:: Re: [DNG] Uptime issue
Top Page
Delete this message
Reply to this message
Author: o1bigtenor
Date:  
To: dng
Subject: Re: [DNG] Uptime issue
On Thu, Oct 17, 2024 at 5:21 PM onefang <onefang_devuan@???> wrote:
>
> On 2024-10-17 06:44:13, o1bigtenor via Dng wrote:
> > On Thu, Oct 17, 2024 at 3:04 AM onefang <onefang_devuan@???> wrote:
> > >
> > > On 2024-09-28 21:20:50, o1bigtenor via Dng wrote:
> > > >    On Sat, Sep 28, 2024 at 8:15 PM nick <[1]nick@???> wrote:

> > > >
> > > >      It's definitely a dream system. I would still suspect it though. My
> > > >      reasoning would be somewhat like this:
> > > >      1. Random lockups are not normal and shouldn't be happening.
> > > >      2. The cause has gotta be either hardware or software.
> > > >      3. If it's hardware it's gotta be one (or more) specific component that
> > > >      is failing, defined by if I replaced that component with an identical
> > > >      unit (of the same manufacturer and model) the problem would go away.
> > > >      4. If it's software it's gotta be a subtle bug or driver
> > > >      incompatibility, sometimes latent bugs can be triggered by unusual
> > > >      combinations eg. Let us say the driver for your AMD graphics card fails
> > > >      when there is 64 GB or more of RAM, just for the sake of example.
> > > >      5. It could also be a matter of settings or configuration eg if your
> > > >      BIOS has configured the RAM for a higher clock than it is specced for,
> > > >      although in this era of autoconfiguration this would probably count as a
> > > >      driver bug.
> > > >      What I would do as a starting point would be to pull out the GPU and
> > > >      half the RAM and use it for a few weeks to see if problem goes away.
> > > >      Does it have internal graphics or do you have an older GPU to use
> > > >      temporarily? If problem recurs swap the RAM for the other half and
> > > >      re-test. You can also try the GPU or RAM in another system to see if
> > > >      problem moves with it. If it turns out to be the GPU then it could be
> > > >      driver issue as drivers are very complex these days. You could try
> > > >      earlier driver or earlier kernel (as you are already doing) but such
> > > >      approach is fraught. Once you narrow down the issue to a specific part
> > > >      or driver its better to take it out of service until a new part or fix
> > > >      is available.
> > > >      In principle you can use the same approach to diagnose CPU or mobo
> > > >      issue, but you would need identical spares which could get costly. If
> > > >      buying spares for testing I would highly recommend to get a PSU first. I
> > > >      haven't been into system building for many years but I have heard that
> > > >      PSU is responsible for a large proportion of faults with modern rigs
> > > >      given how demanding they are on the PSU.
> > > >      I am sure you can solve this. The nightmare is when it happens on a
> > > >      laptop where you really have no option but to try earlier kernels or
> > > >      removing drivers or take the laptop out of service (has happened to me).
> > > >      On a PC it is much easier. Oh yeah another thought: you might try
> > > >      running the dreaded Windows on it for a while. If it still locks up you
> > > >      have eliminated software except possibly for common code in AMD display
> > > >      drivers.

> > > >
> > > >    Given that when I had all four browsers loaded the uptime shrank horribly
> > > >    and my peripheral knowledge of the
> > > >    desire of far too many companies to use javascript to do things and report
> > > >    back to them what they want to know
> > > >    I am far more likely to suspect software than hardware.
> > > >    Just like right now - - I am still stuck at the dredded mz googly's email
> > > >    system and right now ublock origin is
> > > >    telling me that the are some 115 domains linked in. (Now up to 120
> > > >    domains. Now up to 137!!!!)
> > > >    This kind of garbage is likely at the bottom of my issues - - - not that
> > > >    hardware can't be an issue but
> > > >    when hardware is manipulated by unscrupulous companies - - - well the
> > > >    results are issues for users.
> > > >    Sorry - - - I have no spare GPU and I don't know another currently
> > > >    available one that has 5 outputs
> > > >    so that's a doa issue already.
> > > >    I guess what I'm looking for - - - well maybe I need to setup that shared
> > > >    hosting setup that I have been thinking
> > > >    about and get mz googly off my neck. Then if I can find a way to jail any
> > > >    and all browsers then I think my
> > > >    hardware issues would like shrink mightily.
> > > >    Any suggestions on how I can jail any browser?

> > >
> > > Hardcore way - run them in a VM.
> > >
> > I tried lxd some time ago and as a result am quite gun shy to even the idea of
> > using VMs.
> >
> > What particular form of VM are you espousing?
>
> While there are less hardcore ways, I have spent a lot of time with qemu
> VMs for various purposes. So that's my go to for such things.


So what happens if a VM is discombobulated?

>
> On the other hand, I'm considering eventually creating something along
> the lines of Qubes, but Devuan based. A less hardcare solution, but
> that's coz all the hard work is done already. That'll have to wait
> though, got other things to get done before I can start on it.


Sounds interesting - - - do let the rest of us know how things go when you're
working on that.
>
> Search for apt packages with jail in the name, there's a few I think. Or
> containers.
>

Will do!

Thanks a muchly for your input!!