> 
> Hi!
> 
> Trying to kill the keyboard, [EMAIL PROTECTED] produced:
> > > Trying to kill the keyboard, [EMAIL PROTECTED] produced:
> > > > Yea, I'm a nut with IDE, 10 drives on this box, makes for good irc pissing
> 
> > > Well, talk about IRQ latency, especially if you serve a couple
> 
> > Not really... the disks are way faster than needed.
> 
> The disks, yes, but the other stuff might have troubles.  But you
> said you used irqtune

Nope, everything is up to par... I did try not using irqtune... same
results, so it's not an irq latency problem.

> > > I am just thinking at times. :-) Can't stop it then.  Which shrink
> > > do you recommend?  :-)
> 
> > Dr. Ruth? :-)
> 
> Naah.  Sometimes a cigar ist just that, a cigar.  Just ask Freud.

Uh... ok, enuff of this :-) I'll keep it clean *snort*

> > > M/K/G is a bit more human readable, though, so just make it
> > > an option.
> 
> > I was going to do that :-) for all the silly machines.
> 
> So how comes your machines can't multiply by 1024 ... or shifting
> bits left and right.  Mine do that stuff all the time :-) Or do
> you have a machine based on the number 10?  (Does that mean -0 !=
> 0 on it?)

Because it ends up not being the correct # of blocks :-) try it :-)
Mebby I'll make an option to spill how many blocks are available... prolly
make more sense than byttes, and be more useful, and... human readable :-)

> > > I know there's QMAGIC and ZMAGIC but it was new to me the magic
> > > changed the way ext2 works ... :->
> 
> > It can effect how it operates actually... look into /lib again :-)
> > Yeah, i know it's mostly for the tools, however... the kernel can be built
> > a.out, etc...
> 
> The Kernel is not swappable, so what?  It does not need to load
> libraries from HD, it's practically compiled --static (apart
> from modules).  Or am I wrong (again)?

Yup, it's static... but different libs/compilers produce different code an
different optimizations for different cpu's...

> > > Nah, you can break lilo a bit too easy for my taste, e.g. by
> > > defragmenting the disk or moving the kernel.  Loadlin just
> 
> > You have the source... :-)
> > Stop bitching, and write it :-)
> > MAke lilo store it'sself in the entire CYL0 HEAD0 area and ext2fs smart...
> 
> Well, I do have the source but the design options in LILO
> just were that you have locked positions of the kernel & co.
> Changing that would mean writing a new program or hacking ext2
> (and other FS) support into LILO, which is kinda hard in <512
> bytes.  Reserving more for the 1st stage loader means it's harder
> to retrofit.  /boot is a better idea here.  And I lack the time
> to get the experience needed, I'm afraid.

*nod* I can understand that...

> And kexec ... has that ever been implemented anywhere?

Donno...

> > Lots of the "EZ-DRIVE" type programs to this...
> 
> And you get a boot sector virus and can but hope you have
> a backup.  And it's sooo easy to get one.

No dos, no viruses :-) besides, we have tape backups... (ahem)

> > Better yet, mount /boot in it's own partition, with a kernel.
> > I have done this on some of my customer's boxes :-)
> > you could also tell ext2defrag to not move some important files...
> 
> So I need to know all important files by name.  Gaah.  If I
> wanted that kind of manual control as the *only* possibility,
> I should also defrag by hand, using a disk editor. :_)

Not a hard task, actually :-)

> > > See, *you* know what e2defrag can do to your HD if you strike
> > > bad luck, yet you offer it as a good and *safe* solution.
> 
> > I did? I think I offered it more as a last-ditch effort... atleast, that was
> > my intent :-)
> 
> No, it arrived as a 'this is the way to save your soul.  DO that
> if you have problems ... and if you don't have problems, do
> it anyway!'

It could be...

[SNIP]
> > *ROFL* I won't even comment how old I am or how long I've used linux.
> 
> Let me guess, you are 18 and use Linux since ... oh, 1985.  Aeh,
> sorry, wrong place, we weren't discussing the jobs managers want
> to offer or others want to fill.  (8 years java experience ...)

nope... infact I'm a divorced father. :-) surprize...

> > Yes, they are stored off site, and they use the tr-1 drive. Not only is it
> > redundant storage, but the redundant *COPY* can be read by both drives.
> > So, yes, I also have redundant drives :-)
> 
> Ah.  Now guess how many percent of 1 percent of the readers here
> happen to do that.  :-)

Zero, then they bitch when it's thier fault... I'm experienced enough to
know better...

[SNIP]
> > I can see a benifit to that... I think... course, I don't rm much... I just
> > keep adding disks :-) I ordered 3 more 10G tonite :-)
> 
> You are sure you don't want SCSI?  Aren't you running out of
> IRQs and slots now?  Or don't you want to buy a black goat?

1: Yes, I'm sure I do NOT want SCSI
2: IRQ is a non issue for me (relates to top secret stuff)
3: I have a 10 slot ISA backplane
4: I don't like SCSI and I don't want to start a flame war. it's a matter
   of personal preference.

> > > /var always changes, and can overflow.  In your case, it would
> > > overflow /tmp as well.  Ever ran a system with *no* space in /tmp?
> 
> > Nope... I make sure it has enuff elbow room.
> 
> Sure :-) How long are you away from the machine?  And if you
> use find, better hope you have no evil users. :-)

I'm litterally glued to it except on weekends, when I tend to my daughter.

[SNIP]
> > > > Nope... I have many large multimegabyte databases here.
> 
> > Actually, most of it is sequentially read. No, I'm not explaining why or
> > what it is :-)
> 
> Ok, just one question:  How much faster is it read by dd?  

Never timed it. I'm sure it's slower with dd...

> > future... yes, I have to be vague... top secret stuff, sorry :-)
> 
> Oh.  Do I have to burn it before reading?  Or do you have to
> kill me before telling me?

Nope, I just won't tell what it is to save your electrons from burning...

> > > But that's not the problem, since your 386 worked well, you
> > > said, and I guess you have more spare cycles than a whole 386
> > > can offer for backup.
> 
> > If it's NOT the libs, then it MUST be something else? If so, what? :-)
> 
> Now, HERE is a question.  I assume you do religiously retension
> your tapes and clean the heads every couple of hours of
> operation? 

Religiously... infact, my backup script retensions every tape every time it
calculates space left on tape. (once per archive)

> You don't have any temperature/humidity change
> problems with your tapes? 

Nope.

> Your tapes are quality tapes and
> reasonably new (so it's not a (series) of bad tapes that causes
> the problems)?  

3M tapes... 'nuff said.

> You IRQtune the tape on a very high priority.

Nope, but I did try with no change in results.

> Your floppy controler works.  The tape drives are not marginal. 

Correct.

> Well, either I do not spot the right hints now or I happen not
> to have enough information.

Ask me and I'll let you know what you need to know... I do believe that I
have given all info.

> Try waving dead chickens over the
> computer & the tape drives, or sacrifice some blood (yours or
> others).

Dead? No. Cut a live one's head off and splash it with the blood...

> > Anyone else try this? I don't think so... It never hurts to try, because
> > it's already broken...
> 
> Hmmm ... You have (the space for) a backup root?

Yup.

>  Wipe it, copy your current root to there (cp -a or tar | tar or something 
(but not dd)).  Then the backup root is not fragmented.  Boot from it.
> Test it.

It isn't fragged now because it was recovered... :-)
I have to MAKE a fragged fs to test.

[SNIP]

> > > *lightbulb* 
> > > Hey, if it was really fragmentation, how do we explain that one
> > > of them runs OK at 1000 and the other doesn't? 
> 
> > Time sensitive drive? 
> 
> TR-1 should be less sensitive.

The tape, mebby, the drive? doubtful. tr-1's read fine in the tr-3 deck, at
the higher speed... so it's gotta be the tape deck... perhaps the IPS of the
tape is slightly different?

> > > Do you drive them at different IRQ's on separate controlers? 
> 
> > No. I yank the drive off the same cable.
> 
> Hmmm.  
> 
> > > What happens if
> > > you strip down the machine bit by bit until the other tape works
> > > too? 
> 
> > Can't... I have others using this box and feeding 10 others, remember?
> > I'm limited in what I can and can't strip out... Downtime is a no-no.
> 
> Damn, you are stubbornly loyal to your users.  How much do they
> pay you?

More than you could afford. I'll prolly be having ~1000 or more soon too.
This is why I gotta have something that works, untill I get time to get a
better backup device... however, if I can help out with this project with
observations... I can do that :-)

> How many times do they thank you and make you feel special?  

Every day.

> How often do they bring you presents and new hardware?

All the time. I'm serious too! :-)

> :-)
> 
> You could always hook the others on the 386 you used for
> backup :-)

Nope. that would be insulting.

> > > It could be anything (including bad connections), since
> > > you opened the machine and did your stuff.
> 
> > Case opened? no no... my drive sits ontop of the system, and is not in any
> > enclosure.  I have an external connector to the floppy controller.
> > I do this so I don't have to constantly dick around opening the case, even
> > tho it's not a big deal, cause the case is a pizza box style one.
> > So it's the same cables, etc, only the drive changed.
> 
> Ok, recheck THAT connector and it's partners, clean it in
> isopropylalcohol, etc.

Been there, done that. even tried a second cable, which gave same results.

> > > You may know, but does Joe Random Reader know?  What's wrong
> > > with spreading clues (it's how I learned a lot)?
> 
> > Nothing. Infact, I am.... and I'm causing you to think, aren't I?
> 
> Sorry to bust your bubble, but ... problems make me think if
> they are clearly described (I'm not much of a mind reader).  :-)

Better buy a crystal ball...

> > >  And how come you speak Lojban?
> 
> > It's less wordy than Klingon?
> 
> http://xiron.pc.helsinki.fi/lojban/lojbroch.html
> (I think so.  I don't speak Klingon, it has to many ' 's anyway.)

Glanced over that page... kinda funny...

> > Nore will I, actually... Next attempt will be IDE. :-)
> 
> Who pays you, IDE? :-)

Shhhh! Quiet! Let's just say I'm improving the interface, and leave it at
that.

> > > Ah, that's a plan.  Oh, while you are on it, how about building
> > > a complete working ftape/afio ramdisk ... if it's in a ramdisk
> > > it won't hinder the IDE drive, and you can control the amount
> > > of fragmentation easily.
> 
> > WOO, I could do that VERY easily and quickly from my backup partition...
> > just force it to load it into ram... excellent idea!
> 
> Yep, you want the ramdisk as rootdisk.  If you want to use your
> backup root (you can also construct a new one on a free
> partition, disused swapspace, floppy or RAM ... that way it'll
> compress better, because empty areas are really just zeros.  You
> may want to strip everything before/while copying if you want a
> bootdisk):
> 
> Make the filesystem you want to use. 
> See that your Kernel runs ramdisks and initial ramdisks, floppies
>       (for bootdisks) and your choosen FS (minix needs less
>       space) without modules
> dd if=/dev/FILESYSTEM(e.g. /dev/ram) bs=1k (so you see the size) | \
>       gzip (-9 if you want) \
>       > /whereever

Ok, that's great... (I know how to do this... fine for others to see...
no rude comments... etc...)
 
> LILO wants to be told which ram-image to use, so you need a FS
> on the boot medium ... there's a good description in the LILO
> docs IIRC. 
> 
> Oh, and read up on the /linuxrc mechanism. 
> 
> > > Interesting ... you swapout AFTER ftape is loaded and you replace
> > > 'buffer' by more kernel memory buffers.
> 
> > Because I've never had a problem allocating 3 buffer minimum...
> 
> I don't have (usually) unless I have a long uptime.  You need to
> swapout more than 7/8 of the RAM completely(!) to guarantee the
> continous buffers will be found in a worst case scenario.  Yet
> another reason against floppy tapes and 80x86.

With 128M, I don't have that happen :-) Mebby you need more memory :-)

> > >  But I don't understand
> > > why you swapoff-swapon ... you just shorten the amount of RAM
> > > you have aviable.
> 
> > Habit, I guess :-) I don't like touching swap, unless it's absolutly
> > needed... I rather have it chew into the cached disk buffers.
> > Swap can affect total performance on IDE drives. I like to make damn sure
> > I'm not into swap when banging on the tape deck as to ensure that the tape
> > deck gets as much attention as possible from the cpu and os.
> 
> OTOH you shorten ram and can thus force the buffers to be
> flushed.  Since only crap that is not used for a LONG time (esp.
> in your case) lands in swap, just leave it that way.  Using
> buffer will cache almost any problems that can happen with swap
> hits.  At 1000 kbit/s (did I write Mbit earlier? *bonk self*)
> you get 125 Kb/s.  With your 16 DMA buffers you have 16*32Kb =
> 512 Kb buffer (4 seconds).  With a 10 MB buffer (no prob for
> you) you have 'just' 81.9+4 secs reserve.  I doubt that swap can
> take that long.  

Your correct about the timings, I do the buffers like that to prevent the
shoe-shine crapola that ftape users are all to familiar with. The problem is
not that, infact, it's something unrelated... prolly DMA fights with IDE
and other anomolies, like IRQ and the software timings, drives with no
tolerance to timing... It's prolly a combonation of all of it, actually.


> > > > Anything else you need to see ( except my passwords HAHAHAH )? Let me know :-)
> 
> > > Your SSN? :-)
> 
> > *ROFL*
> > Atleast you have humor... nice to see that these days of serious butt
> > nuggets, of whom I dispise... :-)
> 
> Humor?  Damn, I was trying to mask it as sarcasm.  Here goes
> the planet^W^Wmy image. 

Why does one want an image like that?

> But since you asked:  You could wade through the output of
> ftape (with the debug turned all up, perhaps) and pick out
> relevant parts. 

Doing so for whatever reasons makes the tape act worse...

> BTW, do you want to know my SSN?  *looks expectingly*

Phone 1-800-U-BYTE-ME and tell them... :-)

Yours,
-- 
 http://dr.ea.ms       http://IDE.cabi.net       http://startrek.off.net
 http://CPM.doa.org
                                ________________
-= Andrew Kroll =---------------\              /-----------------------------
Tired of Bill Gates?       LL    \            /Think Bill is getting MY CASH??
Win '95 sucks! DOS is OK. LL     II  NNNNN  UU UU  XX XX Linux! A free Un*x
Want to turn your PC intoLL     II  NN NN  UU UU   XXX  clone for 386/486/P5's
a powerful workstation? LLLLL  II  NN NN  UUUUU  XX XX  FINALLY A -=REAL=- OS!
-------------------------------------\    /--= <[EMAIL PROTECTED]> =--
                                      \  /
                                       \/
                        !FREE! At your favorite FTP site!
 

Reply via email to