Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 22 Mar 2005 11:33:18 -0500
From:      Bart Silverstrim <bsilver@chrononomicon.com>
To:        freebsd-questions@freebsd.org
Subject:   Re: Anthony's drive issues.Re: ssh password delay
Message-ID:  <a44e8b1d23bc412823c0f654ac384afa@chrononomicon.com>
In-Reply-To: <802735952.20050322164924@wanadoo.fr>
References:  <423E116D.50805@usmstudent.com> <423EEE60.2050205@dial.pipex.com> <18510151385.20050321193911@wanadoo.fr> <eeef1a4c0503211224572d64e4@mail.gmail.com> <1975192207.20050322041925@wanadoo.fr> <eeef1a4c050322010021fd8eb4@mail.gmail.com> <1688160068.20050322102514@wanadoo.fr> <eeef1a4c050322014420d89861@mail.gmail.com> <1404322406.20050322112613@wanadoo.fr> <c112a9a423c9f4a9702d0e1f959e7b59@chrononomicon.com> <802735952.20050322164924@wanadoo.fr>

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

On Mar 22, 2005, at 10:49 AM, Anthony Atkielski wrote:

> Bart Silverstrim writes:
>
>> And you ran into a snag that you can't work through.
>
> Yes, at least not with the time I have available.

Classic tune.  I play it a lot too.

>> Most people if they were doing this on a lark would either replace the
>> hardware or try a different distro.
>
> There's only one "distro" of FreeBSD.

Yup, that's why I was referring to Linux with the 'distro thing.  Or 
you can try NetBSD.  It came on a CD with this month's Linux Format 
magazine.  Supposed to be compatible with everything short of a 
toaster.

> Replacing the hardware defeats
> the purpose of making good use of existing hardware that still runs
> perfectly.

Well, if it's spewing errors, are you sure that it's running perfectly? 
:-)  Doesn't sound it!

I'd still suspect that there's a problem with the hardware at some 
level and NT just didn't report it to you.  But that's just a WAG.  
That's also why I suggested Linux.  See if a version of Linux reports 
errors as well.  Two to one saying something is wrong with the 
hardware, you might want to look at the hardware.  If it's just 
FreeBSD, then avoid FreeBSD on that hardware.

>> When I ran into that error before, I remember seeing another person
>> post to a list saying that NT doesn't report the reset error.  Linux
>> did.  Maybe this is what you're running into?
>
> Well, if NT didn't report the error, and I never lost the data, how do 
> I
> get FreeBSD to stop reporting the error?

Obvious answer is grep and commenting the fields out of the source code 
:-)

Obvious concern is that it is a warning that something's wrong (or not 
set up correctly) and it may fail or cause problems down the road.

>> And the design of the OS may be 20 years old, but the OS most people
>> are using isn't.
>
> So?

So, you're saying it's 20 years old.  I said it's based on a 20 year 
old *design* that still works very well, but if the OS were 20 years 
old, it wouldn't run on anything being released today hardware-wise.  
Thus the OS isn't 20 years old.  Hell, cars are *basically* the same as 
they first were.  But I don't think you can swap parts from a model T 
into an 04' vette and expect it to keep running properly.  Unless maybe 
the seat can be retrofitted.

>> Is the hardware on the compatibility list?  What is it?
>
> An AIC7880 SCSI controller.

Well, is there a way to dump the code from that controller and compare 
it to another one that is known not to be tampered with by the OEM?  
Anyone out there on the list have one they can do this to?

> Anything can happen with bad hardware.  That doesn't mean that every
> error you see is bad hardware.

No, but you eliminate the obvious and the easiest-to-fix parts first.

Digging through source code and going through debug cycles for legacy 
hardware isn't exactly time well spent by most sane opinions, when the 
fix could be a ten minute swap of a drive or something like that.

> When the processor failed on my old server, I got segment violations in
> all sorts of programs.  Does this mean that if I get a segment 
> violation
> on the new server, I should replace all the hardware--since I've seen
> segment violations before when a a processor fails?

Actually, I think most troubleshooters would get another processor 
first (or check that cooling was adequate to the CPU and memory) if 
that was what fixed it last time.  Next would be memory.  So right off 
the bat you have fans to check, processor, and memory.  None of which 
are the fault of the OS, all very likely culprits.

>> ... you simply refuse to believe the possibility that something could
>> have been wrong but NT didn't *TELL* you about it, and then want to
>> launch into an attack on the OS in a list where people are running
>> FreeBSD quite happily on a wide range of hardware.
>
> If no data corruption occurred, there was nothing wrong.  If NT doesn't
> have to tell me about it, neither does FreeBSD.

That's an interesting philosophy.  I already posted about Linux 
messaging that the controller was in need of constant resets.  The 
drive was *FAILING*.  NT didn't say anything.  If you like sticking 
your head in a hole then go ahead and comment out the code that give 
the error and all will be well.  Most UNIX people like having 
diagnostic errors in the logs to troubleshoot things.

I didn't lose data when that had happened to me.  But the failure was 
*coming*.  I had oddball pauses in the OS, I had clicking noises 
occasionally happening as the drive reset...but didn't lose data.  So 
go ahead and press your luck.  Your hardware, your choice.  Maybe it's 
a nothing error, like the ones I see when a bridged set of network 
cards on the network are making the ARP daemon go nuts on one of the 
servers with <ip address> switched from <MAC1> to <MAC2> error 
messages.

> But nobody actually knows what is wrong.  Nobody has any clue.  That's
> the problem.

Can you contact whoever the developer is that's in charge of SCSI work? 
  Ask the right person and I bet they could help.  That's about as 
obvious as "I found it in the last place I looked!" :-)

> Lots of people are willing to wildly speculate on this or
> that hardware problem in order to create the impression that they know
> what they are talking about, but careful scrutiny of what they say
> reveals that they are totally ignorant of the real problem.

*sigh*  they were offering assistance in the form of "when I had a 
similar problem, I tried X, and it worked...try X."  You tell them 
they're idiots.  Now, I can't imagine WHY people would get a bad 
impression of you.  This is a free list for trying to help people.  To 
"try to create the impression they know what they're talking about", 
they won't achieve much on a mailing list for other geeks and admins.  
That's also why most of them have no trouble telling you that you've 
treated people like crap and can shove that server where the sun 
doesn't shine if you have so little respect for their suggestions, 
since they aren't paid to get insulted or thanked on this list.  
Whether you get your stuff to work isn't their problem.  They were just 
offering free advice.  Want paid support?  Doesn't Walnut Creek still 
do that?  There's other consulting firms too who I'm sure would work 
with you on it once the contract is inked.  Then you have every right 
to raise a stink.

> They have
> no idea but cannot bring themselves to say so.  And their idea of
> troubleshooting is to replace hardware forever because they don't know
> what the software is doing.

Maybe it's a legacy from supporting Windows, or it's the fact that 
sometimes it's just *faster* to replace an ill-supported video card 
with one that works rather than work on a new driver.  I mean, *duh*.

>> The stories just don't jive.  You're implying it doesn't work and is
>> crap because all your hardware isn't working with it.  This list is
>> populated by people that have worked out problems and configuration
>> errors and are running it without trouble.
>
> On the same hardware?

Evidently not.  It's legacy hardware you're using! :-)

>> Something just intuitively tells me that at this point you're more
>> concerned with pissing people off to make a point.
>
> No, although I do get tired of talking to people who sound like they're
> still in grade school.  I'm accustomed to working with professionals,
> and professionals address the problem, they don't shoot the messenger.

Really?  I'd didn't get that impression after a few ad hominem attacks 
on people in the list.  Although I'm having fun with the occasional 
barb being tossed just because it appears that you have insulted so 
many on this list that you're rapidly losing hope of getting anyone to 
help you if for no other reason than you seem unwilling to work with 
their suggestions in the first place and attack people for being idiots 
when you aren't getting them to dive right in with the debugger for an 
eight-year-old HP system.

The grade school approach would have been to tell you to just go away 
and take your server with you right off the bad.  To the contrary, Ted 
has continued to offer suggestions despite your attacks, and other 
people have just started ignoring you, while a couple are sitting back 
and enjoying whatever fallout comes from your quandary since they have 
better things to devote time to than getting insulted.

>> Probably the best thing for you to try is Linux Knoppix on your 
>> server,
>> see if it boots and sees errors.
>
> I suppose I may be forced to try something else; it doesn't look like
> I'm ever going to get any help with FreeBSD.

Haven't you ever used Knoppix?  It's liveboot.  it *doesn't install* to 
your system.  Just runs straight from CD.  That's the point.  Easy way 
to try things, try to recover from problems, test for problems, trial 
Linux, etc...?  You're telling me you've never heard of it before?

> The problem is, most open-source communities are populated by the same
> sort of adolescent mentalities; all they want to do is defend their
> beloved favorite hardware and software and attack anyone who disagrees
> with them.

Quite the leap.  You're insinuating everyone is a doodyhead on the list 
while at the same time saying that they just want to defend their OS 
because you can't convince the list users to dive in with debugger and 
screwdriver in hand to fix the problem you're having specific to your 
machine, a machine that you said yourself would be difficult to get 
another of the same configuration for testing.   If it's difficult to 
get it for testing, how in hell do you expect the developers to 
reproduce the problem?  Ship them your server?  Your current approach 
seems to be tell them their nitwits then ask for help...how ridiculous 
is that approach?

The only "attacks" I saw on you were retaliation for your own 
inflammatory comments.  I don't recall EVER seeing someone on this list 
start flaming on someone without that someone baiting others into it.  
I've seen rude, I've seen curt, but not flames.  That tells me that 
when flames are billowing the other person brought their own gasoline 
to the party.

As far as the "open-source communities" are concerned, they're not 
doing it for a specific person.  Different people have their own 
agendas, but make no mistake...they do this because they want to work 
with the software.  They're not out to make a buck, they're not out to 
please the desktop users who insist on clicking on porn popups and 
viruses in email.  They really don't care if you want to use it or not. 
  They didn't force you to use it!  You came to the OS yourself.  If you 
don't like it, switch to something else...LIS, try the Knoppix CD to 
see if the error is there too!

> It's all one big weekend hobby to them, or one sacred
> religion, or both--but it's certainly not professional IT.

They're not driven by profit.  Deal with it!  And if you want to 
maintain that proprietary is the way to go because they aren't leaping 
in to fix a bug on an 8 year old computer you can't find a duplicate of 
for testing *yourself*, then that's fine with them.  Maybe you should 
tell Google or Yahoo that their OS is unsuitable for professional 
production environments.  Meanwhile I came back monday to a locked up 
Win2k laptop. Hmm...

Did you even look at the link/google searches I sent previously for 
people to ponder over with some critical thinking, the Shatter attack 
and Why I Hate MS pages?

> That's why
> open-source is never likely to replace proprietary software ... nobody
> in the open-source community is willing to grow up.

You're entitled to your opinion.  If it's worth getting this worked up 
over, try emailing the dev team and ask them nicely to look at it with 
you.  Stomping your feet and sulking while screaming about everyone 
ELSE being childish isn't going to get you very far.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?a44e8b1d23bc412823c0f654ac384afa>