgrade stuff when you don't need it is the best kind of reason to upgrade lol.
Oh yeah. Upgrade happening this weekend. I'm Excite.
Dats a big cooler.
Dats a big cooler.
Okay, so last night was hell. What didn't go wrong?!?
First off, the board arrived with an old BIOS, really old, like 1.0 old. This was a possibility. So I knew I'd have to update the BIOS. Except, there were two problems preventing it from booting and completing the BIOS upgrade.
Due to the BIOS being old, it did not recognize my CPU or RAM. It reported my 24 core monster as a paltry 8 core EPYC 7232p.
Problem #1: 1 of the 6 32GB modules was DoA. So I only had 160GB of ram (poor me lol).
Problem #2: Apparently I cannot install either of my HBA adapters into the PCIe 8x slots. Doing so causes a POST problem and the system locks up before ever booting. Because it did not boot, it did not think the BIOS update was successful. Which it was.
Problem #3: My Rail kit took a shit, it warped and the rails are now fucked. Further, looks like NORCO went out of business, so I can't get a replacement rail-kit. My only option is an aftermarket set. I hope I can unrack it when the time comes and that its not stuck in the rails.
I may look into replacing this chassis with a Supermicro 4U chassis. I'll keep an eye on Ebay and see if one shows up for decent. Usually they are $700-$1000 for an empt system. But it would certainly be nicer quality and feature wise than the NORCO is.
So, after playing both Musical Memory Module to verify the bad RAM stick, and having to diagnose the POST issue, I spent nearly 6 hours last night sorting it out. I went to bed at 4AM this morning BTW.
But here I am, this monster system is now running. Just for shits & giggles, I threw a 4K transcode at it for about 15 minutes. This is the CPU usage graph. Yellow line is 100%. However, measures each core as its own 100%, so technically 100% real usage gets reported as 4800%.
Damned thing didn't even BLINK at the load. Its barely legible on the graph.
I can tell you one thing that DID blink. The Power usage. Damned system idles at 280watts now, before it was 140w. EEEK.
First off, the board arrived with an old BIOS, really old, like 1.0 old. This was a possibility. So I knew I'd have to update the BIOS. Except, there were two problems preventing it from booting and completing the BIOS upgrade.
Due to the BIOS being old, it did not recognize my CPU or RAM. It reported my 24 core monster as a paltry 8 core EPYC 7232p.
Problem #1: 1 of the 6 32GB modules was DoA. So I only had 160GB of ram (poor me lol).
Problem #2: Apparently I cannot install either of my HBA adapters into the PCIe 8x slots. Doing so causes a POST problem and the system locks up before ever booting. Because it did not boot, it did not think the BIOS update was successful. Which it was.
Problem #3: My Rail kit took a shit, it warped and the rails are now fucked. Further, looks like NORCO went out of business, so I can't get a replacement rail-kit. My only option is an aftermarket set. I hope I can unrack it when the time comes and that its not stuck in the rails.
I may look into replacing this chassis with a Supermicro 4U chassis. I'll keep an eye on Ebay and see if one shows up for decent. Usually they are $700-$1000 for an empt system. But it would certainly be nicer quality and feature wise than the NORCO is.
So, after playing both Musical Memory Module to verify the bad RAM stick, and having to diagnose the POST issue, I spent nearly 6 hours last night sorting it out. I went to bed at 4AM this morning BTW.
But here I am, this monster system is now running. Just for shits & giggles, I threw a 4K transcode at it for about 15 minutes. This is the CPU usage graph. Yellow line is 100%. However, measures each core as its own 100%, so technically 100% real usage gets reported as 4800%.
Damned thing didn't even BLINK at the load. Its barely legible on the graph.
I can tell you one thing that DID blink. The Power usage. Damned system idles at 280watts now, before it was 140w. EEEK.
get starlinkOh lawd dats beefy.
My have to get my own local server setup, especially when this house goes through and my internet consists of 30Mb DSL.
I did have fairly nice ball bearing RL-26 rails for the case. But they be broke now.
I ordered a set of rails, but it cost me $70. Ouch.
its amazing how expensive fucking server rails are. its nuts.I did have fairly nice ball bearing RL-26 rails for the case. But they be broke now.
I ordered a set of rails, but it cost me $70. Ouch.
They weren't even that high of a quality, just that they were made for the Norco 4224 chassis. However, Norco be out of business now and I can' find a set anywhere. I honestly never expected the rails of all things to fail.
Okay,
So, I'm getting in here, and the performance of this server is just out-right INSANE. Incredibly, ungodly fast. Now that it has everything of Plex cached in memory since its been running for a few hours, its really, REALLY snappy.
Also, I hooked up the 10Gig-E line direct between the MacMini, and the Plex.
DAMN boy, look at the transfer speeds go! 1.25GB/s peak. Averaged around 1.1GB/s. I can now edit directly from the mini if I want.
But what will really happen, is now that I've got enough memory to run a Windows guest on the server, I"ll probably attach my USB stuff to it, pass-thru, work directly off the server in a VM.
So, I'm getting in here, and the performance of this server is just out-right INSANE. Incredibly, ungodly fast. Now that it has everything of Plex cached in memory since its been running for a few hours, its really, REALLY snappy.
Also, I hooked up the 10Gig-E line direct between the MacMini, and the Plex.
DAMN boy, look at the transfer speeds go! 1.25GB/s peak. Averaged around 1.1GB/s. I can now edit directly from the mini if I want.
But what will really happen, is now that I've got enough memory to run a Windows guest on the server, I"ll probably attach my USB stuff to it, pass-thru, work directly off the server in a VM.
Thread Info
-
Created -
3KReplies
-
83KViews
-
Participants list
