Jump to content

help to configurate server


debianuser
 Share

Recommended Posts

Ok guys...

I finally bought that server...

1Gb dual channel - Pentium D 3 Ghz - Asus P5WD2 Premium Motherboard and 500 Gb SATA - 3 Gb/s

I installed Debian on it and as far as i tested.. the performances are really not what I expected...

is it because some more tweaking need to be done especially to use the potential of the dual core and the SATA hdd?

if yes could you please guide me or give advice..

my SATA HDD is set up on 2 Primary channel...

thanks a lot

Link to comment
Share on other sites

:lol: Hooray for marketing!

SATA-3GB... You know what the 3GB/s is? The speed at which the drive's buffer (typically 8 MB. 16mb tops) can be accessed. Burst rate at that. The actual drive speed (as in getting previously unread data from the actual platter) will be more in the low 10s of megabytes. Use Bonnie++ to get some actual figures.

As for the dual-core, make sure you're running an SMP-capable kernel. When you run "cat /proc/cpuinfo | grep processor" there should be 2 lines (processor 0 and processor 1). If there isn't, you're not running an SMP-capable kernel.

Link to comment
Share on other sites

:lol: Hooray for marketing!

SATA-3GB... You know what the 3GB/s is? The speed at which the drive's buffer (typically 8 MB. 16mb tops) can be accessed. Burst rate at that. The actual drive speed (as in getting previously unread data from the actual platter) will be more in the low 10s of megabytes. Use Bonnie++ to get some actual figures.

As for the dual-core, make sure you're running an SMP-capable kernel. When you run "cat /proc/cpuinfo | grep processor" there should be 2 lines (processor 0 and processor 1). If there isn't, you're not running an SMP-capable kernel.

Hello Cooper - thanks for man for replying I was getting desperate here.

I got the for the cpu... one thing I did not get is about the SATA.. I use debian 3.1 with 2.6 kernel.. could it be that the drivers are just not working correctly?

and top of the head... that is really matter if i put the SATA disk on the secondary master channel? like is the speed affected whenever it's on the primary channel or secondary channel? that should not matter much, right?

thanks guys!

Link to comment
Share on other sites

SATA doesn't have the concept of 'primary' and 'secondary'. Your disk is the only device on the cable (there are exceptions, but they're pricy) so it's got a direct connection with the controller. Your controller presumably has 4 SATA ports to it. Each runs equally fast unless the chipset is crappy, or the system's bus has become sturated (not particularly likely at this stage).

Now with PATA, there IS the concept of primary and secondary drive on the cable. Because those 2 drives are duking out who gets to transfer data across the wire at any given point, any setup that involves a Primary and Secundary device on an IDE cable will be sub-optimal.

For some quick figures, run 'hdparm -tT /dev/sd<whatever>' which will perform a quick and dirty read test. For more meaningful numbers install and run Bonnie++ which will create files and directories on your system in an effort to measure total file system performance.

Please post the results, and specify which part of it is disappointing to you.

Link to comment
Share on other sites

SATA doesn't have the concept of 'primary' and 'secondary'. Your disk is the only device on the cable (there are exceptions, but they're pricy) so it's got a direct connection with the controller. Your controller presumably has 4 SATA ports to it. Each runs equally fast unless the chipset is crappy, or the system's bus has become sturated (not particularly likely at this stage).

Now with PATA, there IS the concept of primary and secondary drive on the cable. Because those 2 drives are duking out who gets to transfer data across the wire at any given point, any setup that involves a Primary and Secundary device on an IDE cable will be sub-optimal.

For some quick figures, run 'hdparm -tT /dev/sd<whatever>' which will perform a quick and dirty read test. For more meaningful numbers install and run Bonnie++ which will create files and directories on your system in an effort to measure total file system performance.

Please post the results, and specify which part of it is disappointing to you.

alright.. I will do the test and post them...

one more thing I need to specifiy is why then in the BIOS - my ide atapic cd rom is a primary master and my hdd is a secondary master?

thanks

Link to comment
Share on other sites

alright Cooper

I did the test - for the cpu - it was only one line

cat /proc/cpuinfo | grep processor

processor       : 0

this is the test of the hard disk

hdparm -tT /dev/hdc1



/dev/hdc1:

 Timing cached reads:   4700 MB in  2.00 seconds = 2348.01 MB/sec

 Timing buffered disk reads:   10 MB in  3.12 seconds =   3.21 MB/sec

so for the cpu - i just need to install a SMP-capable kernel?

what do you think of the performances of the disk?

thanks mate

Link to comment
Share on other sites

I did the test - for the cpu - it was only one line

cat /proc/cpuinfo | grep processor

processor       : 0

So you're running a uni-processor kernel when you have a dual-processor on board. You need to install a different kernel that supports SMP. In the mean time you're only using 1 of the 2 cores and things will be understandibly slower.

this is the test of the hard disk

hdparm -tT /dev/hdc1



/dev/hdc1:

 Timing cached reads:   4700 MB in  2.00 seconds = 2348.01 MB/sec

 Timing buffered disk reads:   10 MB in  3.12 seconds =   3.21 MB/sec

Christ in a basket, I can shout morse code faster than that!

But let's start with the simple bit first. You know that SATA harddisk you've got? Well, it isn't. You've got a PATA harddisk (you know. Wide gray cable as opposed to a thin one). And from the looks of it you've got the DMA off.

Read the man page on hdparm. Start off with running it with only the device as parameter. No switches. Go from there. Presumably you want to try the '-d1 -X66' parameters. Note that getting aggressive with these settings can easily lock up your system at the very first contact with the harddisk after running the command, so a quick 'sync' before trying it would be a very good idea.

You should eventually see something like this:

 Timing cached reads:   2284 MB in  2.00 seconds = 1141.93 MB/sec

 Timing buffered disk reads:   70 MB in  3.07 seconds =  22.79 MB/sec

That's an old dog of a drive. 20 gigger which should tell you enough. Here's the same on a 200 gig SATA drive, though please understand, it IS a Western Digital drive (aka: rubbish!)

 Timing cached reads:   2280 MB in  2.00 seconds = 1139.93 MB/sec

 Timing buffered disk reads:  176 MB in  3.03 seconds =  58.12 MB/sec

Here's the same for a quality 200 gig Seagate drive:

 Timing cached reads:   3880 MB in  2.00 seconds = 1939.60 MB/sec

 Timing buffered disk reads:  224 MB in  3.01 seconds =  74.47 MB/sec

The reason I know you're not using SATA is because SATA falls under the SCSI subsystem in Linux, and as a result your device node should've been /dev/sda instead of hda.

Link to comment
Share on other sites

Hello Cooper

thanks for the reply

I do use a SATA ( i think I know the difference betweeb SATA and IDE ;-) )

well I tried to install SUSE and the device as you said is "sda" ... tried Ubuntu Server version.. the same "sda"

so is Debian having problem to recogize the SATA disk.. i though that was fixed starting with kernel 2.6

with Suse or Ubuntu .. the dual core is recognized

and under both linux dist. the hdparm gives

Timing cached reads:   4704 MB in  2.00 seconds = 2352.00 MB/sec

 Timing buffered disk reads:   204 MB in  3.02 seconds =   67.55 MB/sec

so what do you think?

i did a test to copy a 800 MB file and it copied it from a mac client to the server in fairly 20-30 sec

please give feedbacks

thanks a lot

Link to comment
Share on other sites

Then the only thing I can think of is that in the kernel you originally used you managed to activate the setting labelled:

Support for SATA (deprecated; conflicts with libata SATA driver)

So yeah, bad idea to use that.

The speeds for the harddisk now look about right. Cached reads are quicker than mine because you're using SATA-II (3gb) whereas I'm using SATA-I (1.5 GB/s). The speed of the medium itself is where I pretty much expect it to be for a drive that size.

800 megs in half a minute equates to about 26 MB/s in writing speed. That's not bad. Since you did this across a network, there might be a little deviance somewhere there, so just to make sure, run Bonnie++. It'll do a variety of local reading and, more importantly, writing tests on your actual filesystem, so you're not only testing the disk itself, but any settings of the filesystem that you may have toyed with aswell.

Link to comment
Share on other sites

Thanks Cooper for all the help you've given me so far...

I have gone to install ubuntu server since debian messes up my hard disk... too bad :cry:

The ubuntu server version seemed a bit weird... the first thing I had to do was to install xinetd while inetd.conf was present on the system... and man... what the hec of a job to install that laser printer... looks like ubuntu server is a bit messed up.... have you ever tried it?

I dunno how i feel about running ubuntu as server... would they be an alternative since Debian cann't be installed (old kernel.. not good drivers)..

I though of Suse since the german government uses it...

what do you think? any recommendations....

thank you so much!

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

 Share

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...