On Thursday 04 March 2010, birger wrote:
>On Wed, 2010-03-03 at 22:47 -0500, Gene Heskett wrote:
>> In this case, same motherboard, same sata0 connector. This board has 6
>> or 7 sata ports. 4 in use ATM.
>
>But not the same drive, right?
No, a 60 day old, 500Gb 3Gbs drive that failed, vs a fresh 1T 3Gbs drive,
both Seacrates.
>So I wouldn't exclude the rather obvious explanations that the drive
>could be slower, failing or that the drive could be set to some
>'compatibility mode' geometry through jumpers that yielded a non-optimal
>file system. Another possible culprit would be the cables.
>
>I think that blaming diskdruid for this kind of speed difference without
>actually considering the more obvious possibilities is a bit weird. Did
>you repartition your slow disk manually and rsync back to see if you got
>it faster?
By then the drive was out of spares, and I seriously doubt it could have
completed the writeback. The clock was ticking and I was getting emails from
smartd at 20 minute intervals. Junk is junk.
>Errors in file system layout on old (slow) disks *can* make a big
>difference. One problem is that some disks have been set up to lie about
>their real geometry in order to work with an old BIOS. Optimizing a file
>system layout for such a disk would be impossible without knowing the
>details of how the disk maps the advertised geometry to the real one.
>Check your disk documentation and use the optimal jumper settings if
>your BIOS can handle it.
What disk docs? The only thing they put in the box these days is the legal
disclaimers. As for the bios, well, its an ASUS, what can I say. The latest
bios update took 3 burns before it worked right.
>When you really know what is going on right down to the platter you can
>optimize a lot. I wrote my own program to low-level format floppies on
>my BBC B back in the mists of time. 30% speed increase for sequential
>IO. :-)
Done that quite a bit on os9, now nitros9, so that is not news. Fiddling
with the interleave settings and segment allocation sizes can make a lot of
diff for floppy media. In fact, here is a short screen snip of a minicom
screen here on this box:
{t2|08}/DD/NITROS9/3.2.9:
That prompt came from a CoCo3 in the basement. Clock speed? 1.79 mhz.
And another:
{t3|07}/DD/NITROS9/3.2.9/MODULES/SCF:dir
Directory of . 2010/03/03 09:35
cogrf.io covdg.io covdg_small.io cowin.io ftdd.dd
joydrv_6551L.sb joydrv_6551M.sb joydrv_6552L.sb joydrv_6552M.sb joydrv_joy.sb
keydrv_cc3.sb n0_scdwn.dd n10_scdwn.dd n11_scdwn.dd n12_scdwn.d
[...]
FWIW, I wrote joydrv_6551L.sb, which allows me to use a serial mouse on a
TRS-80 Color Computer.
That particular link, 't3', has bluetooth for its cabling.
>For the case you refer to I would still first check and recheck the
>configuration of that disk, and then try a fresh F12 install. And I
>would just use the default DD partitioning.
First error, recommending DD. It will not allow a /boot partition to exceed
199 megabytes. My present, and fully functional /boot is 400. And its 56%
used. I should to some housekeeping I guess. Delete the old 2.6.32 kernels,
they are so old hat in 2010.
It will not allow swap to be the next partition, in order to put it on a
faster area of the disk. To DD, swap must be on the center of the disk, its
slowest possible position. I could go on, but its coals to newcastle on this
list.
>If the disk really is weird
>enough that DD can't optimize for it I would just buy a new disk. They
>are so cheap these days I wouldn't bother spending much time on it.
Exactly what I did. That disk was one of the 'needed a software update'
models, which I did do to it. Funny thing, the oldest drive in this box is a
400Gb Deathstar. I believe, if there is SS still available when it retires,
it may qualify. The one 1T WD drive isn't a year old yet, and already been
warrantied once, nor is either of the 1T Seacrate's.
As far as claiming DD 'optimizes', that is a stretch, applicable to Joe
Lunchbucket if he doesn't know any better. He, if he gives a s--t, will
learn eventually. But he is also used to being picked on, cuz he'll be
thinking in a winderz frame of mind. I am not, and never will be, a member
of that camp.
Some folks forget that not all of us are escapees from the M$ camp. I was
carving code for an RCA 1802 in the late 70's, and by the mid-80's had
'graduated' to a multi-user, multi-tasking os, writing what I needed in
several languages including assembly. When it became obvious that the 'coco'
didn't have the stuff to do real networking, then a big box amiga came to
live here, one that I just sold last week. From there to linux in 1998, I've
owned one winders machine in that time frame, a laptop I bought and
immediately put fedora on. Now, even the coco can be a web server, they
wrote the server in basic09 in about an hour.
>birger
>
Thanks, and be well.
--
Cheers, Gene
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Duty, n:
What one expects from others.
-- Oscar Wilde
--
users mailing list
users@xxxxxxxxxxxxxxxxxxxxxxx
To unsubscribe or change subscription options:
https://admin.fedoraproject.org/mailman/listinfo/users
Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines
