headers
Chris Fowler | 1 Apr 2003 01:15
Favicon

Re: CF compared to IDE

More info on the DOM

On Mon, 2003-03-31 at 15:45, Andy Warner wrote:

> Hi Jeff, please to meet you.
> 
> DOM write speeds are variable, but somewhat similar those of
> a raw HDA, but obviously all modern drives have a significant
> buffer between the host interface and the HDA; DOMs do not
> have any such buffering. Hence the write rate is usually on a
> par with a streaming data rates (not burst) to disk. Throughput
> can be bursty, as a sector erase takes place, or garbage
> collection takes place. Chris & I have seen evidence that
> some DOM units do not support DMA (or that the support breaks
> under some circumstances) - I don't know if this would be
> important to you.
> 
> I have not seen any DOM units that offer write speeds similar
> to the newer ultra compact flash units. I guess they'll appear
> if the DOM formfactor is judged a successful niche.
> 
> I have no idea what your application is, but obviously you
> need to pay close attention to write activity (even if it's
> just housekeeping activity.)
> -- 
> Andy Warner
> 
> andyw@...	 	Link Margin		(612) 801-8549
Permalink | Reply | 0 comments |
headers
Chris Ricker | 1 Apr 2003 01:16
Picon
Favicon

Re: Mem. reqs. for RH9 (and other distros)

On Mon, 31 Mar 2003, Fulton Green wrote:

> But that's not the point of this note.  Instead, it's about the machine
> requirements:
>    http://www.RedHat.com/software/linux/technical/
> 
> Does anyone else find it interesting that even the *text* mode now
> requires up to 64 MB of RAM?
> 
> If anyone has any insight on this (that means you, Chris Ricker :), I'd
> love to hear it.  I can only guess that the new threading stuff bumped
> up the memory usage.

I don't know that I have special insight on it ;-). At any rate, my guess
would be that it's mainly due to how anaconda (RH installer) does batch RPM
installs -- batched rpm installs take memory (amount proportional to number
of packages being installed), and that's gotten worse with new rpm's
concurrent access feature

You can get alternate installers for RHL which do things in a
less-memory-intensive fashion. Check out RULE
<http://www.rule-project.org/en/index.php>. The slinky stuff should mostly
work for 9 now, or soon if not yet (though I wouldn't necessarily expect
miniconda to be ported to 9).

> Or, perhaps, RH8 also required 64 MB minimum, though I remember a time
> when RHL only req'd 32 MB.

8 could install in 32 megs, with a little luck. RULE made it consistently do 
so (and also gave you the ability to install on < 586)....
(Continue reading)

Permalink | Reply | 0 comments |
headers
Chris Fowler | 1 Apr 2003 01:21
Favicon

Re: CF compared to IDE

Two key points about the PQI modules.

1) They do not support DMA

2) Depending on model, you can get units that use power
   on IDE PIN 20 to power them.   If your motherboard 
   supports (doubtfully) power on pin 20, then you will not
   need to plug the DOM unit into the PC PS.  We have a 
   embedded type board that supports pin 20 power and a Tyan
   MB that does not.

I've been satisified with read and write speeds.

You can on ebay purchase 3.5" CF -> IDE converters that will allow you
to take a standard CF unit and plug it into your IDE.  This maybe an
option for those that can not get or do not want DOM modules.

On Mon, 2003-03-31 at 16:14, hbbs@... wrote:
> Maybe I spoke too soon.
> 
> I Googled myself a datasheet for the PQI DOMs and I'm seeing a read of 2.5MB/2
> and a wrote of 1.2MB/s - I have 340MB IDE HDDs that go about that fast.
> 
> - Jeff
> > Wow, this totally rocks.  Just one of these things - even a small one - could do
> > a lot to enhance the performance of older PCs.  
> > 
> > - Jeff
> > 
> > > The DOM is a CF->IDE converter in a compact format.  It has a IDE
(Continue reading)

Permalink | Reply | 0 comments |
headers
Jim Popovitch | 1 Apr 2003 01:50
Picon
Favicon

RE: Redhat ISOs Available

Please let us know what the b/w peaks at tonight.  :)

THANKS!

-Jim P.

> -----Original Message-----
> From: ale-admin@... [mailto:ale-admin <at> ale.org]On Behalf Of Jonathan
> Glass
> Sent: Monday, March 31, 2003 5:49 PM
> To: ale@...
> Subject: [ale] Redhat ISOs Available
> 
> 
> If you want the RH 9.0 ISOs via HTTP, please feel free to download them
> from http://ibb-250.ibb.gatech.edu/. 
> 
> The MD5SUMS are in the "MD5SUMS" file.  How original.  :)
> 
> Thank you
> -- 
> Jonathan Glass
> Systems Support Specialist II
> Institute for Bioengineering & Bioscience
> Georgia Institute of Technology
> 404.385.0127
>
Permalink | Reply | 1 comment |
headers
Armsby John-G16665 | 1 Apr 2003 02:39

Occasional Extensive Disk I/O Question

I realize this is an akward question to answer but I will do my best to pose an intelligent question.

I have RedHat 8.0 running apache on ext3, 256 megs of RAM, standard desktop installation, addition of
apache, MySQL, postfix.  The web server gets hit a few hundred times a day.

Problem:  Occasionally after a reboot command the machine hard drive light illuminates for 3-4 hours. 
Eventually it goes off.  It is doing it as I write......

51 processes: 49 sleeping, 2 running, 0 zombie, 0 stopped
CPU states:  0.1% user,  0.1% system,  0.0% nice, 99.6% idle
Mem:   255452K av,  185172K used,   70280K free,       0K shrd,   21400K buff
Swap:  522104K av,       0K used,  522104K free                  141296K cached

  PID USER     PRI  NI  SIZE  RSS SHARE STAT %CPU %MEM   TIME COMMAND
 1201 root      15   0  1020 1020   840 R     0.3  0.3   0:00 top
    1 root      15   0   476  476   424 S     0.0  0.1   0:06 init
    2 root      15   0     0    0     0 SW    0.0  0.0   0:00 keventd
    3 root      15   0     0    0     0 SW    0.0  0.0   0:00 kapmd
    4 root      34  19     0    0     0 SWN   0.0  0.0   0:00 ksoftirqd_CPU0
    5 root      15   0     0    0     0 SW    0.0  0.0   0:00 kswapd
    6 root      25   0     0    0     0 SW    0.0  0.0   0:00 bdflush
    7 root      15   0     0    0     0 SW    0.0  0.0   0:00 kupdated
    8 root      25   0     0    0     0 SW    0.0  0.0   0:00 mdrecoveryd
   12 root      15   0     0    0     0 SW    0.0  0.0   0:00 kjournald
   68 root      16   0     0    0     0 SW    0.0  0.0   0:00 khubd
  160 root      15   0     0    0     0 SW    0.0  0.0   0:00 kjournald
  445 root      16   0  1044 1044   764 S     0.0  0.4   0:00 dhclient
  486 root      15   0   576  576   492 S     0.0  0.2   0:00 syslogd
  490 root      15   0   428  428   376 S     0.0  0.1   0:00 klogd
  507 rpc       15   0   532  532   460 S     0.0  0.2   0:00 portmap
(Continue reading)

Permalink | Reply | 3 comments |
headers
Jeffrey B. Layton | 1 Apr 2003 03:05

Re: Occasional Extensive Disk I/O Question

Armsby John-G16665 wrote:

>How can I find out what is going on?  Where do I look to so what process is writing/reading to the disk?
>
>Any help would be appreciated.
>

A couple of questions. What filesystem are you running?
Also, when the disk light is on, can you hear anything out
of the disk? Also, try running 'vmstat' to see what's going
on. Try something like 'vmstat 2 10'. Try it a couple of
times. What happens if you do a 'sync' when this is going
on?

Jeff
Permalink | Reply | 1 comment |
headers
Armsby John-G16665 | 1 Apr 2003 03:17

var/messages question

another question.....

My web server (Redhat 8.0, 256meg, apache) runs a C++ script which ran perfectly up until recently.  When I do
a query which will return a LOT of results (Google like search of an intralinet), the output from the server
seems to lock up.... I looked a the /var/messages and see this:

Mar 31 17:31:04 xena kernel: eth0: Transmit error, Tx status register 82.
Mar 31 17:31:04 xena kernel: Probably a duplex mismatch.  See Documentation/networking/vortex.txt
Mar 31 17:31:04 xena kernel:   Flags; bus-master 1, dirty 1254(6) current 1254(6)
Mar 31 17:31:04 xena kernel:   Transmit list 00000000 vs. cf0cb380.
Mar 31 17:31:04 xena kernel:   0:  <at> cf0cb200  length 800005ea status 000105ea
Mar 31 17:31:04 xena kernel:   1:  <at> cf0cb240  length 800005ea status 000105ea
Mar 31 17:31:04 xena kernel:   2:  <at> cf0cb280  length 800005ea status 000105ea
Mar 31 17:31:04 xena kernel:   3:  <at> cf0cb2c0  length 800005ea status 000105ea
Mar 31 17:31:04 xena kernel:   4:  <at> cf0cb300  length 800005ea status 000105ea
Mar 31 17:31:04 xena kernel:   5:  <at> cf0cb340  length 800005ea status 800105ea
Mar 31 17:31:04 xena kernel:   6:  <at> cf0cb380  length 800005ea status 000105ea
Mar 31 17:31:04 xena kernel:   7:  <at> cf0cb3c0  length 800005ea status 000105ea
Mar 31 17:31:04 xena kernel:   8:  <at> cf0cb400  length 800005ea status 000105ea
Mar 31 17:31:04 xena kernel:   9:  <at> cf0cb440  length 800005ea status 000105ea
Mar 31 17:31:04 xena kernel:   10:  <at> cf0cb480  length 800005ea status 000105ea
Mar 31 17:31:04 xena kernel:   11:  <at> cf0cb4c0  length 800005ea status 000105ea
Mar 31 17:31:04 xena kernel:   12:  <at> cf0cb500  length 800005ea status 000105ea
Mar 31 17:31:04 xena kernel:   13:  <at> cf0cb540  length 800005ea status 000105ea
Mar 31 17:31:04 xena kernel:   14:  <at> cf0cb580  length 800005ea status 000105ea
Mar 31 17:31:04 xena kernel:   15:  <at> cf0cb5c0  length 800005ea status 000105ea
Mar 31 17:31:10 xena kernel: eth0: Transmit error, Tx status register 82.
Mar 31 17:31:10 xena kernel: Probably a duplex mismatch.  See Documentation/networking/vortex.txt
Mar 31 17:31:10 xena kernel:   Flags; bus-master 1, dirty 1260(12) current 1260(12)
Mar 31 17:31:10 xena kernel:   Transmit list 00000000 vs. cf0cb500.
(Continue reading)

Permalink | Reply | 2 comments |
headers
Chris Fowler | 1 Apr 2003 03:36
Favicon

Re: Occasional Extensive Disk I/O Question

How much memory do you have and what are you running?  I see this alot
if I have KDE+setiathome running on a box and do not use it for a couple
days.  If I come back and move the mouse to turn off the KDE
screensaver, I get extensive i/o for about 30 seconds.

On Mon, 2003-03-31 at 20:05, Jeffrey B. Layton wrote:
> Armsby John-G16665 wrote:
> 
> >How can I find out what is going on?  Where do I look to so what process is writing/reading to the disk?
> >
> >Any help would be appreciated.
> >
> 
> A couple of questions. What filesystem are you running?
> Also, when the disk light is on, can you hear anything out
> of the disk? Also, try running 'vmstat' to see what's going
> on. Try something like 'vmstat 2 10'. Try it a couple of
> times. What happens if you do a 'sync' when this is going
> on?
> 
> 
> Jeff
> 
> 
> _______________________________________________
> Ale mailing list
> Ale@...
> http://www.ale.org/mailman/listinfo/ale
Permalink | Reply | 0 comments |
headers
Christopher Bergeron | 1 Apr 2003 04:14

help! i broke CTRL-C


Heck, at this point even a binary "filename" that should grep the 
scripts for could at least give me a start...

Much thanks,
-CB
Permalink | Reply | 6 comments |
headers
Christopher Bergeron | 1 Apr 2003 04:17

Re: var/messages question


Just my 2 copper plated, nickel cored, lincoln headed cents -
-CB

Armsby John-G16665 wrote:

|another question.....
|
|My web server (Redhat 8.0, 256meg, apache) runs a C++ script which ran 
perfectly up until recently.  When I do a query which will return a LOT 
of results (Google like search of an intralinet), the output from the 
server seems to lock up.... I looked a the /var/messages and see this:
|
|
|Mar 31 17:31:04 xena kernel: eth0: Transmit error, Tx status register 82.
|Mar 31 17:31:04 xena kernel: Probably a duplex mismatch.  See 
Documentation/networking/vortex.txt
|Mar 31 17:31:04 xena kernel:   Flags; bus-master 1, dirty 1254(6) 
current 1254(6)
|Mar 31 17:31:04 xena kernel:   Transmit list 00000000 vs. cf0cb380.
|Mar 31 17:31:04 xena kernel:   0:  <at> cf0cb200  length 800005ea status 
000105ea
|Mar 31 17:31:04 xena kernel:   1:  <at> cf0cb240  length 800005ea status 
000105ea
|Mar 31 17:31:04 xena kernel:   2:  <at> cf0cb280  length 800005ea status 
000105ea
|Mar 31 17:31:04 xena kernel:   3:  <at> cf0cb2c0  length 800005ea status 
000105ea
|Mar 31 17:31:04 xena kernel:   4:  <at> cf0cb300  length 800005ea status 
000105ea
(Continue reading)

Permalink | Reply | 0 comments |

Gmane