Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 22 Mar 2005 10:53:07 -0500
From:      Bart Silverstrim <bsilver@chrononomicon.com>
To:        freebsd-questions@freebsd.org
Subject:   Re: Anthony's drive issues.Re: ssh password delay
Message-ID:  <1bcc63264925d9b965671502cd627f45@chrononomicon.com>
In-Reply-To: <1574051488.20050322161955@wanadoo.fr>
References:  <20050321095647.R83831@makeworld.com> <LOBBIFDAGNMAMLGJJCKNIEOAFAAA.tedm@toybox.placo.com> <1907678552.20050322101315@wanadoo.fr> <9c07bb562656cd8cce0cd70dff184b93@chrononomicon.com> <1574051488.20050322161955@wanadoo.fr>

next in thread | previous in thread | raw e-mail | index | archive | help

On Mar 22, 2005, at 10:19 AM, Anthony Atkielski wrote:

> Bart Silverstrim writes:
>
>> Or it gave warnings that NT didn't.  Or it showed problems that NT
>> didn't.
>
> Unless someone can tell me what these messages mean, they are useless 
> to
> me, warnings or not.
>
>> If it worked so well, why not put NT back on the machine and try
>> running a battery of tests and diagnostics on the machine with NT to
>> see if it was masking a problem or not?
>
> I don't have any T&D software for NT.

Then you could purchase some or look to see if there's shareware to run 
the disks through their paces.

Or you can look for a boot disk from the drive or controller 
manufacturer.  They often have test utilities available for download.

Point is you're having trouble with FBSD and kept pining about NT 
working better, so the best solution (especially after the number of 
people you've insulted here) would be to go back to that software and 
see if everything is indeed copasetic under that OS with diagnostic 
software.  You've insulted your way into the killfiles of several 
people here on the list so they are ignoring you...seems like Ted is 
the only one trying other suggestions for you at this point.

> The smartctl utility looks very interesting, but I don't know to what
> extent I can trust it.

Um...little secret...you can't trust software 100% to diagnose 
problems.  Just like memory testers.  Just because software memory 
tester says it's okay, doesn't mean it's okay.  If it finds a problem, 
you probably have a problem.  If it says it's fine, it means you may 
have a problem.

It's just one tool.

If you want something you can trust, it costs several thousand dollars 
for specialized diag equipment.  Be my guest :-)  Usually for most 
people it's cheaper to just narrow down the part having problems and 
replace it.

>> I don't remember ever losing data while using DOS...
>
> I don't recall any specific instances of that, either.

Well, with that track record, who needs FreeBSD OR Windows?? :-)

>> Really?  Windows' problems are fixed by the benevolent leader Allchin
>> in MS?
>
> No, but the support organizations of major vendors don't behave like
> ill-mannered schoolboys when they are asked to show some responsibility
> for their software.

I think you could also look back at some of your own posts and see that 
you're not entirely innocent in the matter.  Several of your posts have 
been inflammatory.  None of the people on this list are obligated to 
help anyone.  In general they ask for some gratitude and respect and in 
return give advice and pointers.  On top of that, none of them forced 
you to use FreeBSD.  If it doesn't work, try a different version or a 
different OS.  They have no "responsibility" to their software on your 
hardware when you weren't forced to use it in the first place.

>> Or they could cover the other bases...that perhaps BSD is saying
>> something that NT didn't report.
>
> Rest assured, they will not do that.  They will cancel the FreeBSD
> rollout and return to NT.  And I couldn't blame them.

I don't think that was the context I was stating that in.  I was giving 
the possibility that NT was hiding the error while FreeBSD was 
informing you of a potential problem.  If people testing for a rollout 
want to roll back or go to another platform, that's their 
business...like I've said before the FreeBSD team isn't out to rule the 
world, they're just working on an OS.

>> Yes, obviously FreeBSD is worthless.
>
> Not worthless ... defective.  On this particular machine, the SCSI
> problems are too serious to allow it to be used for any kind of
> production.  I suppose a company could just swap out hardware over and
> over until it got lucky.  Or it could run a different OS.

Well, if it's similar to the NT idea I proposed, you could comment out 
the portion of the driver that reports the error, and voila'!  Fixed.

> I've been a bit luckier on my production machine, which has brand-new
> hardware.  Still, I'm getting mysterious SATA errors from time to time,
> one of which crashed the system.

Sata is *relatively* speaking, new.  And I've seen other posts about 
questions with SATA controllers and chipsets and support.  It's being 
fleshed out so I'm not surprised there's problems with some.  Is the 
SATA controller on the compat list for 5.3?

> As usual, nobody has a clue as to
> what's causing them, and once again, the Great Satan is the hardware.

If it's new or non-tested, I'm not surprised.

> Even though the only common point between the two machines is that they
> are running FreeBSD, somehow four separate disk drives and two
> controllers must be magically failing simultaneously.

Only thing in common?  Despite one being "near new" and the other being 
"eight years old"?

One cutting edge, one legacy, if only there were a middle ground where 
FreeBSD seemed to be comfortable running without errors...hmmm.....

>> I know that I routinely scour the OS source code for problems when one
>> of our cobbled workstations acts up under Windows.  Oh, wait, I was
>> just reminded that usually it's a network card or video card that goes
>> bad in old systems that prevents them from working.  Or a bad memory
>> stick, or a processor that overheated.  But maybe those errors were
>> caused by a bad Windows driver?  I don't know.  Just know that when I
>> replace the part, the OS must suddenly have the bugs fixed in the
>> source code too.
>
> Tell me the exact part that is failing when I get these SCSI errors.

Check the restarter coil.  It's always troublesome in my Apple systems.

>> Really?  Here I thought that was called "troubleshooting", especially
>> if the hardware in question is on the support hardware list and people
>> on lists with the same hardware aren't having that problem.
>
> People with the same hardware ARE having that problem.  But nobody ever
> resolved it for them, either.  It was the usual story about how it must
> be a hardware problem.

So you've narrowed it down to more likely being the software.

Is it possible that there was another reason they didn't pursue fixing 
that particular make and model of controller's driver?

>> Silently resetting without telling you?
>
> I'm talking about the quirks that the drivers take into account for
> different types of hardware.  FreeBSD, like many other operating
> systems, _does_ contain special code to workaround hardware
> idiosyncrasies, contrary to the implication that only Windows does this
> and that it is somehow a Bad Thing.

Only time I remember saying that to anyone was with working about 
*software* idiosyncrasies.  How often do FBSD/Linux bend over backwards 
because someone's version of Open Office stopped running after an 
update?

I don't actually remember anyone saying workarounds for hardware bugs 
being A Bad Thing.  But it's probably happened.

>> Or you find someone with the same hardware setup and install the OS to
>> see if it's reproducible.
>
> Others have already encountered this error, on the same hardware, as I
> recall from my newsgroup and Web searches on the subject.
>
> In any case, finding an identical machine today would be problematic.

Which would also strengthen the argument to let it go if it were very 
old hardware and developers can't even *get* the hardware to test with. 
  If you feel that strongly, contact the devel team and ask *nicely* for 
someone to try looking at it and send them your computer...

For most people, the effort isn't worth it; they'd get a newer server 
with hardware specifically on the compat list.  But to each their own.  
Can you try the liveboot CD idea with Linux?  Maybe you'd be happier 
with that.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?1bcc63264925d9b965671502cd627f45>