Anybody know how extract the exact number of kilobytes
written to a disk (say /dev/hda) since boot??
iostat will tell you the number of _blocks_ but sez that
"a block is of indeterminate size"
The best I can come up with is a fudged value: multiplying
the number of blocks times the average block
Somewhere between v1.14 (rh6.1) and v2.04 (rh7.1), bash stops
providing per-line history when doing multi-line "compound
command" type stuff like...
for i in *; do
echo $i
done
That is to say, with v1.14, I can recall each individual line
listed above with the up arrow key; whereas v2.05 g
okay, i'm pissed and have a lot to do before turkey
day vaca so i'm being lazy and buggin you all...
i just tried to rip a harry potter cd with itunes
and it wedged real good--i want to get it on my ipad
before tomorrow so traveling alone with my daughter
is a breeze. (and i wanna listen to the n
> On Sun, Nov 24, 2002 at 05:27:56PM -0600, steve rader wrote:
> > It's been about a dozen years since I got my BS in
> > CompSci, but my recollection is...
> >
> > used is used by user-space processes
> > sharedis used for shared memor
The "free" command reports real memory util as
used, shared, buffers and cached. But what are the
definitions of the terms? There's no mention of them
in the man pages for free, top, or vmstat.
It's been about a dozen years since I got my BS in
CompSci, but my recollection is...
used is
> From: Douglas J Hunley
> anyone got decent pointers on setting up and configuring raid using
> disksuite on solaris? thanks!
it's not exactly doc, but here's my recipe for
cooking solaris7 software raid1 partitions with
disksuite... and a little perl script that you can
use under cron so yo
> > begin steve rader <[EMAIL PROTECTED]> spewed forth:
> > Here's another reason to avoid dd'ing whole partitions: since
> > it copies the actual file system, it forces the destination
> > file system to be exactly the same size as the source.
>
So I'm out in the brave new world of... redhat 7.2...
and I find the brute-force-and-ignorance way to get
"ls" to use old-school (case sensitive) sort order
is to unset LANG.
Is the most elegant solution to use
LC_COLLATE=POSIX
?
Floundering in a i18n world... whatever that means...
no, rea
Here's another reason to avoid dd'ing whole partitions: since
it copies the actual file system, it forces the destination
file system to be exactly the same size as the source.
In other words, if you dd a 2 gig file system into a 4 gig
partition then you end up with a 2 gig file system (and thus
> > > > >On Wed, 6 Nov 2002, steve rader wrote:
> > > > >Also, here's one more reason to avoid tar: it has a 100 char
> > > > >max file name length restriction. I suspect cpio and certainly
> > > > >dd don't have th
> > On Wed, 6 Nov 2002, steve rader wrote:
> > >Also, here's one more reason to avoid tar: it has a 100 char
> > >max file name length restriction. I suspect cpio and certainly
> > >dd don't have that problem.
> Net Llama! wrote:
> > H
> From: Net Llama!
> OK, thanks for the explanation. My solaris knowledge is admittedly weak.
> Anyway, doesn't iostat do this under Linux?
Looks like linux iostat still can't report % io wait
cuz the kernel disk subsystem doesn't count "jiffies"
(ms or cycles or clock ticks or somesuch?) whe
> > Net Llama! wrote:
> > Errr...what is % io wait?
> From: Jim Bonnet
> I believe he is referring %wio which is the percentage of time the cpu
> is idle with processes waiting on I/O..
right... imho, it's a _critical_ statistic for system
performance tuning... if there's, a, ah, "signif
About four years ago, mercilessly bugged Stephen Tweedy about
the lack of % io wait counters in the ext2 file system kernel
stuff. IIRC, he said he might put it in.
/proc/partitions and sysstat-4.0.1's iostat on 2.4.x systems
lead me to think we still can't watch/measure % io wait
under Linux.
> > > > From: steve rader
> > > > - (cd /master; tar -cf - .) | (cd /slave; tar -xpf -)
> > Net Llama! <[EMAIL PROTECTED]> wrote:
> > It should be faster with dd:
> > dd if=/dev/sda7 of=/dev/sdb1 bs=1024
> From: Roger Oberholtzer
> It's entirely possible... I figured it out... I have an expect
> script to write the partition table and a bash script to do
> the rest... the write-MBR-on-/dev/?db magic is...
>
> mount /dev/hdb1 /slave
> vi /slave/etc/lilo.conf # set...
>disk=/dev/hdb
>[...]
>root=/dev/h
> From: Jim Bonnet
> you've lost me.. What are you trying to do?
>
> Are you just trying to clone a disk so that you can distribute that
> clone to like hardware?
I'm cloning /dev/sda7 onto /dev/sdb such that I can then use
/dev/sdb as /dev/sda... in otherwords, I take a blank (or M$
infeste
> From: steve rader
> - (cd /master; tar -cf - .) | (cd /slave; tar -xpf -)
Btw, will this step go faster if I dd?? If so, what's
the correct way to copy, say, /dev/sda7 to /dev/sdb1?
(Assume both partition have the same si
>> > From: Net Llama!
>> > Drive designations are contreolled by your BIOS. You can't change these
>> > via software.
>> From: steve rader
>> Yea--understood. Are you saying it's not possible
>> to write a MBR to a non-boot disk
Does anyone know how (if possible) to tell lilo to write a MBR
to, say /dev/sdb (the "second disk" eg bios=0x81) such that it
will boot as /dev/sda?? I think I have tried all reasonable
combinations of chroot and lilo.conf disk= and bios= tricks
with no success.
Failing that, does anyone have a
20 matches
Mail list logo
