Forum: Author Hangout

that's a pretty hard question. it seems to crash when the vm system is loaded, but that may only be because it is loaded most of the time. but as the timing of the crash is random that is not a valid measure. I'd have to leave it down for a month or two to exclude it from being a cause and that is not doable. I did manage to rule out chrome with a few crashes before it was loaded.

I have a reasonable suite of programs that run all or almost all the time. however I can't suffer the app downtime to rule each one out if it, as sometimes happens it takes 2 months for a crash to happen. I can't be without those apps for that period of time.

other apps include discord, steam, outlook, foobar2k, vlc and occasionally Libreoffice.

What OS (and what version there of) are you running.

On a MS Win system, the two most likely problems are a software compatibility issue and a problem with one or more applications using too much memory. Windows can crash if memory is at 100% utilization.

How often do you reboot the system? When you get the multiple failures in a day, does that happen when the system has been up for a particularly long time?

Another possibility is a hardware problem. I had a similar problem with seemingly random crashes of the video drivers. It could go weeks without an issue and then crash several times in one day.

It turned out the problem was that the fan on the co-processor cpu on the video card had failed. Heat buildup was causing random faults in the video card.

the 'host' is Win 11 - 22h2 and the 'guest' is a Ubuntu derivative.

system has 64GB ram, I'm yet to run out.

the system hard boots every crash. it's like hitting the reset button, normally what I do though is turn it off for a minute or so. this doesn't really seem to make any difference.

I've run memtest86+ for 72 hrs, but I'm looking for another tester to be more thorough and rule that out.

the system hard boots every crash. it's like hitting the reset button, normally what I do though is turn it off for a minute or so. this doesn't really seem to make any difference.

That wasn't the question. Not counting crashes, how often do you reboot the system?

Is it a desktop or a laptop?

it's a desktop. I only power cycle it for system updates and the crashes when they occur. in the past my uptime record, for a different (MS) system was 98 days and I only turned it off then because it needed to be moved. that was an XP system. I've run every version of windows from 3.1/NT3.51 except 8 and 8.1, though I have worked on them. that's a lot of system builds. I've run into a great many peculiar behaviours but this one seems to defy simple explanation.

Not having any trace aside from an ID 41 Kernel-Power is usually an Hardware failure, maybe the power supply.

There are some rare instance in which a BSOD fail to leave a trace. To rule that out, you should disable the automatic restart in case of system failure.

If it is indeed a traceless BSOD, the lack of trace lies probably with a third party driver. Try to stay away from non microsoft NVMe drivers, SSD toolbox or accelerator feature, get rid of the Intel Storage driver (unless it is used for raid).

Which hypervisor are you running?

Is your ubuntu VM sharing files with the Windows Install like it can be done with wsl2 or docker? In that case avoid sharing of cloud storage OneDrive/GoogleDrive/Dropbox/iCloud trough that feature.

I do get the odd blue screen, but they are less than 1% of the crashes, in fact had one today which was interesting because every reboot after it went straight into BIOS, until I hard powered it off for 5 minutes.

the only non-standard drivers (windows wise) are the Nvidia one for the GPU. everything else is straight from the motherboard manufacturer. there's no RAID or accelerators, I'm using 'storage spaces' instead. no sharing between any of the VMs and the host, they're all mostly isolated though I have linux services installed it doesn't point at anything.

The hypervisor is VMWare Workstation 16 and has been updated to two new dot versions.

I have suspected the PSU, maybe a bad cap or similar for a while but don't have a spare. I do intend to get a new one, likely upgrading from 850W to 1KW PCI 3.0, just no funds atm

straight from the motherboard manufacturer

Those I avoid like the plague. They are more often than not buggy as hell and giant security holes.
Motherboard Driver also replace the Microsoft drivers by intel/AMD one which are also crap at least for the storage/thuderbolt/MEI parts.

when I say straight from the manufacturer I am referring to the non-standard hardware that Windows 11 couldn't detect. I've learnt long ago to trust a 'windows' driver over OEM unless there is some sort of shenanigans with the OEM stuff because of some slightly different implementation of the product stack.

that said back in the 9x and early 'common core' days it was the other way round. M$ drivers were absolute shite then and the internet was relatively immature so multi-sourcing wasn't as easy as it is today. I do remember 'borrowing' drivers from other versions to get hardware on the local system to 'work'.

Vista was a particular minefield for this, at least the first 12 months, one of the reasons for its failure. in addition to unrealistic system requirements, it was very good if you could throw enough resources at it. but Christ did you have to go big or go home.

RAID drivers are a key example of this and one of the reasons I've moved on from RAID to WSS, that and destructive rebuilds with most consumer RAID sub-systems.

Enterprise gear is different but not in the budget for this box. there is a different 'project' that will likely get enterprise RAID 60

VMWare Workstation 16

That's an horrible idea:
* You are not running on a validated reference design.
* You are not running the latest version 17.x
* You have to disable all Hyper-V features on Windows 11
That include, sandbox, Host Guardian, wsl2, hyper-v, Defender memory integrity, and the list goes on...

Also, considering that Broadcom (future owner of vmware) are there to milk the cow and announced plan (falsely rescinded?) to lay off developer and drop low end product like Workstation, I would plan to get out of that ecosystem asap.

I'll look into ESXi to put on my dual Xeon rig, main concern would be remoting into it, I've had troubles doing that to a *nix machine here and there. This particular flavour doesn't seem to like remote access

I've used Hyper-V in a few forms, bare-metal and as a service type 2. Virtual Box and VMWare.

I really didn't like Virtual Box.

Hyper-V seemed to have issues when the network changed.

The VMWare 'machine' is a carry over from a different host. I've not found an 'affordable' way to convert VMWare to VHDx. Iwas planning to migrate to 17 but the last minor update to 16.x caused major issues, so I've not gone to 17 yet.

Have you considered ditching M$ entirely? For home use, I started with win3.0 (unix variants at work). With the disaster of win8 I gave up. Now all my client computers are linux (mostly linux mint and some ubuntu). I've been quite happy with linux mint for the last 10+ years; for whatever help that gives you.

Not gone full *nix for reasons of compatibility, mainly for steam library use. I do have a few other M$ apps that there is no *nix alternative, but mostly as a semi-serious gamer (steam library valued at over $25K) and that doesn't include my other libraries.

*nix wise I've used, bare metal: System V, SUSE, Redhat/Centos, Ubuntu/Mint/other binary compatibles, Puppy and a couple of others that I can't seem to recall. Guest wise the list is similar mostly concentrated around Ubuntu binaries.

I've never had a personal 8/8.1 system but I have had nearly every other system including NT variants prior to XP. also run DR-DOS, OS/2 Warp and BeOS in the 'non-standard' list.

but the compatibility scope of windows holds me to it fairly hard.

Yes, I empathize. I managed to find, and then learn, adequate replacements for everything but Sony Vegas (non-linear video editor). There are probably 10-20 Linux video editors; but I got so proficient with Vegas, and now edit so rarely, that I haven't had the desire to learn a new package. It's on my (someday) To Do List.

only one piece of hardware has changed since initial build, changing from a 10 series to a 30 series card. drivers have changed every time an update has dropped. the crashes have been a consistent issue from day one.

This has been my most problematic PC ever, since '90/91 when I changed over from my trusty ol' Apple IIs, well one legit and one clone which I used for about 10 years without any issues ever.

yer, Mac OSX is based on NExT which is based on Rapsody which is based on *nix. my main problems with the Apple ecosystem are lack of product range (walled garden) and limited combustibility.

that said I do have a 24" iMac sitting to the side. all it needs is a new HDD, but from memory won't run anything newer than 10.10 iirc.

the CPU is a Core 2 model which makes it pretty dated. I'd be interested in putting *nix on that, maybe. it's waiting on some desk space and that HDD as mentioned.

only one piece of hardware has changed since initial build, changing from a 10 series to a 30 series card. drivers have changed every time an update has dropped. the crashes have been a consistent issue from day one.

So the crashes started before that change? No use in trying to revert to the 10 series then.
Since it's from day one it's most likely a hardware issue. If you have enough memory try removing the memory chips one by one and run a while with a little less memory.
I have a new machine that's now for the second time back to the company that custom assembled it (warranty) and the first time a bad memory chip caused problems that sound a little like you are experiencing. If none of the memory chips causes a problem try other hardware that you have laying around and can be a temporary replacement.

I've thought about memory and ran a 72hr looped test with Memtest86+. I want something more thorough but not sure what to use. A memory test is, along with a CPU 'burn-in' something I do every build right off the bat. system passed all of those fine. when I originally built it up, I did heavily modify the Windows install with a utility I got off github, but looking back at it there were no similar complaints. since then I've done a repair install, keeping apps and docs. did that for a different problem where I couldn't get on to the desktop. that was fixed but the random reboots continue. So I'm thinking either mobo or psu as the most likely culprits. after 24h1 drops I'm definitely doing a bare metal reinstall, by then I'll also likely have some testing h/ware a mate is donating to me. not sure what's in the bundle but I'm eager to do some swapouts then.

Facit, with MS Windows you can't know what really caused the problems!

Every day I'm glad I ditched the MS crap a long, long time ago. XP was the last I used while transitioning to Linux.

Any time I click on Edge's icon it tries to start, shows shortly a blank window and dies silently

Just kill ALL Edge process then restart it.
If needed, you can repair the install in the parameters Apps.

yer, I looked at Proxmox as a way to virtualise a set of services on a single physical, mainly file server and a collection of 'other' services such as BDC, IIS and Mail were amongst the list when I looked. Pass through the controller card and other hardware, NICs perhaps I'm not sure about virtualising a gaming rig with a passthrough. How well do those sorts of things cope with triple or more monitor setups?

I guess I'd just have to get creative with drive arrangements etc

NICs perhaps I'm not sure about virtualising a gaming rig with a passthrough. How well do those sorts of things cope with triple or more monitor setups?

Shouldn't be a problem. You can add/change/remove/configure all hardware in the KVM configuration. If you create the VM with all three monitors present they should all three appear in the VM. Windows should simply recognize all monitors that are added to the hardware configuration as if they were 'real' monitors. I've done it with two monitors although I only use Linux systems.