It's been a long time since I've encountered this kind of obtuseness. This is not new, but I thought that this kind of blindness was in the rear-view mirror. I guess not.
One of their motherboards keeps locking up on me. After some back and forth their last reply ended the discussion thread with a:
# Since we do not fully support and validate Linux, we cannot offer full # support on Linux. Please try a Windows base OS to check if you are still # experiencing the same instability issue.
Just thought I'd mention it here, so that other DIYers might take this into consideration. Looking at their support page, it's listing new BIOS updates every month.
On 5/16/22 16:02, Sam Varshavchik wrote:
It's been a long time since I've encountered this kind of obtuseness. This is not new, but I thought that this kind of blindness was in the rear-view mirror. I guess not.
One of their motherboards keeps locking up on me. After some back and forth their last reply ended the discussion thread with a:
# Since we do not fully support and validate Linux, we cannot offer full # support on Linux. Please try a Windows base OS to check if you are still # experiencing the same instability issue.
Just thought I'd mention it here, so that other DIYers might take this into consideration. Looking at their support page, it's listing new BIOS updates every month.
Hi Sam V.,
If a motherboard can run Windows, it can run Linux. Not necessarily the other way around though. Windows has artificial blocks (8 & 9 gen processors and TPM modules).
What you are experiencing is the atrocious customer service since the work-at-home started. Gigabyte and Intel a YUGE offenders. You got a "blow off" response. Anything to NOT answer and NOT help. What is unusual is that GIGABYTE actually answered back.
I have a question into them as to the replacement for their W480M, which is no longer produced. The lazy so-and-so sent me back the advertising brochure for the W480M. When I called him on it, he would not budge. I did figure it out (C246M-WU4).
When I asked Intel which processors supported on that motherboard supported ECC, they sent me all the processors. When I called him on it, he also would not budge. Ya, i3 and ECC, my aunt fanny. (I did eventually figure it out: it is their new "e" series xeon processors.)
Business rule: relationships outlive transactions. These zero and blow off customer service companies better be darned careful or they will not have businesses after too long.
So anyway, as a occasional system builder, I have found Gigabyte's motherboards to be of pretty good overall quality.
Now providing you did not populate your motherboard with cheap memory, used processors, bargain SSD drives, here are some things to try.
1) download any of the Fedora Live Spins and dd them to a a 8+ GB flash drive. Boot off it. And torture the thing. If it does not freeze up, your hardware is okay, except for your hard drive. Then you dnf gsmartcontrol and test your hard drive with it.
https://spins.fedoraproject.org/ # dd if=spin.iso of=/dev/sd[x]
Sub appropriate names in the above.
2) if you are feeling really adventurous and have a spare hard drive kicking around, you could download and install Windows 10 (no TPM hassle)
https://www.microsoft.com/en-us/software-download/windows10ISO
and cut it to and 8+ GB flash drive with woeusb (Internet connection required):
# dnf install WoeUSB # dd bs=4096 if=/dev/zero of=/dev/sd[x] # woeusb --device Win10_21H2_English_x64.iso /dev/sd[x] --target-filesystem NTFS
or whatever the iso is called. You do not need a license for this. Windows will just not let you set a wall paper. Woopie-do.
If you do not feel like installing Windows on a spare hard drive, disconnect your Linux drive, fire up the Windows installer and jet let it sit there. See if you freeze up.
If it is your hardware, try substituting Kingston memory. Crucial if you absolutely you have to.
For SSD Drive, I only recommend Samsung. They just work. No drama.
And remember. You pay for what you get. Well, except for my advice!
:-)
HTH, -T
On 5/16/22 18:27, Tom Horsley wrote:
On Mon, 16 May 2022 18:17:53 -0700 ToddAndMargo via users wrote:
For SSD Drive, I only recommend Samsung. They just work. No drama.
But if you need to update firmware on a samsung ssd you have to run their windows only utility (at least the last time I looked).
I busted their chops on that one already.
On the other hand, I have never had to upgrade their firmware.
I have a bootable flash drive with DOS 6.22 on it for upgrading bios'es Supermicro's motherboards.
These guys can be rather shortsighted.
On Mon, 16 May 2022 18:32:29 -0700 ToddAndMargo via users wrote:
I have a bootable flash drive with DOS 6.22 on it for upgrading bios'es Supermicro's motherboards.
At least with motherboards all the ones I've had in the last few years can update directly from a usb stick with no dos/windows/anything required.
On 5/16/22 18:17, ToddAndMargo via users wrote:
On 5/16/22 16:02, Sam Varshavchik wrote:
It's been a long time since I've encountered this kind of obtuseness. This is not new, but I thought that this kind of blindness was in the rear-view mirror. I guess not.
One of their motherboards keeps locking up on me. After some back and forth their last reply ended the discussion thread with a:
# Since we do not fully support and validate Linux, we cannot offer full # support on Linux. Please try a Windows base OS to check if you are still # experiencing the same instability issue.
Just thought I'd mention it here, so that other DIYers might take this into consideration. Looking at their support page, it's listing new BIOS updates every month.
Hi Sam V.,
If a motherboard can run Windows, it can run Linux. Not necessarily the other way around though. Windows has artificial blocks (8 & 9 gen processors and TPM modules).
What you are experiencing is the atrocious customer service since the work-at-home started. Gigabyte and Intel a YUGE offenders. You got a "blow off" response. Anything to NOT answer and NOT help. What is unusual is that GIGABYTE actually answered back.
I have a question into them as to the replacement for their W480M, which is no longer produced. The lazy so-and-so sent me back the advertising brochure for the W480M. When I called him on it, he would not budge. I did figure it out (C246M-WU4).
When I asked Intel which processors supported on that motherboard supported ECC, they sent me all the processors. When I called him on it, he also would not budge. Ya, i3 and ECC, my aunt fanny. (I did eventually figure it out: it is their new "e" series xeon processors.)
Business rule: relationships outlive transactions. These zero and blow off customer service companies better be darned careful or they will not have businesses after too long.
So anyway, as a occasional system builder, I have found Gigabyte's motherboards to be of pretty good overall quality.
Now providing you did not populate your motherboard with cheap memory, used processors, bargain SSD drives, here are some things to try.
- download any of the Fedora Live Spins and dd them
to a a 8+ GB flash drive. Boot off it. And torture the thing. If it does not freeze up, your hardware is okay, except for your hard drive. Then you dnf gsmartcontrol and test your hard drive with it.
https://spins.fedoraproject.org/ # dd if=spin.iso of=/dev/sd[x]
Sub appropriate names in the above.
- if you are feeling really adventurous and have a
spare hard drive kicking around, you could download and install Windows 10 (no TPM hassle)
https://www.microsoft.com/en-us/software-download/windows10ISO
and cut it to and 8+ GB flash drive with woeusb (Internet connection required):
# dnf install WoeUSB # dd bs=4096 if=/dev/zero of=/dev/sd[x] # woeusb --device Win10_21H2_English_x64.iso /dev/sd[x] --target-filesystem NTFS
or whatever the iso is called. You do not need a license for this. Windows will just not let you set a wall paper. Woopie-do.
If you do not feel like installing Windows on a spare hard drive, disconnect your Linux drive, fire up the Windows installer and jet let it sit there. See if you freeze up.
If it is your hardware, try substituting Kingston memory. Crucial if you absolutely you have to.
For SSD Drive, I only recommend Samsung. They just work. No drama.
And remember. You pay for what you get. Well, except for my advice!
:-)
HTH, -T
Are all your fans working? Frozen fans on video cards are a YUGE cause of freeze ups.
On Mon, 16 May 2022 18:30:10 -0700 ToddAndMargo via users wrote:
Are all your fans working? Frozen fans on video cards are a YUGE cause of freeze ups.
So is the nouveau driver :-(. I run nouveau on every new fedora install until the video and/or system locks up, then I install nvidia drivers from rpmfusion and never have any additional video problems. (In the latest install nouveau can't run X at all and picks the wrong resolution under wayland, so I didn't run it very long :-).
On 5/16/22 18:43, Tom Horsley wrote:
On Mon, 16 May 2022 18:30:10 -0700 ToddAndMargo via users wrote:
Are all your fans working? Frozen fans on video cards are a YUGE cause of freeze ups.
So is the nouveau driver :-(. I run nouveau on every new fedora install until the video and/or system locks up, then I install nvidia drivers from rpmfusion and never have any additional video problems. (In the latest install nouveau can't run X at all and picks the wrong resolution under wayland, so I didn't run it very long :-).
Good point.
A live of Xfce or MATE would be a great test as neither uses Wayland
On 5/16/22 18:30, ToddAndMargo via users wrote:
On 5/16/22 18:17, ToddAndMargo via users wrote:
On 5/16/22 16:02, Sam Varshavchik wrote:
It's been a long time since I've encountered this kind of obtuseness. This is not new, but I thought that this kind of blindness was in the rear-view mirror. I guess not.
One of their motherboards keeps locking up on me. After some back and forth their last reply ended the discussion thread with a:
# Since we do not fully support and validate Linux, we cannot offer full # support on Linux. Please try a Windows base OS to check if you are still # experiencing the same instability issue.
Just thought I'd mention it here, so that other DIYers might take this into consideration. Looking at their support page, it's listing new BIOS updates every month.
Hi Sam V.,
If a motherboard can run Windows, it can run Linux. Not necessarily the other way around though. Windows has artificial blocks (8 & 9 gen processors and TPM modules).
What you are experiencing is the atrocious customer service since the work-at-home started. Gigabyte and Intel a YUGE offenders. You got a "blow off" response. Anything to NOT answer and NOT help. What is unusual is that GIGABYTE actually answered back.
I have a question into them as to the replacement for their W480M, which is no longer produced. The lazy so-and-so sent me back the advertising brochure for the W480M. When I called him on it, he would not budge. I did figure it out (C246M-WU4).
When I asked Intel which processors supported on that motherboard supported ECC, they sent me all the processors. When I called him on it, he also would not budge. Ya, i3 and ECC, my aunt fanny. (I did eventually figure it out: it is their new "e" series xeon processors.)
Business rule: relationships outlive transactions. These zero and blow off customer service companies better be darned careful or they will not have businesses after too long.
So anyway, as a occasional system builder, I have found Gigabyte's motherboards to be of pretty good overall quality.
Now providing you did not populate your motherboard with cheap memory, used processors, bargain SSD drives, here are some things to try.
- download any of the Fedora Live Spins and dd them
to a a 8+ GB flash drive. Boot off it. And torture the thing. If it does not freeze up, your hardware is okay, except for your hard drive. Then you dnf gsmartcontrol and test your hard drive with it.
https://spins.fedoraproject.org/ # dd if=spin.iso of=/dev/sd[x]
Sub appropriate names in the above.
- if you are feeling really adventurous and have a
spare hard drive kicking around, you could download and install Windows 10 (no TPM hassle)
https://www.microsoft.com/en-us/software-download/windows10ISO
and cut it to and 8+ GB flash drive with woeusb (Internet connection required):
# dnf install WoeUSB # dd bs=4096 if=/dev/zero of=/dev/sd[x] # woeusb --device Win10_21H2_English_x64.iso /dev/sd[x] --target-filesystem NTFS
or whatever the iso is called. You do not need a license for this. Windows will just not let you set a wall paper. Woopie-do.
If you do not feel like installing Windows on a spare hard drive, disconnect your Linux drive, fire up the Windows installer and jet let it sit there. See if you freeze up.
If it is your hardware, try substituting Kingston memory. Crucial if you absolutely you have to.
For SSD Drive, I only recommend Samsung. They just work. No drama.
And remember. You pay for what you get. Well, except for my advice!
:-)
HTH, -T
Are all your fans working? Frozen fans on video cards are a YUGE cause of freeze ups.
I almost forgot. When you freeze up, can you do a <ctrl><alt><f2> ?
To get back out, do a <ctrl><alt><f1>
ToddAndMargo via users writes:
Are all your fans working? Frozen fans on video cards are a YUGE cause of freeze ups.
It's a brand new Radeon 7750 card. I have not closely inspected its tiny fan, but everything's quiet on that front. The system hangs when it's idle, pretty much overnight. I keep the system doing nothing, and it freezes up randomly every 3-10 days. The video is frozen, no pings from the network, etc…
It's either this card, the Ryzen 3950X CPU, or the Gigabyte motherboard. I think I can exclude the RAM, I tried running with alternating two of the four RAM sticks, and it still hangs. I got a cheap Ryzen 3 CPU and now swapped the CPU, waiting to see what happens.
I need to prove this to my own satisfaction, but it has to be the crap Gigabyte motherboard. I should get a cheap Radeon 5450 card today, and after swapping that only the motherboard is left as the common denominator.
Their site shows Gigabyte pushing out BIOS updates every two months, on average. I just saw that they have a new one, since I loaded the latest one at the start of all this. Why? Why do you need BIOS updates six times a year? This is not a sign of a mature product.
Hi
I used to build workstations and servers for the Computacenter in St Albans England. 200 machines a night. I've just had a long session with a Gigabyte board and an NVIDIA card. A GK107GL Quadro K600 card which refused to work with anything. I tried Debian 11. Linux Mint. Arch Linux and Fedora 35. Every time I installed a system with this hardware it installed the wrong driver or the monitor was set at the wrong refresh rate. I nearly gave up and bought a different NVIDIA card. Finally got it to work with Ubuntu 22.04 which set the correct refresh rate and installed the correct driver which is the 470.103.01. It was a long head banging session to work that out. The Nouveau driver nearly worked but kept on locking up an crashing.
Richard
https://www.sheflug.org.uk/indexpage
On 5/17/22 03:13, Sam Varshavchik wrote:
ToddAndMargo via users writes:
Are all your fans working? Frozen fans on video cards are a YUGE cause of freeze ups.
It's a brand new Radeon 7750 card. I have not closely inspected its tiny fan, but everything's quiet on that front. The system hangs when it's idle, pretty much overnight. I keep the system doing nothing, and it freezes up randomly every 3-10 days. The video is frozen, no pings from the network, etc…
It's either this card, the Ryzen 3950X CPU, or the Gigabyte motherboard. I think I can exclude the RAM, I tried running with alternating two of the four RAM sticks, and it still hangs. I got a cheap Ryzen 3 CPU and now swapped the CPU, waiting to see what happens.
I need to prove this to my own satisfaction, but it has to be the crap Gigabyte motherboard. I should get a cheap Radeon 5450 card today, and after swapping that only the motherboard is left as the common denominator.
Their site shows Gigabyte pushing out BIOS updates every two months, on average. I just saw that they have a new one, since I loaded the latest one at the start of all this. Why? Why do you need BIOS updates six times a year? This is not a sign of a mature product.
How about <ctrl><alt><f2> or a <ctrl><alt><f3>
To get back out, do a <ctrl><alt><f1> or a <ctrl><alt><f2>
What shows on the monitor when it freezes?
ToddAndMargo via users writes:
On 5/17/22 03:13, Sam Varshavchik wrote:
ToddAndMargo via users writes:
Are all your fans working? Frozen fans on video cards are a YUGE cause of freeze ups.
It's a brand new Radeon 7750 card. I have not closely inspected its tiny fan, but everything's quiet on that front. The system hangs when it's idle, pretty much overnight. I keep the system doing nothing, and it freezes up randomly every 3-10 days. The video is frozen, no pings from the network, etc…
It's either this card, the Ryzen 3950X CPU, or the Gigabyte motherboard. I think I can exclude the RAM, I tried running with alternating two of the four RAM sticks, and it still hangs. I got a cheap Ryzen 3 CPU and now swapped the CPU, waiting to see what happens.
I need to prove this to my own satisfaction, but it has to be the crap Gigabyte motherboard. I should get a cheap Radeon 5450 card today, and after swapping that only the motherboard is left as the common denominator.
Their site shows Gigabyte pushing out BIOS updates every two months, on average. I just saw that they have a new one, since I loaded the latest one at the start of all this. Why? Why do you need BIOS updates six times a year? This is not a sign of a mature product.
How about <ctrl><alt><f2> or a <ctrl><alt><f3>
To get back out, do a <ctrl><alt><f1> or a <ctrl><alt><f2>
What shows on the monitor when it freezes?
No, everything's frozen, no response. It usually hangs overnight, with the monitor sleeping. No response from the keyboard, no pings from the network. The only thing to do is to hit reset and reboot.
Nothing gets logged in syslog. journalctl shows the normal noise, then dead silence.
On 5/17/22 04:46, Sam Varshavchik wrote:
ToddAndMargo via users writes:
On 5/17/22 03:13, Sam Varshavchik wrote:
ToddAndMargo via users writes:
Are all your fans working? Frozen fans on video cards are a YUGE cause of freeze ups.
It's a brand new Radeon 7750 card. I have not closely inspected its tiny fan, but everything's quiet on that front. The system hangs when it's idle, pretty much overnight. I keep the system doing nothing, and it freezes up randomly every 3-10 days. The video is frozen, no pings from the network, etc…
It's either this card, the Ryzen 3950X CPU, or the Gigabyte motherboard. I think I can exclude the RAM, I tried running with alternating two of the four RAM sticks, and it still hangs. I got a cheap Ryzen 3 CPU and now swapped the CPU, waiting to see what happens.
I need to prove this to my own satisfaction, but it has to be the crap Gigabyte motherboard. I should get a cheap Radeon 5450 card today, and after swapping that only the motherboard is left as the common denominator.
Their site shows Gigabyte pushing out BIOS updates every two months, on average. I just saw that they have a new one, since I loaded the latest one at the start of all this. Why? Why do you need BIOS updates six times a year? This is not a sign of a mature product.
How about <ctrl><alt><f2> or a <ctrl><alt><f3>
To get back out, do a <ctrl><alt><f1> or a <ctrl><alt><f2>
What shows on the monitor when it freezes?
No, everything's frozen, no response. It usually hangs overnight, with the monitor sleeping. No response from the keyboard, no pings from the network. The only thing to do is to hit reset and reboot.
Nothing gets logged in syslog. journalctl shows the normal noise, then dead silence.
HAve yo tried turning off all teh "sleep" and "suspend" features and set everything to always on.
Also, do you have a UPS power supply? Is your VCR blinking?
ToddAndMargo via users writes:
HAve yo tried turning off all teh "sleep" and "suspend" features and set everything to always on.
Also, do you have a UPS power supply? Is your VCR blinking?
Yes, the whole thing's on a UPS. I adjusted all the appropriate sleep and suspend settings. Its console is attended, and I ssh in. It does not go to sleep on its own.
This is a hardware fault, somewhere. I doubt that it's the video card, but I have a spare on the way. It's either a bad motherboard, a bad CPU, or an incompatibility between the motherboard and the CPU. I swapped the CPU, and I have nothing to do but fiddle my thumbs for the next two weeks.
On 5/17/22 05:41, Sam Varshavchik wrote:
I swapped the CPU, and I have nothing to do but fiddle my thumbs for the next two weeks.
My experience with such a test is that it will be the one time when it takes three times as long to reproduce and I jump to conclusions thinking I have found the problem and get my heart broken.
Be patient!
On Tue, 17 May 2022 07:46:09 -0400 Sam Varshavchik mrsam@courier-mta.com wrote:
No, everything's frozen, no response. It usually hangs overnight, with the monitor sleeping. No response from the keyboard, no pings from the network. The only thing to do is to hit reset and reboot.
Nothing gets logged in syslog. journalctl shows the normal noise, then dead silence.
Maybe switch to a console before leaving it for the night. And run something in the console to constantly monitor the system, like top, but more informative. Then when it freezes, the last things running and their issues are already on screen.
I vaguely recall having an issue with a screensaver having a memory leak, and when the system was down for longer periods of time, it would do a freeze like this. I don't recall the exact details. There was no OOM killer then, and I'm not sure the OOM killer would take care of that.
And I had something like this happen when a power supply was going flaky. The voltages had drifted out of spec as it decayed.
SSH into the system to see if it responds?
Tough problem to diagnose.
setup crashdumps. there is a wiki someplace for how to set them up and how to test that crashdumps are working right.
And leave the screen on the text console, you may get a kernel dump. If you get a kernel dump and/or kernel message on the screen odds are it is a software issue.
Usually on a fatal hw issue (PCIe error, machine check exceptions) the bios will reset the machine immediately and you won't get a crashdump/message on the screen but the machine will be found up but rebooted.
None of the MB vendors price support(they only provide warranty of the hw, and will usually immediately claim it is a software issue) into their products and will look for any excuse to say not their issue.
On Tue, May 17, 2022 at 8:03 AM stan via users < users@lists.fedoraproject.org> wrote:
On Tue, 17 May 2022 07:46:09 -0400 Sam Varshavchik mrsam@courier-mta.com wrote:
No, everything's frozen, no response. It usually hangs overnight, with the monitor sleeping. No response from the keyboard, no pings from the network. The only thing to do is to hit reset and reboot.
Nothing gets logged in syslog. journalctl shows the normal noise, then dead silence.
Maybe switch to a console before leaving it for the night. And run something in the console to constantly monitor the system, like top, but more informative. Then when it freezes, the last things running and their issues are already on screen.
I vaguely recall having an issue with a screensaver having a memory leak, and when the system was down for longer periods of time, it would do a freeze like this. I don't recall the exact details. There was no OOM killer then, and I'm not sure the OOM killer would take care of that.
And I had something like this happen when a power supply was going flaky. The voltages had drifted out of spec as it decayed.
SSH into the system to see if it responds?
Tough problem to diagnose. _______________________________________________ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam on the list, report it: https://pagure.io/fedora-infrastructure
On Tue, May 17, 2022 at 10:03 AM stan via users < users@lists.fedoraproject.org> wrote:
On Tue, 17 May 2022 07:46:09 -0400 Sam Varshavchik mrsam@courier-mta.com wrote:
And I had something like this happen when a power supply was going flaky. The voltages had drifted out of spec as it decayed.
I too have seen that.
SSH into the system to see if it responds?
If your network supports OvrC you can set it to notify you when a system goes offline.
Tough problem to diagnose.
Agree (but an interesting challenge)
You can try to determine if the problem occurs randomly at a low rate rate (likely hardware) or more deterministically after a long period of uptime by rebooting on a schedule. Random issues are more likely hardware (e.g., power supply) while deterministic is software (e.g., memory leak).
Getting crashdumps or console messages can help pinpoint the time. Some people have used a video camera set for time lapse to capture interesting details from the screen.
I've had systems that crashed when IT hit it with a periodic network scan. I used the time correlation with network logs on a second system to determine the cause (faulty implementation of SNMP). You need a system watching the network and a way to pinpoint the time of the failure.
On 5/17/22 06:02, stan via users wrote:
SSH into the system to see if it responds?
I can run Thunar and geany and a few other simple interface programs over teh Intrnet to customer's sites wi6th SSH and X11.
Here is a sample thunar run line:
thunar ssh://50.zzz.xxx.yyy:abcd/home/rrrr/Documents
OR just fire them off from the command line with with the -x switch:
ssh -l rrrr -t -X -p xxxx 192.168.255.102
You can probably get xsensors to fire off this way.
On 5/17/22 03:13, Sam Varshavchik wrote:
Ryzen 3950X CPU
I have found that AMD processors' only reason for existence is to keep Intel's prices down.
It is not that AMD can't make a good CPU, it is because they are expected to be cheaper and so everything else that goes with them is also expected to be cheaper. Makes for a cheaper, cheaper, cheaper house of cards after a bit that kinds, sorta works.
When I build system, I stick with Intel chipsets and CPUs.
I have had thermal problems with a particular Supermicro board, but I forget which one.
ToddAndMargo via users writes:
On 5/17/22 03:13, Sam Varshavchik wrote:
Ryzen 3950X CPU
I have found that AMD processors' only reason for existence is to keep Intel's prices down.
It is not that AMD can't make a good CPU, it is because they are expected to be cheaper and so everything else that goes with them is also expected to be cheaper. Makes for a cheaper, cheaper, cheaper house of cards after a bit that kinds, sorta works.
I have another 16 core/32 thread ripper, a slightly different model, 2950X. It's been rock solid. I'm not sure what motherboard it is, I had it built for me.
On 5/17/22 03:13, Sam Varshavchik wrote:
I keep the system doing nothing, and it freezes up randomly every 3-10 days.
On Windows machines, if fans are frozen and not cooling properly, you turn your back to them for a bit, then turn back, touch the keyboard, and they freeze up. Gives the user a complex thinking they did something to the computer.
AMD's notoriously run hot.
I adobe "xsensors" for checking heat on Linux machines.
Is your case closed when this happens? Open cases do not cool properly.
ToddAndMargo via users writes:
On 5/17/22 03:13, Sam Varshavchik wrote:
I keep the system doing nothing, and it freezes up randomly every 3-10 days.
On Windows machines, if fans are frozen and not cooling properly, you turn your back to them for a bit, then turn back, touch the keyboard, and they freeze up. Gives the user a complex thinking they did something to the computer.
AMD's notoriously run hot.
I adobe "xsensors" for checking heat on Linux machines.
Is your case closed when this happens? Open cases do not cool properly.
I have plenty of cooling on this thing, five 120mm fans. Two came with the case, it had space for four more, and I shoved three more in. I had to get 1:3 fan power multipliers, to get all the fans plugged in (and I checked the motherboard's manual to make sure it can supply sufficient amperage to all of the fans).
Plus the monster heat sink I strapped on the 16 core CPU, with its own fan.
On 5/17/22 04:57, Sam Varshavchik wrote:
Is your case closed when this happens? Open cases do not cool properly.
I have plenty of cooling on this thing, five 120mm fans. Two came with the case, it had space for four more, and I shoved three more in. I had to get 1:3 fan power multipliers, to get all the fans plugged in (and I checked the motherboard's manual to make sure it can supply sufficient amperage to all of the fans).
Plus the monster heat sink I strapped on the 16 core CPU, with its own fan.
Awesome. You can never have enough circulation, if you can stand the noise. Love PWM fans.
ToddAndMargo via users writes:
On 5/17/22 04:57, Sam Varshavchik wrote:
Plus the monster heat sink I strapped on the 16 core CPU
Not to ask too stupid a question, but did you remember the thermal grease? xsensors will tell you.
Not only did I remember it, but when ordering all the parts I ordered an extra tube of it, even though the heat sink had a layer of goo pre-applied on it. This came in handy when I had to remove it and reinstall it a few times. Despite the fact that I was putting on at least twice as much as the pre-applied layer, the extra tube has enough for at least 6-8 "here-goes- nothing"s.
This is going to come down between the mobo and the CPU. So, which one's more likely to be junk? My money's on the mobo. It's only a question of the precise nature of the problem. Either it's plain junk, or it has a difference of opinion with this particular 16-core Ryzen 3950X CPU model. Only time will tell. The replacement 4 core Ryzen 3 has been idling for two days now. I'll wait three weeks. It never lasted longer than three without locking up.
Even if it stays up I'll get another mobo and plug the 3950X on it, with the same hardware, and see what happens. I have a compatible MSI board somewhere around here. I actually tried it first but couldn't get it to work, and got this Gigabyte deal, but I'll give the MSI another go, or get a third one. I'm very happy with the 16-core Ryzen 2950X in my older gear, and I have more confidence in the CPU, especially seeing how unhelpful Gigabyte support turned out to be.
On 5/17/22 17:12, Sam Varshavchik wrote:
ToddAndMargo via users writes:
On 5/17/22 04:57, Sam Varshavchik wrote:
Plus the monster heat sink I strapped on the 16 core CPU
Not to ask too stupid a question, but did you remember the thermal grease? xsensors will tell you.
Not only did I remember it, but when ordering all the parts I ordered an extra tube of it, even though the heat sink had a layer of goo pre-applied on it. This came in handy when I had to remove it and reinstall it a few times. Despite the fact that I was putting on at least twice as much as the pre-applied layer, the extra tube has enough for at least 6-8 "here-goes-nothing"s.
This is going to come down between the mobo and the CPU. So, which one's more likely to be junk? My money's on the mobo. It's only a question of the precise nature of the problem. Either it's plain junk, or it has a difference of opinion with this particular 16-core Ryzen 3950X CPU model. Only time will tell. The replacement 4 core Ryzen 3 has been idling for two days now. I'll wait three weeks. It never lasted longer than three without locking up.
Even if it stays up I'll get another mobo and plug the 3950X on it, with the same hardware, and see what happens. I have a compatible MSI board somewhere around here. I actually tried it first but couldn't get it to work, and got this Gigabyte deal, but I'll give the MSI another go, or get a third one. I'm very happy with the 16-core Ryzen 2950X in my older gear, and I have more confidence in the CPU, especially seeing how unhelpful Gigabyte support turned out to be.
I am thinking motherboard too. CPU are very high quality and I have only seen one bad on in 28 years.
Gentlemen:
Its not just the heat sink, consider also the heat sink fan? is it big enough?
is it working properly, at full Speed? Also do you have case fans?
Again working properly?
Is the heat sink clogged with dirt and dust? Remember spring is here and were getting warmer days. This could explain your problem!
Remember a 16 core processor put out a lot of heat. _BTW on some of these_ _multi-core devices require liquid cooling._ Check your manufacturer's requirements.
Thomas Dineen
On 5/17/2022 5:12 PM, Sam Varshavchik wrote:
ToddAndMargo via users writes:
On 5/17/22 04:57, Sam Varshavchik wrote:
Plus the monster heat sink I strapped on the 16 core CPU
Not to ask too stupid a question, but did you remember the thermal grease? xsensors will tell you.
Not only did I remember it, but when ordering all the parts I ordered an extra tube of it, even though the heat sink had a layer of goo pre-applied on it. This came in handy when I had to remove it and reinstall it a few times. Despite the fact that I was putting on at least twice as much as the pre-applied layer, the extra tube has enough for at least 6-8 "here-goes-nothing"s.
This is going to come down between the mobo and the CPU. So, which one's more likely to be junk? My money's on the mobo. It's only a question of the precise nature of the problem. Either it's plain junk, or it has a difference of opinion with this particular 16-core Ryzen 3950X CPU model. Only time will tell. The replacement 4 core Ryzen 3 has been idling for two days now. I'll wait three weeks. It never lasted longer than three without locking up.
Even if it stays up I'll get another mobo and plug the 3950X on it, with the same hardware, and see what happens. I have a compatible MSI board somewhere around here. I actually tried it first but couldn't get it to work, and got this Gigabyte deal, but I'll give the MSI another go, or get a third one. I'm very happy with the 16-core Ryzen 2950X in my older gear, and I have more confidence in the CPU, especially seeing how unhelpful Gigabyte support turned out to be.
users mailing list --users@lists.fedoraproject.org To unsubscribe send an email tousers-leave@lists.fedoraproject.org Fedora Code of Conduct:https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines:https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives:https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam on the list, report it:https://pagure.io/fedora-infrastructure
Thomas Dineen writes:
« HTML content follows » Gentlemen:
Its not just the heat sink, consider also the heat sink fan? is it big enough?
That thing is a monster.
is it working properly, at full Speed? Also do you have case fans?
Again working properly?
Yes, and yes. And the lock-ups occur when the system is idle overnight and does absolutely nothing whatsoever.
Is the heat sink clogged with dirt and dust? Remember spring is here and were getting warmer days. This could explain your problem!
It's brand new and clean.
Remember a 16 core processor put out a lot of heat. BTW on some of these multi-core devices require liquid cooling. Check your manufacturer's requirements.
If it was crashing under load then this would be something to explore. But it's not.
On Tue, 17 May 2022 20:46:02 -0400 Sam Varshavchik wrote:
Yes, and yes. And the lock-ups occur when the system is idle overnight and does absolutely nothing whatsoever.
That's the sort of thing that would make me want to run "journalctl -l -f" on an xterm remotely connected from another system that stays up to see if the same stuff shows up before each crash.
Tom Horsley writes:
On Tue, 17 May 2022 20:46:02 -0400 Sam Varshavchik wrote:
Yes, and yes. And the lock-ups occur when the system is idle overnight and does absolutely nothing whatsoever.
That's the sort of thing that would make me want to run "journalctl -l -f" on an xterm remotely connected from another system that stays up to see if the same stuff shows up before each crash.
I'm going to try this, even though "-b -1 -r" showed nothing after the fact: normal noise, then dead silence. Perhaps the journal entries were not flushed to disk, but this might catch it.
On Mon, 2022-05-16 at 19:02 -0400, Sam Varshavchik wrote:
One of their motherboards keeps locking up on me. After some back and forth their last reply ended the discussion thread with a:
# Since we do not fully support and validate Linux, we cannot offer full # support on Linux. Please try a Windows base OS to check if you are still # experiencing the same instability issue.
I take issue with this on a customer service stand point.
As the customer, it is NOT YOUR JOB to diagnose faulty hardware. Probably most of their customers would be unable to do that. You should be able to return non-working equipment and THEY should assess it. You should also be able to return it to the vendor without being hassled about it, nor being told to go back to the manufacturer instead (in my country, that's actually illegal to be fobbed off that way).
Of course you mightn't trust them, and it's in your interests to do your own tests, first. But the obligation should be on THEM.
When I've dealt with good companies, I've asked them about non-working equipment, they've given me a few simple to do "have you tried this?" questions, and when I confirmed I'd done what I could, they simply told me to bring it back. If they'd asked me to jump through hoops I would have given them a serve over it, but they didn't. I delightedly name one, Jaycar, as a company that will try their level best to get out of warranties, and act illegally about it.
On Mon, May 16, 2022 at 8:02 PM Sam Varshavchik mrsam@courier-mta.com wrote:
It's been a long time since I've encountered this kind of obtuseness. This is not new, but I thought that this kind of blindness was in the rear-view mirror. I guess not.
One of their motherboards keeps locking up on me. After some back and forth their last reply ended the discussion thread with a:
# Since we do not fully support and validate Linux, we cannot offer full # support on Linux. Please try a Windows base OS to check if you are still # experiencing the same instability issue.
As suggested, you should be able to do this without purchasing a license, and it would give you a way to install firmware updates.
Just thought I'd mention it here, so that other DIYers might take this into consideration. Looking at their support page, it's listing new BIOS updates every month.
It would be worth searching for other reports of problems with the same motherboard regardless of OS. Also look for details of the BIOS updates. Modern hardware is pretty uniform, so the same problem often crops up on other instances. Non-reproducible problems tend to be with cables and connectors.
Try stripping down to the most basic configuration and swapping the remaining add-ons (GPU, mass storage, cables).
-
George N. White III -
Richard Ibbotson -
Roger Heflin -
Sam Varshavchik -
Samuel Sieb -
stan -
Thomas Dineen -
Tim -
ToddAndMargo -
Tom Horsley