From you discription of the drives thrashing and things taking too long to load, there are two things to check out. First, do you have a swap partition. Second look in the yast hardware section and check if you have dma enabled. If your hard drive is running in PIO mode instead of using bus-mastering, it will be slow.
Another thing to check is in the power saving section. Does the drop down selection under the AC power side have performance selected?

Were there any background cron processes running at the time. For example, there is a process that records the names of all of the files on your hard drive. On a fresh install, this may run for a while before the database is updated. Later on It won't
need to update as much. On a server that is on 24/7 the daemon would run around 3 in the morning, however with a computer that is turned on or off, the anacron service will kick in and start the updatedb command. So if the computer is doing two disk intensive operation at once, you can get thrashing.

You can check which application is slowing down the computer by using the the top command.

i assume you are using a RPM type package with no auto depenedency solving ?

there are pograms whach can do this for you, for example emerge in gentoo. (im not recomending you try gentoo, because it is aimed at experts)

but to install a program.. ANY porgram... i just type...

emerge (program name)

then the program and its dependency's will automatically be downloaded and installed...
so whatever your porgram is, i could have started the install in 2 seconds..... but depending on the internet connection and porcessor speed, it will take a little more time to install,,, but while thats happening, it be wwathching ty, or posting here...

what distro are you using ? there will rpobably be such a porgragram as gentoo's emerge for your distro.

Okay, to get back to my original predicament...

Clearly the best thing to do would be to install more RAM. Installing Slackware / Gentoo / VectorLinux / [insert non-newbie distro here] is not really an option, given that I have enough trouble getting easy-peasy SUSE to work. The problem is that right now I am a poor student and I just can't afford the RAM. So, to go thru your ideas for speeding up SUSE (for which I am very grateful)...


Postfix disabled, as recommended. Which of the following would it be safe to disable?:

"cron" - cron job service;
"fbset" - framebuffer setup;
"nfsboot";
"nscd" - name service cache daemon;
"portmap" - DARPA port to RPC program number mapper;
"random" - script to snapshot random state and reload it at boot time;
"resmgr" - state resource manager for device file access;
"smbfs" - import remote SMB / CIFS (MS Windows) file systems;
"splash" - splash screen setup;
"splash_early" - kills animation after network start;
"splash_late" - starts animation before shutdown;
"sshd" - start the sshd daemon


The asterisk is next to "mdma2". There is udma1 and udma2 but no udma5. All this, I'm afraid, means nothing to me (I must emphasise this again...) But whatever DMA is, it seems to be working.


Yes.


This was interesting. Right now, I have nothing but Firefox running. Here are the memory figures:

27% firefox-bin
10% X
4% kdeinit
..followed by around 40 other items, including some with multiple instances (such as "kdeinit" and "mingetty"). Could this yield any clues? At the very least there is one process running (which?) which accesses the hard-drive ("tchik tchik") every 4 seconds, all the time.

Two other possibly-related problems I can't resist burdening you with:

1. In Firefox there is a 'mousestroke delay' - as in, every manipulation I make with the mouse (menus, buttons) takes a second to register. This is really very annoying, and it doesn't happen at all in Konqueror.

2. In GAIM there is a keystroke delay, which makes typing fast all but impossible.

Are these directly related to the speed issue or can I zap them somehow? Are they not to do with 'interrupt requests' or some such thing?

As usual, all ideas would be really, really appreciated...

Rollo

rollo, you need to post some of these files so we can look at them. I'd like to see the output of "/sbin/hdparm -I /dev/hda", to do that su to root and issue this command "/sbin/hdparm -I /dev/hda > hdparm.txt" (no quotes)this will create a file called hdparm.txt in your home directory, copy and paste that with your next post. You can also look at what DMA the harddrive is set at in windows and compare between windows and Linux. I think it can go to udma instead of mdma which would probably solve your problem.

Few things i do on a stock install of slack.

1. Upgrade Fluxbox to 0.9.10
2. Recompile the Kernel, Removing bloat and optomize it for i686.
3. Sometimes, I recompile X for speed. But, normally i don't.

zajelo3 thanks very much for your help.


/dev/hda:

ATA device, with non-removable media
Model Number: IBM-DADA-26480
Serial Number: ZC4ZCHD7013
Firmware Revision: AD6OA4AA
Standards:
Used: ATA/ATAPI-4 T13 1153D revision 17
Supported: 4 3 2 1
Configuration:
Logical max current
cylinders 13424 13424
heads 15 15
sectors/track 63 63
--
CHS current addressable sectors: 12685680
LBA user addressable sectors: 12685680
device size with M = 1024*1024: 6194 MBytes
device size with M = 1000*1000: 6495 MBytes (6 GB)
Capabilities:
LBA, IORDY(can be disabled)
Buffer size: 460.0kB bytes avail on r/w long: 4 Queue depth: 1
Standby timer values: spec'd by Vendor, no device specific minimum
R/W multiple sector transfer: Max = 16 Current = 16
Advanced power management level: 128 (0x80)
DMA: sdma0 sdma1 sdma2 mdma0 mdma1 *mdma2 udma0 udma1 udma2
Cycle time: min=120ns recommended=120ns
PIO: pio0 pio1 pio2 pio3 pio4
Cycle time: no flow control=240ns IORDY flow control=120ns
Commands/features:
Enabled Supported:
* NOP cmd
* READ BUFFER cmd
* WRITE BUFFER cmd
* Host Protected Area feature set
* Look-ahead
* Write cache
* Power Management feature set
Security Mode feature set
* SMART feature set
Advanced Power Management feature set
Security:
supported
not enabled
not locked
frozen
not expired: security count
not supported: enhanced erase
18min for SECURITY ERASE UNIT.


What do you make of that?

I googled it and found a post where a guy with the same HD posts his hdparm info. If you look at the DMA modes his is set at udma2. For some reason yours isn't being set correctly. When you get it set right it will make a noticeable improvement. Let me look into how you reset this setting and I'll post with that info. Unless somebody else knows how to do it off the top off there head. I know I've read stuff about this on the Linux Audio Users list as checking dma settings is one of the first places to look when diagnosing poor recording performance. Anyway I'll look into it.

Model=IBM-DADA-26480, FwRev=AD6OA42A, SerialNo=ZC2ZCHL0493
Config={ HardSect NotMFM HdSw>15uSec Fixed DTR>10Mbs }
RawCHS=13424/15/63, TrkSize=0, SectSize=0, ECCbytes=4
BuffType=DualPortCache, BuffSize=460kB, MaxMultSect=16, MultSect=16
CurCHS=13424/15/63, CurSects=12685680, LBA=yes, LBAsects=12685680
IORDY=on/off, tPIO={min:240,w/IORDY:120}, tDMA={min:120,rec:120}
PIO modes: pio0 pio1 pio2 pio3 pio4
DMA modes: sdma0 sdma1 sdma2 mdma0 mdma1 mdma2 udma0 udma1 *udma2
Drive Supports : ATA/ATAPI-4 T13 1153D revision 17 : ATA-1 ATA-2 ATA-3 ATA-4
Kernel Drive Geometry LogicalCHS=789/255/63 PhysicalCHS=13424/15/63

Oh and before you change any disk settings go ahead and run "/sbin/hdparm -Tt /dev/hda" as root and record the results.

Ok, that was painless, here's a couple links to where I got my info from:

http://linux.oreillynet.com/lpt/a/li...29/hdparm.html
http://ccrma.stanford.edu/planetccrm...00000000000000

Anyway, the way to change the DMA setting with hdparm is to use the "X" option, like this:
"/sbin/hdparm -X33 /dev/hda" Issue this command as root and it should change the dma setting for you. But it will only set it to "udma1" not "udma2" as the specs on your hard drive say that it's an ATA-33 and that corresponds to "udma1". I guess the other guy had his set wrong when it showed "udma2". There are more tweaks in those pages that I gave the links for, so read them and see what else you can do. If that command changes the dma setting you will wan't to add that command to a startup script, so when you boot your machine it gets executed. On FC2 I would need to add it to "/etc/rc.d/rc.local" but I don't know what it would be on Suse. If it doesn't work you might only have a 40 wire ide cable, so then you would need to buy and install a 80 wire cable to be able to adjust the dma setting.

/dev/hda:
Timing buffer-cache reads: 184 MB in 2.03 seconds = 90.52 MB/sec
Timing buffered disk reads: 4 MB in 5.84 seconds = 701.48 kB/sec

Did that. Here's the new report.

/dev/hda:

ATA device, with non-removable media
Model Number: IBM-DADA-26480
Serial Number: ZC4ZCHD7013
Firmware Revision: AD6OA4AA
Standards:
Used: ATA/ATAPI-4 T13 1153D revision 17
Supported: 4 3 2 1
Configuration:
Logical max current
cylinders 13424 13424
heads 15 15
sectors/track 63 63
--
CHS current addressable sectors: 12685680
LBA user addressable sectors: 12685680
device size with M = 1024*1024: 6194 MBytes
device size with M = 1000*1000: 6495 MBytes (6 GB)
Capabilities:
LBA, IORDY(can be disabled)
Buffer size: 460.0kB bytes avail on r/w long: 4 Queue depth: 1
Standby timer values: spec'd by Vendor, no device specific minimum
R/W multiple sector transfer: Max = 16 Current = 16
Advanced power management level: 128 (0x80)
DMA: sdma0 sdma1 sdma2 mdma0 *mdma1 mdma2 udma0 udma1 udma2
Cycle time: min=120ns recommended=120ns
PIO: pio0 pio1 pio2 pio3 pio4
Cycle time: no flow control=240ns IORDY flow control=120ns
Commands/features:
Enabled Supported:
* NOP cmd
* READ BUFFER cmd
* WRITE BUFFER cmd
* Host Protected Area feature set
* Look-ahead
* Write cache
* Power Management feature set
Security Mode feature set
* SMART feature set
Advanced Power Management feature set
Security:
supported
not enabled
not locked
frozen
not expired: security count
not supported: enhanced erase
18min for SECURITY ERASE UNIT.

So it only changed the setting from mdma2 to mdma1.

Did a few more speed tests - results vary wildly, not sure this proves anything.

/dev/hda:
Timing buffer-cache reads: 188 MB in 2.02 seconds = 93.27 MB/sec
Timing buffered disk reads: 12 MB in 3.02 seconds = 3.98 MB/sec

/dev/hda:
Timing buffer-cache reads: 156 MB in 2.03 seconds = 76.67 MB/sec
Timing buffered disk reads: 4 MB in 4.17 seconds = 983.11 kB/sec

/dev/hda:
Timing buffer-cache reads: 172 MB in 2.02 seconds = 84.99 MB/sec
Timing buffered disk reads: 6 MB in 3.73 seconds = 1.61 MB/sec


You're right, tweaking this setting might be a sneaky way to speed up my snail-slow system. But I read all the warnings and I don't fancy risking system failure typing commands I don't really understand!

So far everything is still working.. Is there a difference between UDMA and MDMA? Any more ideas about relatively unrisky commands to try to shift the settings?

Regarding the Suse equivalent of Fedora's boot script file, do you think it could be /etc/init.d/boot.local? At the moment there's nothing in it but some blurb including this:

# /etc/init.d/boot.local
#
# script with local commands to be executed from init on system startup
#
# Here you should add things, that should happen directly after booting
# before we're going to the first run level.
#

Try "/sbin/hdparm -X66 /dev/hda" as I just read another post where your model of disk is being set to udma2. But I would say this is definetly your speed problem. You went from mdma2 to mdma1 and that's going in the wrong direction we want to be at udma2 or at least udma1.

Okay, doing this at the command line produced the information:

/dev/hda:
setting xfermode to 66 (UltraDMA mode2)

But the new information file indicates:

DMA: sdma0 sdma1 sdma2 mdma0 mdma1 *mdma2 udma0 udma1 udma2

I.e., back to original setting.

Some new speed tests for comparison:

/dev/hda:
Timing buffer-cache reads: 220 MB in 2.02 seconds = 108.77 MB/sec
Timing buffered disk reads: 12 MB in 3.19 seconds = 3.76 MB/sec

/dev/hda:
Timing buffer-cache reads: 192 MB in 2.01 seconds = 95.44 MB/sec
Timing buffered disk reads: 18 MB in 3.09 seconds = 5.83 MB/sec

/dev/hda:
Timing buffer-cache reads: 180 MB in 2.01 seconds = 89.57 MB/sec
Timing buffered disk reads: 8 MB in 4.31 seconds = 1.86 MB/sec

/dev/hda:
Timing buffer-cache reads: 164 MB in 2.02 seconds = 81.00 MB/sec
Timing buffered disk reads: 10 MB in 3.13 seconds = 3.20 MB/sec

Could that be an improvement, or has nothing changed? Any ideas about the config file?

Gotta go to bed, it's pretty late over here.. Will do more speed comparison tests using the last command tomorrow and post the results here.

I really appreciate all your assistance zajelo3. Good night...

Yes that sounds like the equivalant of /etc/rc.d/rc.local. You really should be getting alot better disk performance. Try looking in your system logs and see if there's any reference to "DMA" anywhere. You could try "/sbin/hdparm -X100 /dev/hda" and see if that gives any gains.

I'm running SuSE 9.1 and don't consider it "bloated".
When I installed Redhat, it loaded a lot more extra programs than SuSE. I had to manually install a lot more stuff for SuSe. I just wish I had the right driver for my old G773 ViewSonic monitor, the one that came with SuSE doesn't work!

/sbin/hdparm -X100 /dev/hda produces:

>setting xfer mode to 100 (unknown, probably not valid)

And doesn't really affect the transfer speed. Nor does there seem to be any difference between speeds at X66 and X33 (have done a few tests) - but then perhaps that's because -X66 is not really changing the setting to UDMA despite the feedback. It's possibly a hardware problem because the feedback contradicts the information log.

Where might I find these logs? I really don't know where to look for stuff on Linux and and this speed running a search of the whole system would put the thing out of action for an hour...