PDA

View Full Version : Building a quiet Mini-ITX Server with ClarkConnect



mherger
2007-06-27, 05:19
Where has the original thread gone? I don't know. Wanted to remove my own posting. The board gave me unknown power - and upon click on "Remove Post" the whole thread was gone :-(.

I'm really sorry for this as I know this was one of the (if not THE) most popular threads on this board. I don't know whether there's a kind soul in Mountain View able to restore it.

Meanwhile I've restored its content from Google and my browser's cache:

http://www.herger.net/slim/BuildingaquietMini-ITXServerwithClarkConnect.htm

vrobin
2007-06-27, 05:47
You should get the last version from google cache before it's gone and post it again there...

mherger
2007-06-27, 06:02
> You should get the last version from google cache before it's gone and
> post it again there...

I added the last four or five posts from my browser's cache. It should be
complete.

--

Michael

-----------------------------------------------------------------
http://www.herger.net/SlimCD - your SlimServer on a CD
http://www.herger.net/slim - AlbumReview, Biography, MusicInfoSCR

vrobin
2007-06-27, 06:23
The whole thread can be read there, but it won't last forever, you should copy paste it there:

http://209.85.129.104/search?q=cache:ijb-CNLxx4cJ:forums.slimdevices.com/archive/index.php/t-31072.html+Building+a+quiet+Mini-ITX+Server+with+ClarkConnect&hl=en&ct=clnk&cd=2&client=firefox-a

minilinked there:
http://minilien.com/?behqUQ4MBH

mherger
2007-06-27, 06:35
> The whole thread can be read there, but it won't last forever, you
> should copy paste it there:

It's missing the latest four postings of the past few days. As I mentioned
you'll find the _full_ thread on the very first link I posted (it's based
on yours).

--

Michael

-----------------------------------------------------------------
http://www.herger.net/SlimCD - your SlimServer on a CD
http://www.herger.net/slim - AlbumReview, Biography, MusicInfoSCR

vrobin
2007-06-27, 07:32
It looks like I've read your message too fast... sorry for this ;)

Robin

mick_w
2007-07-13, 05:58
Michael - thanks for rescuing the old thread - I've now added a copy to the links page on the website...

I thought it might be worth adding this post with a fresh link to the site:


***** http://www.ulverston.myzen.co.uk/mini-itx/index.htm *****


I've also added a new link to a Dutch translation that Ramon Boerman has added on his website.

Mick

mherger
2007-07-15, 11:17
> Michael - thanks for rescuing the old thread

....that's the minimum I could do after accidentally deleting it :-/.

Michael

Fifer
2007-10-23, 05:37
A heartfelt thanks to mick for his excellent instructions (and to others who helped me along the way with some Linux dummy questions). My mini-ITX server is now up and running and the performance running SC7 is far superior to my TS-101 running 6.5.1 (in fact, it's much quicker than the TS-101 running 6.3.x).

After some initial problems getting CC to install on my eBay purchased EPIA M10000 Nehemiah (overcome by transferring the HD to another machine and doing the install there), I found mick's instructions to be very clear, simple, comprehensive and accurate. The whole thing cost less than £150 (case, board and memory under £100 from eBay and a new Maxtor 250GB disk at £45) and despite my initial reservations, SC7 seems to run just fine in 256MB of RAM. A complete rescan of 4500 tracks took about 18 minutes which is slower than mick's setup, but still respectable. Everyday operation with the remote, the web interface and with Jive all seem perfectly responsive and normal. Scanning might be quicker with more memory, but a complete rescan from scratch is a reasonably infrequent operation so it's probably not necessary.

I opted to use one disc for the OS and the music (my Cubit case only takes one HD), but the HD seems only to be accessed during playback or scanning and not at all when idle. As it happens, the HD is quieter than the processor fan (which is very quiet and effectively inaudible when the box is located under my desk). Noise wasn't a major issue for me in any case as the server is in my study on the first floor and I listen to music downstairs. However, it is very quiet which is a bonus.

So, for under £150, I have a high performance, low power, small form-factor SlimServer box with power enough to cope with future SC (or SC) developments for the next wee while. I'm very happy.

Now to install AlienBBC and get print serving working ...

MrStuff
2007-12-25, 10:03
Following inspiration from Mick I have finally built myself a dedicated file/slimserver box based around Via hardware and Clarkconnect. However it just doesn't appear to be stable or
responsive enough and I need to understand if I've made a
fundamental mistake before struggling to resolve the issues I see.

Can someone take a look at this and let me know if what I am
trying to do is sensible or not?

Here goes some information:

Hardware:
Via EN12000EG board
1GB Ultra low profile DDR2 (533MHz)
(Ran memtest86+ for 6 hours with no errors)
Old Fujitsu MHT2040AT 40GB 2.5" HDD (system)
(Possible repalce with SSD if prices fall next year)
New Western Digital WD10EACS 1GB HDD (main file storage)
Pioneer DVR-K17 DVD Drive
Travla C158 (60W) case (fan disconnected)
Temperatures well monitored - never gets very high
(CPU ~45C, M/B ~66C, HDD's ~42C)

Software:
ClarkConnect 4.2
Slimserver 6.5.1 (as I've been using on my FC5 box)
Mplayer/AlienBBC (installtion/versions as suggested by Mick)
GkrellM-Daemon 2.2.4 for remote status monitoring
lm_sensors 2.8.7
Various tweaks to webconfig for logging system/harddrive temps

Clients:
1*SB1
2*SB2
Linux (FC5 box) (Samba, SSH & Gkrellm)
Win XP Pro Box (Samba & Softsqueeze)
Vista Ultimate laptop (Samba & SoftSqueeze)

Goal:
Virtually silent
Centralised file storage (currently on FC 5 box)
Slimserver to support 3 simultaneous streams (Flac/AlienBBC)
Central mail server with anti-spam/virus
Imap links to clients on Windows/Linux boxes

Problem(s):

*Everything* works in isolation (inc AlienBBC and MPlayer).

Under heavy Samba load (from WinXP) Slimserver is
unresponsive and music streams break-up/stall
Raising priority of Slimserver in preferences doesn't
seem to work (process doesn't change it's nice value).
Manual re-nicing (-4) of SS does improve responsiveness,
but quickly leads to kernel panic and a need to reboot
machine/fsck storage HDD.

Samba access from FC5 box causes virtually instantaneous
Kernel panic (version mis-match problems?)

Haven't tried access from Vista laptop yet.

Machine has been observed to panic at other times as
well, notably when trying to download cc-devel and when
strange characters are input through keyboard (connected
through KVM which sometimes sends "strange" characters
when switching PC's under control).

signor_rossi
2007-12-26, 06:45
@MrStuff
Doesn't a kernel panic most likely suggest driver problems?
Have you tried another distro already or did you try to get support from ClarkConnect (maybe newer kernel already helps)?
Then 66°C seems a bit high for M/B temperature. On my old P4 M/B i usually have only halve your temperature now.

Bye, signor_rossi.

MrStuff
2007-12-27, 09:58
I haven't tried another distro on this hardware yet. Featurewise ClarkConnect seems to tick all the boxes that I need and I haven't found another distro that comes with almost everything I need in the box.

I would like some feedback from anyone else doing something similar on this class of hardware before I start trying to debug and fix the individual problems. I have a feeling this will take me a lot of time, and if I can't be certain to end up with something that will allow me to run Slimserver without the music breaking up then I'd rather give up now and cut my losses.

BTW - It's now run memtest for nearly 32 hours straight with no errors so it really doesn't look like a memory problem.

MrStuff
2007-12-28, 02:59
OK, so after a week of trying to work out why it wasn't working I finally noticed that the CPU was only running at 400MHz rather than 1200MHz.

Although there was no setting in the BIOS for the CPU multiplier I reset the BIOS to the defaults and the CPU speed was restored (along with who knows what else) and everything now seems to be working and stable.....

Oh, and the performance is much better now as well........

Still some other problems to sort out but nothing that's going to be a showstopper.

mick_w
2007-12-28, 04:03
Nice to see you cracked it...

I've got a similar set-up and everything works fine, (never seen a kernel panic) - but that probably doesn't really help you.

You should be able to stream to a SB while transferring files with Samba without any detrimental effects.

Mick

signor_rossi
2007-12-28, 12:25
Did you upgrade the BIOS right after you built the box? BIOS reset should also be done after that, as stated in the manuals. ;)

Bye, signor_rossi.

house
2007-12-31, 05:40
Any news if it works problems free?

Fifer
2007-12-31, 08:20
Mine has been for several months now if that's any help.

JohnH
2008-01-15, 12:56
I have also built a Mini-ITX using Mick's excellent instructions.
All is working fine.
Would like to use the mobo Wake On Lan function. Have enabled
PowerOn by PCI Card
ACPI function

But it does not work can anybody advise where I may be going wrong?

John

Espen
2008-01-19, 06:59
I have also built a Mini-ITX using Mick's excellent instructions.
All is working fine.
Would like to use the mobo Wake On Lan function. Have enabled
PowerOn by PCI Card ACPI function
But it does not work can anybody advise where I may be going wrong?


I can not help you yet. I have the same problem. I know wake on lan works because if I turn off the computer (plug out the power cord and connect it again), then the machine will be powered on later when I send a wol request to it from an SB3 box (and from a software emulator on another PC).

But, it does not work if I turn off the server from the OS itself (I have mapped the power button to initiate the shutdown procedure). After the machine has been shut down I can not power it up again with wol without the "trick" described above.

Seems the OS-shutdown sets the machine in some state where it does not respond to this signal.

Anyone with success?

Timothy Stockman
2008-01-19, 07:40
Western Digital WD10EACS 1GB HDD (main file storage)
Just picked up one of these "Green Power" drives for the linux box I'm making, and according to all reviews, this drive is quiet and consumes lower power than any other 1GB drive. This of course, translates into less heat produced. It does run at 5400 RPM, but since the bit density is very high to achieve 1GB, the actual bit rate to/from the disc is comparable to a 7200 RPM, 500 GB drive.

I should have the machine going mid-week next week. If everything I've read is true, this should currently be about the best drive for the music drive of a SqueezeCenter box.

Espen
2008-01-19, 09:26
I can not help you yet. I have the same problem. I know wake on lan works because if I turn off the computer (plug out the power cord and connect it again), then the machine will be powered on later when I send a wol request to it from an SB3 box (and from a software emulator on another PC).

But, it does not work if I turn off the server from the OS itself (I have mapped the power button to initiate the shutdown procedure). After the machine has been shut down I can not power it up again with wol without the "trick" described above.

Seems the OS-shutdown sets the machine in some state where it does not respond to this signal.

Anyone with success?

Well quoting myself now. After googling a little bit I solved this by modifying the /etc/init.d/halt script. At the very end I

1) removed the "-i" from the HALTARGS variable
2) Then called

/sbin/ifconfig eth0 192.168.1.10 up
/usr/sbin/ethtool -s eth0 wol g

before

exec $command $HALTARGS

It seems the hardcoded IP address does not really matter. That is not the address the server is using (I use DHCP so it may change) so I have not bothered to determine what address was assigned to the machine right before exit. I use the 192.168.1 segment though.


Together with the "Server Power Control" plug-in I can now remotly control power on/off from the remote.

JohnH
2008-01-19, 15:42
Espen - thanks for the help

Made changes as per post and all working now.

dan325it
2008-01-20, 11:22
Thanks a lot to Mick for the great website with plain English instructions.

I'm planning to get one of the mini-itx boards with on-board RAID1 support. I'll use an old laptop drive for OS and two of the WD green hard drives for music files and backup. Is there anything I should be aware of? Will I still be able to spin down the drives when idle?

Thanks!

flashgordon
2008-02-05, 13:20
Well quoting myself now. After googling a little bit I solved this by modifying the /etc/init.d/halt script. At the very end I

1) removed the "-i" from the HALTARGS variable
2) Then called

/sbin/ifconfig eth0 192.168.1.10 up
/usr/sbin/ethtool -s eth0 wol g

before

exec $command $HALTARGS

It seems the hardcoded IP address does not really matter. That is not the address the server is using (I use DHCP so it may change) so I have not bothered to determine what address was assigned to the machine right before exit. I use the 192.168.1 segment though.


Together with the "Server Power Control" plug-in I can now remotly control power on/off from the remote.

Good work Espen.

I could not get this to work, so if anyone else is having trouble implementing Espen's fix for WOL, try these instructions here:- http://forums.fedoraforum.org/archive/index.php/t-122902.html

They are very similar to Espen's, but I think the thing that helped getting mine to work was updating the rc.local file.

Krishna
2008-03-13, 11:39
I recently built a mini-ITX system (based on mick_w's website).

Here is what I used:
Case + Motherboard + Memory [$290]
Travla C158 Mini-ITX Case with 60/90W PS
http://www.logicsupply.com/products/c158b_60w ($107)
Via EPIA CN10000 Fanless C7 Mini-ITX
http://www.logicsupply.com/products/ln10000eg ($137)
1GB Ultra-low profile DDR2 522 ($46)
http://www.logicsupply.com/products/128mlq64v5vl

Hard Disk Drives [$134]
OS Drive: Seagate Momentus 5400.3 ST960813AS 2.5" 60GB 5400rpm 8MB SATA-150
http://www.geeks.com/details.asp?inv...13AS-R&cat=HDD ($34)
Data Drive: Samsung Spinpoint T-Series HD501LJ 3.5” 500GB 7200rpm 16MB SATA-300
http://www.newegg.com/Product/Produc...82E16822152052 ($105)

Cables [$10]
2x SATA to Molex power supply adapter cable & SATA data cable ($5 each)

Total = $434

Everything works fine and the instructions were very clear.
Thanks :)

Mnyb
2008-03-15, 09:18
I managed to build one off those mini ITX machines. Thankyou all for the knowledge you have especially Mick and his website.

I've basically got it to do what i want, i got Squeeze and Samba working.

i added a little more security by using the built in firewall and hostbased protection ("hosts allow" & "hosts deny" lines in smb.conf)

I have the VIA EPIA EN12000EG and CC 4.2 SP1.

however
I have a couple of small problems to resolve and i think the are related?

"shutdown -h now" or shutdown from dashboard won't power of the computer. Have to use power button after shutdown.

The power button won't initiate shutdown at all ?
I installed acpi to no avail, any insights ? I think i have the BIOS settings correct, anybody with the same motherboard.

I tried to implement Espens little trick to get wol to work.
I had to try some things suggested by the link provided by flashgordon http://forums.fedoraforum.org/archiv.../t-122902.html i don't dare to take the last step edit the rc.local file, heres why, when I run the suggested commands from shell (just to try out) i get this result:
# echo -n PCI0 >> /proc/acpi/wakeup
-bash: /proc/acpi/wakeup: No such file or directory
theres not even an acpi dir there ?

I'm chicken out because of another problem :-)
the video drivers is not ok, when CC loads i get "out of range" after a while and no picture(I tried some thing's suggested by the cc forum, no result). This is no biggie as i intend to run it headless, but if I mess to much with the NIC, and add some not so good shutdown's, I find that i cant connect remotely and i'm not able to see whats going on via the monitor ?

I've seen that a good thing is to add acpi=force to the kernel line in the /boot/grub/menu.lst ? can anyone suggest an exact syntax so i don't mess with the boot (remember i have no picture just the ssh tool )

Cheers

Mikael

Mnyb
2008-03-15, 10:24
Power down is now working, I added acpi=force to the kernel line in /boot/grub/menu.lst


"kernel /vmlinuz-2.6.18-53.1.13.2.cc ro root=/dev/VolGroup00/LogVol00 video=vesafb vga=0x313 acpi=force"

/Mikael

edit/
This also solves half the wol problem as suddenly there is a /proc/acpi/wakeup directory after each boot
but still no wol ?

Mnyb
2008-03-16, 07:54
Wol solved, i post info in another tread, nothing to do with the server as acpi=force took care off all problems apparently, then I could apply the necessary changes to use wol, as suggested by the earlier posts.

Mnyb
2008-03-20, 11:02
I got the video working by adding this video=vga16fb vga=771 to the kernel line /boot/grub/menu.lst

kernel /vmlinuz-2.6.18-53.1.13.2.cc ro root=/dev/VolGroup00/LogVol00 video=vga16fb vga=771 acpi=force

Anyone else having an HP1940 LCD monitor can use this with his/her.
VIA EPIA-EN12000EG,EDEN V4/1200Mhz mobo

Other alternatives might be (i'm no expert just try) :

# VESA framebuffer console @ 1024x768x64k
# vga=791
# VESA framebuffer console @ 1024x768x32k
# vga=790
# VESA framebuffer console @ 1024x768x256
# vga=773
# VESA framebuffer console @ 800x600x64k
# vga=788
# VESA framebuffer console @ 800x600x32k
# vga=787
# VESA framebuffer console @ 800x600x256
# vga=771
# VESA framebuffer console @ 640x480x64k
# vga=785
# VESA framebuffer console @ 640x480x32k
# vga=784
# VESA framebuffer console @ 640x480x256
# vga=769
# vga=ask

I also bought the right kind of memory on the Mini-ITX site

The no working memory from Crucial.

1GB 1RX8 PC2-4200U-444-12-D0
1GB, DDR2 533, CL4

A working memory from Crucial.

1GB 2RX8 PC2-4200U-444-12-D0
1GB, DDR2 533, CL4

Note the 2RX8 :)

The wol issue was this http://forums.slimdevices.com/showthread.php?t=44843

I also made the PowerCtrl plug in work's but the instructions are outdated and not necessarily accurate with SC or Clarkconnect.

Instructions with my comments after #

Install the contents to /usr/local/slimserver/Plugins/PowerCtrl #Try squeezecenter :)

Manually delete the /usr/local/slimserver/Cache/plugin-data.yaml file.
# No such file in my SC7.01

As root, add the slimserver user to the /etc/shutdown.allow file:
touch /etc/shutdown
echo “slimserver” >>/etc/shutdown
#No such file in CC4.2

edit the /etc/sudoers file:
Comment out the line with: Defaults requiretty
#I don't have this line in CC4.2

Add:

%slimserver ALL=NOPASSWD:/sbin/shutdown -h now
%slimserver ALL=NOPASSWD:/sbin/shutdown -r now

# try squeezecenter instead:)

restart slimserver and voilà!"

This migth be obvius for some, but as i asked this questions and found the answers myself. It would be nice to post them

Andy D
2008-03-28, 02:41
Hi Mick,

Thanks for writing the tutorial - I moved house back last year and have been wondering what to do about a new server ever since as the PC now lives in another room, but I think you've just helped me make up my mind!

I'm a bit concerned about SSD wear rates and performance and I've got a few questions on things you don't mention:

1. What filesystem did you use on the SSD - ext2 or ext3?
2. Did you add the noatime option to the fstab entry for the SSD?
3. Do you know which drive the MySQL DBs are stored on?

EDIT : just found the answers to 1 and 2 on the Tweaking ClarkConnect page!

Cheers

Andy

Mnyb
2008-03-28, 11:35
Just a little suggestion:

this is suggested by Mick, for mounting the harddrive:

/dev/sda1 /mnt/HardDrive ext2 defaults 0 0

I would have used at least:

/dev/sda1 /mnt/HardDrive ext2 defaults 1 1

which gives you regular disk checks after a X boots.
I configured the amount to 50 boots with tune2fs -c 50 /dev/sda1.
The disk check is slow and boring but it's not that often, be aware that the check is performed during boot.
As my objective was a little different I chosed the ext3 partition:

/dev/sda1 /mnt/music1 ext3 defaults,noatime 1 1

Then thanks again for the idea, CC4.2 is rock solid

Andy D
2008-03-29, 10:16
Which drive does the swap space live on the SSD or the hard drive?

Thanks

Andy

Mnyb
2008-03-30, 08:36
I installed CC on the SSD before adding the HDD for my music, so basically everything runs from the SSD in my case.

It works without the harddrive, so i don't see why there would be a swap space there. Of course it is probably possible to make it so if you now how.

If I look at my mount point /mnt/music1 in my case theres nothing there but music and playlists

the rest of the file system is on the SSD

Actually i don't now how CC swaps, my ubuntu machine has a swap partition.
CC has LVM (logical volume mapping) i've learned this 5 minutes ago.

Edit: the swap file on my CC is /dev/VolGroup00/LogVol01 swap swap defaults 0 0
I haven't figured how to see which disc a certain logical volume is on ?
But i got so far that i can conclude that my Harddrive is not allocatable for VolGroup00 (vgdisplay test -v)
So i'm 90% sure that my swap is on the SSD drive which is hda in my CC install

So your probably better off asking the CC forum for this.

"fdisk -l" and "lvs -a" commands give some info

Sorry not being able to answer this

Espen
2008-03-31, 00:34
Hi Mick,

Thanks for writing the tutorial - I moved house back last year and have been wondering what to do about a new server ever since as the PC now lives in another room, but I think you've just helped me make up my mind!

I'm a bit concerned about SSD wear rates and performance and I've got a few questions on things you don't mention:

1. What filesystem did you use on the SSD - ext2 or ext3?
2. Did you add the noatime option to the fstab entry for the SSD?
3. Do you know which drive the MySQL DBs are stored on?

EDIT : just found the answers to 1 and 2 on the Tweaking ClarkConnect page!

Cheers

Andy


I have mounted /var/log and /var/run as ramdisks. I still have swap on the
SSD, but it has never been used (I have in 1 GB RAM) so I plan to disable
it. I have the SQL database on SSD and that is perhaps a candidate for
getting over to the harddrive.

After I configuered the system I have realized that it is not fiddling
with the OS that takes time but rather proper tagging of music. So the most
important thing one do is proper backup. But I plan to have an extra SSD
as backup as well. I need less than 2GB so it is not too expensive.

Andy D
2008-04-04, 03:10
Thanks for the answers...

I don't really want to "fork out" for an SSD at the moment so was debating whether to install CC on the hard drive or try fitting in onto a CF card.

I've still not completely made up my mind to go with CC but as I want to add TwonkyVision (or such like) for videos it seems the most likely choice.

Andy

mundge
2008-04-04, 09:08
Hi guys,

FIrst foray into Linux (properly at least) and was going great guns. Very well written tutorial, just what a complete noob like me needs! For some reason though, I can't weem to log into the Web Interface from my Windows PC (running Vista).

Have tried various combinations of https://192.168.1.67:81/admin which should let me access the server should it not? Have also replaced the numbers with mediaserver.home.lan:81, but still with no joy.

I can ping the server from my PC and it shows up on the Router devices screen with the above IP.

Internet Explorer initally says it won't load because of a certificate issue, but I just click contimue, then username (I assume "admin"?) and then my password, but then just end up with blank screen, red address bar, but no errors or anything, just no access.

Any help would be much appreciated.

bert1e
2008-04-04, 12:15
You need to use the username root

mundge
2008-04-04, 13:02
You need to use the username root

Should I have known that or should I hold my head in shame?! Will try when my RAID rebuilds... new HDDSs, typical, think one is bust already!

Mnyb
2008-04-05, 00:03
You can create other users to, that can have access with or without limitations on what they can do.

Another tip on mini-itx boards from VIA and CC4.2 .

check that your RAM is working properly before you do anything with your system.
During installation of CC if you don't do anything i think the swap size will be similar to your RAM size, or in reality what your Motherboard is reporting as your ram size ! I didn't think so I just doubled the number presented to me, not realizing that i had the well known memory incompatibility problem with VIA boards.
approx half of the ram is not showing up. So the resulting swap space was smaller than i would like it to be (I changed to a compatible ram stick now).

As in many linux ? (all ?) system swap is disk partition, or more precisely a logical volume in CC4.2

To change swap size would meant that you have the resize the logical volumes on your main drive and resize the file systems on those logical volumes.

As I haven't found a bomb proof howto on how to do this safely, i let it rest for now as my server is not swapping anyway with 1GB of ram.
I had also put down to much work to get everything else running, so i would not dare to crash it now.

Thanks to Andy D for putting me on the track (I was blissfully ignorant before he posted questions regarding swap space ) :-/

mundge
2008-04-05, 18:02
You need to use the username root

All working really well, can't believe how simple it all seems, helped to a massive degree by the excellent walkthrough provided above.

It seems to be just what I wanted when I started. Will now see if I can get it set up so I can access it from outside the network. Fingers crossed... and thanks!

mundge
2008-04-07, 12:16
Been up and running for a couple of days now from a RAID 1 setup. Was thinking about putting the OS onto a CF card on the IDE - can I just transfer everything but my music etc (that's all I want to keep safe after all) from the RAIDed drives onto the CF card and change the boot order or do I have to reinstall the whole lot? Thought this might make things quieter until I actually have to stream stuff.

It's truly amazing some of the lengths people go to to help us novice users - much appreciated and would be lost without this advice.

James

ik632
2008-04-07, 13:40
I have always had the best luck when doing a fresh install when changing the primary OS media. I have used the IDE-CF adapter and they work pretty well. If your board supports it you can get a USB flash drive and boot from that. I think the 4GB ones are running about $30 these days. That may be a cheaper route if you don't already have the CF card and adapter.

mundge
2008-04-07, 17:04
Am a big fan of digital photography (hence the need for a server to backup - have about 130GB of photos!) and have loads of compact flash cards and bought a CF-IDE adaptor for £4/$8.

Just sent through all my photos and such and bit reluctant to run the risk of having to reinstall and possibly wipe the drives unless I really have to.

I guess if I delete something important on the RAID/OS drives then hard reset and reinstall it shouldn't banjax my data. Might try that tomorrow, worse that happens is I have to recopy everything!

ik632
2008-04-09, 10:16
I don't think the RAID will be affected at all. When I upgrade my NAS (runs FreeNAS) I wipe the install drive and then install fresh leaving the RAID alone until I get to mounting it again.

Andy D
2008-04-17, 10:29
Tried installing on a 2GB CF card and fails to format the card for some reason, so now I've bought a 2GB flash disk module and everything appears OK.

Now just got to work out what I leave on flash and what needs moving to the hard drive.

Espen
2008-04-18, 07:39
Tried installing on a 2GB CF card and fails to format
Now just got to work out what I leave on flash and what needs moving to the hard drive.

I only have music files on the harddrive. One could perhaps make symlinks for
the SQL database and scaled artwork files to folders on the harddrive. I have
not tried it yet. The real value on this system is properly tagged music and
artwork though. It takes time and effort to get it right. So I am less concerned
about the solid state drive and its well-being than the harddrive (make backups).

haver
2008-04-25, 04:40
I followed micks' instruction for building my own music server but trying to install CC 4.2 on a mini-ITX via CN10000EG board using a CompactFlash with a CF to IDE adapter as a boot device. I was able to get through the installation process without a hitch. But when I reboot, it takes forever to do it, and then the computer seems to be freezing at « Creating block devices ». It hangs there for what appears to be an eternity. After more than half an hour I decided to shut down the computer.

Any Idea what is going on here?

(In the meantime, I installed everything on a SATA HDD and it works like a charm!)

mundge
2008-04-25, 06:10
I don't think the RAID will be affected at all. When I upgrade my NAS (runs FreeNAS) I wipe the install drive and then install fresh leaving the RAID alone until I get to mounting it again.

Do you have the OS on a separate disk to the RAID drives or on the same one?

cannydanny
2008-05-29, 14:58
Hi everyone, I've been interested in streaming my music for quite sometime now and i'll shortly be buying a logitech duet to do this. In the meantime i've been reading quite alot about NAS devices,and came across this brilliant how to build a mini atx server running clark connect, i must confess i know absolutely nothing about linux os, but i'm willing to learn at my own pace, which i'm sure will be very slow. Anyway after following the instruction as best as possible i still can't log into clark connect from my remote laptop. I've been trying everything but still no luck so can anyone please give me some much needed and appreciated help.
Here's my info: Clark connect up and running, i've selected DHCP on my router and assigned my routers lan ip address to the nameserver, i've used media-box and home.lan as the server and domain names respectively i've also noted that the lan ip address of 192.168.1.68 has been assigned to it by the router. I can see there is something happening on the clark connect system but as as yet don't understand it, anyway when i try to log into it from mylaptop using http://192.168.1.68:81/admin or http://media-box:81/admin nothing happens.
FYI here's my settings, believe me i tried them in many configurations without success
Router ip 192.168.1.254
subnet mask 255.255.248.0
any advice greatly appreciated,
cannydanny

mundge
2008-05-29, 17:09
Danny - you sound like me mate, know next to nothing about Limux, but I do know that you need to type https not http at the start of the address. I get into mine by typing https://192.168.1.70:81/admin, username root, password whatever.

Hope this helps.

cannydanny
2008-05-30, 03:04
Thx mundge for the quick reply, yes i did notice that on the how to guide and i have tried it but alas still no connection, clark connect was vey simple to install, just can't get into in from my laptop, i guess its something to do with my ip addresses and other settings, still can't understand it though, does anyone have any ideas.
thankyou cannydanny

Espen
2008-05-30, 03:16
Have you from from a local terminal on the clarkconnect box tested with
/sbin/ifconfig ? It will print out the IP address in use (just in
case it is using something else than what you think). Can you ping
other machines from the box (and vice versa)?

cannydanny
2008-05-30, 05:09
Thx Espen, i'm at work now so will check later, I've been reading some instructions on the clarkconnect website on the instructions it said "If your ISP assigns you a static IP, you will need to get these settings from your ISP" I have a static ip address so i guess i need to enter all my settings in manually, is this right?
Is it correct that i can't use DHCP if i have a static ip address?
Perhaps you or another kind member would be good enough to confirm i'm using the correct settings that clarkconnect asks for.
Ip address 87.XXX.XXX.212 (my static ip address)
Netmask 255.255.248.0
Default gateway ip 87.194.xxx.x
primary nameserver NS1.betherenow.co.uk
I'll use media.box as the hostname and home.lan as the domain name.
Does this look right, do i need to enter my routers local ip address anywhere.
Thankyou for your help.
Cannydanny

joncooper
2008-05-30, 05:37
You have a broadband router, so the IP address of your clark connect server will be whatever IP the router assigns via DHCP.

madpatrick
2008-05-30, 06:23
Hi everyone, I've been interested in streaming my music for quite sometime now and i'll shortly be buying a logitech duet to do this. In the meantime i've been reading quite alot about NAS devices,and came across this brilliant how to build a mini atx server running clark connect, i must confess i know absolutely nothing about linux os, but i'm willing to learn at my own pace, which i'm sure will be very slow. Anyway after following the instruction as best as possible i still can't log into clark connect from my remote laptop. I've been trying everything but still no luck so can anyone please give me some much needed and appreciated help.
Here's my info: Clark connect up and running, i've selected DHCP on my router and assigned my routers lan ip address to the nameserver, i've used media-box and home.lan as the server and domain names respectively i've also noted that the lan ip address of 192.168.1.68 has been assigned to it by the router. I can see there is something happening on the clark connect system but as as yet don't understand it, anyway when i try to log into it from mylaptop using http://192.168.1.68:81/admin or http://media-box:81/admin nothing happens.
FYI here's my settings, believe me i tried them in many configurations without success
Router ip 192.168.1.254
subnet mask 255.255.248.0
any advice greatly appreciated,
cannydanny

Danny,

I use CC now for more than 5 years with a lot of pleasure.
You can go to the menu by HTTPS://MEDIA-BOX:81 of HTTP://192.168.1.68:81
(i believe these are you domain name or IP adress of the server.
You can you IP by typing IFCONFIG on the server prompt.

bert1e
2008-05-30, 06:40
I would start by trying to ping the clarkconnect box from your workstation. If that doesnt work it might be a setting in your router preventing you from talking to other hosts within your network.

Also log in on the clarkconnect box as root and try pinging the workstation

chill
2008-05-30, 06:43
Danny

The static IP address that your ISP gives you simply refers to the IP address that your router has when accessed from outside your network, and it isn't relevant to your problem I think. Inside your network, all your IP adresses can be set to virtually whatever you like - there is a router between your network and the outside world, and therefore it doesn't matter if your local IP addresses are the same as everyone else's - they are never seen from outside your network, so there is never a clash.

Many routers will use a standard private IP address of 192.168.0.1, and with a subnet mask of 255.255.255.0 this means that all of your other devices on your network will have IP addresses of the form 192.168.0.n. If your router is using DHCP, then it will choose a value of n (up to 255) for each device that connects. If it isn't, then it's up to you to choose a fixed IP address within this range for each bit of equipment on your network.

You need to check the IP address that your router is using, and find out if it is using DHCP to assign IP addresses locally. All this information is usually available via the router's web interface. There is also usually a page that tells you the IP addresses of all the equipment on your network - referenced by the MAC addresses.

This should enable you to find out either the IP address that has been assigned to your ClarkConnect box automatically, OR the range from which you have to manually assign one.

Chris

Edit: Ahem - looks like I should have read more of the thread before posting - D'oh!

cannydanny
2008-05-30, 11:46
Thankyou all for your quick replies, I'm in, I came home from work reinstalled cc and was able to log on from my remote laptop first time, now onward to the next step.
thankyou all once again.
cannydanny

pichonCalavera
2008-06-04, 14:54
Looking for a solution for the main hard drive, I stumbled upon this new SSD Drive from Intel: http://www.intel.com/design/flash/nand/z-p230/index.htm , more info here too: http://www.linuxdevices.com/news/NS7426360998.html .

Being a SSD design, with 8GB and for 45dlls , I think these would be perfect for a quiet and cheap Hard Drive don't you think? , my only confusion is how would you put this inside a mini-itx case...

unregistered
2008-06-04, 16:50
has anybody gotten this working on clarkconnect to a windows share? My attempts seem to fail miserably...

my /etc/auto.nas
music -fstype=smbfs,guest,rw ://192.168.0.22/music

the mount point shows up in the squeezecenter settings (/mnt/NAS/music), but no music is found when doing a search.

Mounting using smbfs works fine, but I would prefer an automount.

George

Ramage
2008-06-05, 00:34
has anybody gotten this working on clarkconnect to a windows share? My attempts seem to fail miserably...

my /etc/auto.nas
music -fstype=smbfs,guest,rw ://192.168.0.22/music

the mount point shows up in the squeezecenter settings (/mnt/NAS/music), but no music is found when doing a search.

Mounting using smbfs works fine, but I would prefer an automount.

George

I had the same problem, but fixed by using cifs in place of smbfs:
Try:
music -fstype=cifs,guest,rw ://192.168.0.22/music

Don't ask me why, there are some subtle differences between smbfs and cifs, although AFAIK cifs is supposed to replace smbfs.

unregistered
2008-06-05, 13:55
very cool. that did it!

gunnar
2008-06-11, 15:00
Hi! Thanks for these very nice instructions!
I have had my CC 4.2 system running Squeezecenter for a while. I also have Mediatomb running, which I use for streaming media to my PS3.

I installed autofs according to the instructions (I think..), and I get the USB mount point at /mnt/USB. But the folder seems to be empty. I`m trying to mount a ntfs WD-Elements 500GB USB-drive. blkid shows that the USB-drive is /dev/sda1, which I have used in the auto.usb file.

Any suggestions?

Gunnar

Ramage
2008-06-12, 00:12
Hi! Thanks for these very nice instructions!
I have had my CC 4.2 system running Squeezecenter for a while. I also have Mediatomb running, which I use for streaming media to my PS3.

I installed autofs according to the instructions (I think..), and I get the USB mount point at /mnt/USB. But the folder seems to be empty. I`m trying to mount a ntfs WD-Elements 500GB USB-drive. blkid shows that the USB-drive is /dev/sda1, which I have used in the auto.usb file.

Any suggestions?

Gunnar

Can you mount and read/write from your ntfs usb drive using the mount command from a terminal? If you can post your command line here. I think your problem could be the ntfs file system.

I auto mount usb drives on my CC box with no issues but they are FAT, I have not tried ntfs.

gunnar
2008-06-12, 13:10
Can you mount and read/write from your ntfs usb drive using the mount command from a terminal? If you can post your command line here. I think your problem could be the ntfs file system.

I auto mount usb drives on my CC box with no issues but they are FAT, I have not tried ntfs.

I think you are right. I tried to mount manually, and this is what happened:

[root@musikk USB]# mount /dev/sda1 /mnt/USB
mount: fs type ntfs not supported by kernel

But ntfs is listed as supported on the mount man page. I can not use FAT, as I have some video files that are larger than 4GB.

Gunnar

Ramage
2008-06-12, 23:07
I think you are right. I tried to mount manually, and this is what happened:

[root@musikk USB]# mount /dev/sda1 /mnt/USB
mount: fs type ntfs not supported by kernel

But ntfs is listed as supported on the mount man page. I can not use FAT, as I have some video files that are larger than 4GB.

Gunnar

The capability is out there I suggest you start here: http://www.linux-ntfs.org/doku.php. It's getting better all the time.

As I don't use it, I can't give you any direct advice on using ntfs with CC.

Good luck and post your results if you're successful.

gunnar
2008-06-25, 00:41
I have now installed the driver, and I`m able to mount the USB ntfs-drive. But so far I get stuttering when streaming. It plays smoothly for about 1 sec, then it hangs for about 1 sec and so on. I have only tried H264 and MPEG2 videofiles so far.
See some more details here: http://www.clarkconnect.com/forums/showflat.php?Cat=0&Number=101829&page=0&fpart=1&vc=1

Gunnar

rlangset
2008-06-25, 06:39
Have you tried installing and mounting like this (from the ntfs-3g page):


Installation

Linux: Make sure you have the basic development tools and the kernel includes the FUSE kernel module. Then type:

./configure
make
make install # or 'sudo make install' if you aren't root

Please note that NTFS-3G doesn't require the FUSE user space package anymore.

Non-Linux: Please see the OS specific installation and source packages above.


Usage

If there was no error during installation then the NTFS volume can be mounted in read-write mode for everybody as follows. Unmount the volume if it had already been mounted, replace /dev/sda1 and /mnt/windows, if needed.

mount -t ntfs-3g /dev/sda1 /mnt/windows

Please see the ntfs-3g manual page for more options and examples.

You can also make NTFS to be mounted during boot by adding the following line to the end of the /etc/fstab file:

/dev/sda1 /mnt/windows ntfs-3g defaults 0 0

Of course you would replace windows with USB in your case.

Rune

gunnar
2008-06-26, 15:01
Hi Rune!
no, I used the directions in the above link since CC requires a modified RPM. I can both read and write to the drive, so the only problem so far is the stuttering. I have installed the automount package, but so far I have only mounted manually.

Gunnar

japsa1974
2008-08-05, 02:00
OK, so after a week of trying to work out why it wasn't working I finally noticed that the CPU was only running at 400MHz rather than 1200MHz.

Although there was no setting in the BIOS for the CPU multiplier I reset the BIOS to the defaults and the CPU speed was restored (along with who knows what else) and everything now seems to be working and stable.....

Oh, and the performance is much better now as well........

Still some other problems to sort out but nothing that's going to be a showstopper.

Hi I am having the same troubles, Can you please explain how you reset the bios to defaults ?

Cheers
Japsa

gharris999
2008-11-07, 21:47
Is anyone able to WOL a Via EPIA CN board after suspending to S3? I can WOL just fine after a system halt. But WOL doesn't seem to work when in pm-suspend or pm-hibernate.

I'm running Fedora 9 on this system.

Any ideas?

vrobin
2008-11-08, 06:07
I had tracked down problem in the default kernel driver. After tweaking around, I finally managed to make successful WOL one time every two wake up (one works with WOL, the seconds has to be made by hitting power button).

I had better luck with VIA linux ethernet driver but still tried to debug the default one... since then, more than a year ago, I decided that letting the computer up 24/7 wasn't that bad and remains with my ethernet driver half debugged :).

syburgh
2008-11-08, 07:13
pm-utils does not do the same things as the halt init script, which could be the source of your problem. Halt shuts down the network interfaces (and in some cases, also removes the network driver modules from the kernel) but pm-suspend/pm-hibernate does not.

What module/driver supports the network interface? What happens when you remove it before suspend/hibernate and re-insert during resume? Create a file /etc/pm/config.d/modules


# Replace e1000 with the name of your adapter's driver module. For 100Mb adapters, you may need to remove the mii module as well (add it to SUSPEND_MODULES before the adapter module. It will be auto-laoded on resume, so no need to add it)
SUSPEND_MODULES="e1000"
RESUME_MODULES="e1000"


I've needed to do this to make a number of platforms work with WOL from S3/S4, but haven't tried anything Via related. You can read more at http://wiki.syburgh.com/audio0-local

gharris999
2008-11-08, 13:30
syburgh: I spent most of yesterday working through the permutations of your excellent wiki, but to no avail. This is what I've tried so far:

1). Created /etc/pm/config.d/modules ..containing:


SUSPEND_MODULES="uhci_hcd mii via_rhine"
RESUME_MODULES="via_rhine uhci_hcd"

2). Created /etc/pm/sleep.d/15network ..made executable and containing:


#!/bin/bash

# Reconfigure network interfaces after resume to ensure ethtool settings are
# applied (e.g. WOL, speed, duplex, etc.)
#
# Note that this file belongs in /etc/pm/hooks/15network and must be executable

. /usr/lib/pm-utils/functions

restart_network() {
/sbin/service network restart
}

enable_wol() {
/sbin/ethtool -s eth0 wol g
}

case "$1" in
hibernate|suspend)
enable_wol
;;
thaw|resume)
restart_network
;;
*)
;;
esac

exit $?

Removed /usr/lib/pm-utils/sleep.d/10NetworkManager (I'm running this server headless with static config info for eth0 in /etc/sysconfig/networking/devices/ifcfg-eth0 so I'm not using NetworkManager:


# Please read /usr/share/doc/initscripts-*/sysconfig.txt
# for the documentation of these parameters.
TYPE=Ethernet
DNS1=208.67.222.222
DEVICE=eth0
HWADDR=00:40:63:ed:3b:66
ETHTOOL_OPTS="wol g"
BOOTPROTO=none
DNS2=208.67.220.220
ONBOOT=yes
USERCTL=no
PEERDNS=yes
IPV6INIT=no
DNS3=192.168.0.1
DHCP_HOSTNAME=squeezenas
NM_CONTROLLED=no
IPADDR=192.168.0.222
NETMASK=255.255.255.0
GATEWAY=192.168.0.1

When I run pm-suspend and then manually wake up the machine with the power button (WOL fails) this is the resulting pm-suspend.log:

Initial commandline parameters:
Adding parameters
Fri Nov 7 23:39:06 MST 2008: Running hooks for suspend.
/usr/lib/pm-utils/sleep.d/00clear suspend: success.
/usr/lib/pm-utils/sleep.d/01grub suspend: not applicable.
/usr/lib/pm-utils/sleep.d/05led suspend: not applicable.
/etc/pm/sleep.d/15network suspend: success.
/usr/lib/pm-utils/sleep.d/49bluetooth suspend: not applicable.
/usr/lib/pm-utils/sleep.d/50modules suspend: success.
/usr/lib/pm-utils/sleep.d/55battery suspend: success.
/usr/lib/pm-utils/sleep.d/65alsa suspend: not applicable.
/usr/lib/pm-utils/sleep.d/90clock suspend: Shutting down ntpd: [ OK ]
success.
/usr/lib/pm-utils/sleep.d/94cpufreq suspend: success.
/usr/lib/pm-utils/sleep.d/95led suspend: not applicable.
/usr/lib/pm-utils/sleep.d/95packagekit suspend: success.
/usr/lib/pm-utils/sleep.d/99hd-apm-restore.hook suspend: saving level 255 for device sda
Advanced Power Management not supported by device sdb.
success.
/usr/lib/pm-utils/sleep.d/99video suspend: success.
Fri Nov 7 23:39:08 MST 2008: performing suspend
Fri Nov 7 23:39:54 MST 2008: Awake.
Fri Nov 7 23:39:54 MST 2008: Running hooks for resume
/usr/lib/pm-utils/sleep.d/99video resume: success.
/usr/lib/pm-utils/sleep.d/99hd-apm-restore.hook resume: restoring level 255 for device sda

/dev/sda:
setting Advanced Power Management level to disabled
success.
/usr/lib/pm-utils/sleep.d/95packagekit resume: method return sender=:1.16 -> dest=:1.15 reply_serial=2
success.
/usr/lib/pm-utils/sleep.d/95led resume: not applicable.
/usr/lib/pm-utils/sleep.d/94cpufreq resume: success.
/usr/lib/pm-utils/sleep.d/90clock resume: success.
/usr/lib/pm-utils/sleep.d/65alsa resume: not applicable.
/usr/lib/pm-utils/sleep.d/55battery resume: success.
/usr/lib/pm-utils/sleep.d/50modules resume: success.
/usr/lib/pm-utils/sleep.d/49bluetooth resume: not applicable.
/etc/pm/sleep.d/15network resume: Shutting down interface eth0: [ OK ]
Shutting down loopback interface: [ OK ]
Bringing up loopback interface: [ OK ]
Bringing up interface eth0: RTNETLINK answers: File exists
Error adding address 192.168.0.222 for eth0.
[ OK ]
success.
/usr/lib/pm-utils/sleep.d/05led resume: not applicable.
/usr/lib/pm-utils/sleep.d/01grub resume: not applicable.
/usr/lib/pm-utils/sleep.d/00clear resume: success.
Fri Nov 7 23:40:06 MST 2008: Finished.

At this point, my hair is lying in tufts on the floor and my cranium is sporting new bald spots.

Any help you folks can give me in tracking this down would be greatly appreciated. Thanks.

gharris999
2008-11-08, 13:48
The other bits of info I've gathered include:

..from lsmod, looking for the ethernet driver:


...
via_rhine 23304 0
mii 8192 1 via_rhine
...

..info on via_rhine:


filename: /lib/modules/2.6.26.3-29.fc9.i686/kernel/drivers/net/via-rhine.ko
license: GPL
description: VIA Rhine PCI Fast Ethernet driver
author: Donald Becker <becker@scyld.com>
srcversion: 2C8DCA4CAD6478ED0878B5C
alias: pci:v00001106d00003053sv*sd*bc*sc*i*
alias: pci:v00001106d00003106sv*sd*bc*sc*i*
alias: pci:v00001106d00003065sv*sd*bc*sc*i*
alias: pci:v00001106d00003043sv*sd*bc*sc*i*
depends: mii
vermagic: 2.6.26.3-29.fc9.i686 SMP mod_unload 686 4KSTACKS
parm: max_interrupt_work:VIA Rhine maximum events handled per interrupt (int)
parm: debug:VIA Rhine debug level (0-7) (int)
parm: rx_copybreak:VIA Rhine copy breakpoint for copy-only-tiny-frames (int)
parm: avoid_D3:Avoid power state D3 (work-around for broken BIOSes) (bool)

..info from lspci:


00:12.0 Ethernet controller: VIA Technologies, Inc. VT6102 [Rhine-II] (rev 78)
Subsystem: VIA Technologies, Inc. VT6102 [Rhine II] Embeded Ethernet Controller on VT8235
Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping+ SERR- FastB2B- DisINTx-
Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx-
Latency: 32 (750ns min, 2000ns max), Cache Line Size: 32 bytes
Interrupt: pin A routed to IRQ 10
Region 0: I/O ports at cc00 [size=256]
Region 1: Memory at fdffe000 (32-bit, non-prefetchable) [size=256]
Capabilities: [40] Power Management version 2
Flags: PMEClk- DSI- D1+ D2+ AuxCurrent=0mA PME(D0+,D1+,D2+,D3hot+,D3cold+)
Status: D0 PME-Enable- DSel=0 DScale=0 PME-
Kernel driver in use: via-rhine
Kernel modules: via-rhine

..info from /proc/acpi/wakeup:


Device S-state Status Sysfs node
SLPB S5 *enabled
PCI0 S5 disabled no-bus:pci0000:00
USB3 S3 disabled pci:0000:00:10.0
USB4 S3 disabled pci:0000:00:10.1
USB5 S3 disabled pci:0000:00:10.2
USB6 S3 disabled pci:0000:00:10.3
USB7 S3 disabled pci:0000:00:10.4
LAN0 S5 disabled pci:0000:00:12.0
AC97 S5 disabled
MC97 S5 disabled
UAR1 S5 disabled pnp:00:08

..I can change LAN0 via #echo -n LAN0 >/proc/acpi/wakeup ..to:
LAN0 S5 enabled pci:0000:00:12.0
..but I don't know how to change that S5 to a S3.

Finally, here is some of /var/log/messages ..covering the suspend/resume cycle:


Nov 8 13:31:55 squeezenas kernel: uhci_hcd 0000:00:10.3: remove, state 4
Nov 8 13:31:55 squeezenas kernel: usb usb5: USB disconnect, address 1
Nov 8 13:31:55 squeezenas kernel: uhci_hcd 0000:00:10.3: USB bus 5 deregistered
Nov 8 13:31:55 squeezenas kernel: ACPI: PCI interrupt for device 0000:00:10.3 disabled
Nov 8 13:31:55 squeezenas kernel: uhci_hcd 0000:00:10.2: remove, state 4
Nov 8 13:31:55 squeezenas kernel: usb usb4: USB disconnect, address 1
Nov 8 13:31:55 squeezenas kernel: uhci_hcd 0000:00:10.2: USB bus 4 deregistered
Nov 8 13:31:55 squeezenas kernel: ACPI: PCI interrupt for device 0000:00:10.2 disabled
Nov 8 13:31:55 squeezenas kernel: uhci_hcd 0000:00:10.1: remove, state 4
Nov 8 13:31:55 squeezenas kernel: usb usb3: USB disconnect, address 1
Nov 8 13:31:55 squeezenas kernel: uhci_hcd 0000:00:10.1: USB bus 3 deregistered
Nov 8 13:31:55 squeezenas kernel: ACPI: PCI interrupt for device 0000:00:10.1 disabled
Nov 8 13:31:55 squeezenas kernel: uhci_hcd 0000:00:10.0: remove, state 4
Nov 8 13:31:55 squeezenas kernel: usb usb2: USB disconnect, address 1
Nov 8 13:31:55 squeezenas kernel: uhci_hcd 0000:00:10.0: USB bus 2 deregistered
Nov 8 13:31:55 squeezenas kernel: ACPI: PCI interrupt for device 0000:00:10.0 disabled
Nov 8 13:31:55 squeezenas kernel: ACPI: PCI interrupt for device 0000:00:12.0 disabled
Nov 8 13:31:56 squeezenas ntpd[1759]: ntpd exiting on signal 15
Nov 8 13:31:57 squeezenas NET[2161]: /etc/sysconfig/network-scripts/ifdown-post : updated /etc/resolv.conf
Nov 8 13:31:58 squeezenas kernel: PM: Syncing filesystems ... done.
Nov 8 13:32:44 squeezenas kernel: Freezing user space processes ... (elapsed 0.00 seconds) done.
Nov 8 13:32:44 squeezenas kernel: Freezing remaining freezable tasks ... (elapsed 0.00 seconds) done.
Nov 8 13:32:44 squeezenas kernel: Suspending console(s)
Nov 8 13:32:44 squeezenas kernel: sd 2:0:0:0: [sdb] Synchronizing SCSI cache
Nov 8 13:32:44 squeezenas kernel: sd 2:0:0:0: [sdb] Stopping disk
Nov 8 13:32:44 squeezenas kernel: sd 0:0:0:0: [sda] Synchronizing SCSI cache
Nov 8 13:32:44 squeezenas kernel: sd 0:0:0:0: [sda] Stopping disk
Nov 8 13:32:44 squeezenas kernel: serial 00:08: disabled
Nov 8 13:32:44 squeezenas kernel: ACPI: PCI interrupt for device 0000:00:10.4 disabled
Nov 8 13:32:44 squeezenas kernel: ACPI: PCI interrupt for device 0000:00:0f.1 disabled
Nov 8 13:32:44 squeezenas kernel: ACPI: PCI interrupt for device 0000:00:0f.0 disabled
Nov 8 13:32:44 squeezenas kernel: pci_set_power_state(): 0000:00:00.0: state=3, current state=5
Nov 8 13:32:44 squeezenas kernel: ACPI: Preparing to enter system sleep state S3
Nov 8 13:32:44 squeezenas kernel: Disabling non-boot CPUs ...
Nov 8 13:32:44 squeezenas kernel: ACPI: PCI Interrupt 0000:00:0f.0[B] -> Link [LNKB] -> GSI 11 (level, low) -> IRQ 11
Nov 8 13:32:44 squeezenas kernel: ACPI: PCI Interrupt 0000:00:0f.1[A] -> Link [LNKA] -> GSI 10 (level, low) -> IRQ 10
Nov 8 13:32:44 squeezenas kernel: ACPI: PCI Interrupt 0000:00:10.4[C] -> Link [LNKC] -> GSI 11 (level, low) -> IRQ 11
Nov 8 13:32:44 squeezenas kernel: serial 00:08: activated
Nov 8 13:32:44 squeezenas kernel: sd 0:0:0:0: [sda] Starting disk
Nov 8 13:32:44 squeezenas kernel: ata1.00: ACPI cmd ef/03:0c:00:00:00:a0 filtered out
Nov 8 13:32:44 squeezenas kernel: ata1.00: ACPI cmd ef/03:45:00:00:00:a0 filtered out
Nov 8 13:32:44 squeezenas kernel: ata1.00: configured for UDMA/100
Nov 8 13:32:44 squeezenas kernel: sd 0:0:0:0: [sda] 39070080 512-byte hardware sectors (20004 MB)
Nov 8 13:32:44 squeezenas kernel: sd 0:0:0:0: [sda] Write Protect is off
Nov 8 13:32:44 squeezenas kernel: sd 0:0:0:0: [sda] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
Nov 8 13:32:44 squeezenas kernel: sd 2:0:0:0: [sdb] Starting disk
Nov 8 13:32:44 squeezenas kernel: ata3: link is slow to respond, please be patient (ready=0)
Nov 8 13:32:44 squeezenas kernel: ata3.00: revalidation failed (errno=-2)
Nov 8 13:32:44 squeezenas kernel: ata3: failed to recover some devices, retrying in 5 secs
Nov 8 13:32:44 squeezenas kernel: ata3.00: configured for UDMA/133
Nov 8 13:32:44 squeezenas kernel: sd 2:0:0:0: [sdb] 1953525168 512-byte hardware sectors (1000205 MB)
Nov 8 13:32:44 squeezenas kernel: sd 2:0:0:0: [sdb] Write Protect is off
Nov 8 13:32:44 squeezenas kernel: sd 2:0:0:0: [sdb] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
Nov 8 13:32:44 squeezenas kernel: Restarting tasks ... done.
Nov 8 13:32:45 squeezenas kernel: USB Universal Host Controller Interface driver v3.0
Nov 8 13:32:45 squeezenas kernel: ACPI: PCI Interrupt 0000:00:10.0[A] -> Link [LNKA] -> GSI 10 (level, low) -> IRQ 10


I really have no idea which of all that is significant.

..but, on the off chance that I messed up the one thing that I've noticed here, I'll change my /etc/pm/config.d/modules file to reference "via-rhine" rather than "via_rhine" as it does now.

gharris999
2008-11-08, 14:06
Nope. That didn't do it.

This is what I find interesting from /var/log/mesages:


Nov 8 13:51:05 squeezenas kernel: ACPI: PCI interrupt for device 0000:00:10.0 disabled
Nov 8 13:51:05 squeezenas kernel: ACPI: PCI interrupt for device 0000:00:12.0 disabled
Nov 8 13:51:06 squeezenas NET[3031]: /etc/sysconfig/network-scripts/ifdown-post : updated /etc/resolv.conf

If acpi is disabling the interrupt for the built-in nic (device 12), how can a WOL wakeup signal from the device wake the system? Have I got this right? Is this the crux of the problem?

If so, what do I need to do to tell acpi to keep its mitts off the nic?

syburgh
2008-11-08, 14:13
So to replay:

WOL does work on this platform because it wakes after a clean shutdown
via_rhine does not appear to enable WOL when ethtool is run or after module unload (you've confirmed this in your output)
It looks like you are applying the ethtool settings twice (once at system startup with ETHTOOL_OPTS and again before suspend in /etc/pm/sleep.d/15network). It shouldn't hurt, but don't want anyone who finds this thread to think it's important


Things to post:

"ethtool eth0" output from immediately after resume (are negotiated speed and duplex the same as before suspend?)


More to test, if you haven't tried already

What happens when you run "service networks stop" the suspend section of /etc/pm/sleep.d/15network and "service network start" after resume? (keep the module SUSPEND_MODULES/RESUME_MODULES section). This snippet looks interesting
Bringing up interface eth0: RTNETLINK answers: File exists
Error adding address 192.168.0.222 for eth0
Have you tried some of the --quirk-s3* options for pm-suspend? Am grasping, but it could conceivably alter the way the system handles interrupts during S3 depending on your BIOS

syburgh
2008-11-08, 14:29
Based on the driver source, I might expect to see this output (via-rhine.c (http://www.gelato.unsw.edu.au/lxr/source/drivers/net/via-rhine.c))

Woke system up. Reason: Magic packet

Some example output

My lspci -vv snippet for RTL8111C


[root@audio0 ~]# lspci -vv
...
01:00.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL8111/8168B PCI Express Gigabit Ethernet controller (rev 02)
Subsystem: Micro-Star International Co., Ltd. Unknown device 4180
Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B-
Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- >SERR- <PERR-
Latency: 0, Cache Line Size: 32 bytes
Interrupt: pin A routed to IRQ 169
Region 0: I/O ports at e800 [size=256]
Region 2: Memory at febff000 (64-bit, non-prefetchable) [size=4K]
Region 4: Memory at fdff0000 (64-bit, prefetchable) [size=64K]
Expansion ROM at febc0000 [disabled] [size=128K]
Capabilities: [40] Power Management version 3
Flags: PMEClk- DSI- D1+ D2+ AuxCurrent=375mA PME(D0+,D1+,D2+,D3hot+,D3cold+)
Status: D0 PME-Enable- DSel=0 DScale=0 PME-
Capabilities: [50] Message Signalled Interrupts: 64bit+ Queue=0/0 Enable-
Address: 0000000000000000 Data: 0000
Capabilities: [70] Express Endpoint IRQ 1
Device: Supported: MaxPayload 256 bytes, PhantFunc 0, ExtTag-
Device: Latency L0s <512ns, L1 <64us
Device: AtnBtn- AtnInd- PwrInd-
Device: Errors: Correctable- Non-Fatal- Fatal- Unsupported-
Device: RlxdOrd+ ExtTag- PhantFunc- AuxPwr- NoSnoop-
Device: MaxPayload 128 bytes, MaxReadReq 4096 bytes
Link: Supported Speed 2.5Gb/s, Width x1, ASPM L0s L1, Port 0
Link: Latency L0s <512ns, L1 <64us
Link: ASPM Disabled RCB 64 bytes CommClk- ExtSynch-
Link: Speed 2.5Gb/s, Width x1
Capabilities: [b0] MSI-X: Enable- Mask- TabSize=2
Vector table: BAR=4 offset=00000000
PBA: BAR=4 offset=00000800
Capabilities: [d0] Vital Product Data

Here is the dmesg results for a couple of suspend/resume cycles:


[root@audio0 ~]# dmesg | grep 01:00.0
ACPI: PCI Interrupt 0000:01:00.0[A] -> GSI 16 (level, low) -> IRQ 169
PCI: Setting latency timer of device 0000:01:00.0 to 64
PM: Writing back config space on device 0000:01:00.0 at offset f (was 1ff, writing 10a)
PM: Writing back config space on device 0000:01:00.0 at offset c (was 0, writing febc0000)
PM: Writing back config space on device 0000:01:00.0 at offset 8 (was c, writing fdff000c)
PM: Writing back config space on device 0000:01:00.0 at offset 6 (was 4, writing febff004)
PM: Writing back config space on device 0000:01:00.0 at offset 4 (was 1, writing e801)
PM: Writing back config space on device 0000:01:00.0 at offset 3 (was 0, writing 8)
PM: Writing back config space on device 0000:01:00.0 at offset 1 (was 100000, writing 100007)
PM: Writing back config space on device 0000:01:00.0 at offset f (was 1ff, writing 10a)
PM: Writing back config space on device 0000:01:00.0 at offset c (was 0, writing febc0000)
PM: Writing back config space on device 0000:01:00.0 at offset 8 (was c, writing fdff000c)
PM: Writing back config space on device 0000:01:00.0 at offset 6 (was 4, writing febff004)
PM: Writing back config space on device 0000:01:00.0 at offset 4 (was 1, writing e801)
PM: Writing back config space on device 0000:01:00.0 at offset 3 (was 0, writing 8)
PM: Writing back config space on device 0000:01:00.0 at offset 1 (was 100000, writing 100007)
PM: Writing back config space on device 0000:01:00.0 at offset f (was 1ff, writing 10a)
PM: Writing back config space on device 0000:01:00.0 at offset c (was 0, writing febc0000)
PM: Writing back config space on device 0000:01:00.0 at offset 8 (was c, writing fdff000c)
PM: Writing back config space on device 0000:01:00.0 at offset 6 (was 4, writing febff004)
PM: Writing back config space on device 0000:01:00.0 at offset 4 (was 1, writing e801)
PM: Writing back config space on device 0000:01:00.0 at offset 3 (was 0, writing 8)
PM: Writing back config space on device 0000:01:00.0 at offset 1 (was 100000, writing 100007)


It's not certain that the ACPI messages you identified as the cause, but it's worth exploring. I use "pm-suspend --quirk-s3-bios", which might also be worth trying.

gharris999
2008-11-08, 15:16
RE before & after suspend ethtool eth0: no differences (checked with diff)


Settings for eth0:
Supported ports: [ TP MII ]
Supported link modes: 10baseT/Half 10baseT/Full
100baseT/Half 100baseT/Full
Supports auto-negotiation: Yes
Advertised link modes: 10baseT/Half 10baseT/Full
100baseT/Half 100baseT/Full
Advertised auto-negotiation: Yes
Speed: 100Mb/s
Duplex: Full
Port: MII
PHYAD: 1
Transceiver: internal
Auto-negotiation: on
Supports Wake-on: pumbg
Wake-on: g
Current message level: 0x00000001 (1)
Link detected: yes

quirk-checker.sh reports:


Checking your system...

WARNING: You have no quirks!

Suspend should work!

Using either --quirk-s3-bios nor --quirk-s3-mode doesn't seem to make any difference in S3 lan wake-ability.

Re service network stop, etc.. I'm not sure what you're getting at here. I didn't expect that ethtool would work with the network stopped and ifconfig not showing a eth0 device...but it does work. I'm afraid I need a little more direction as to what to try.

Re expected output: where would one expect to see this?

Re lspci -vv:


00:12.0 Ethernet controller: VIA Technologies, Inc. VT6102 [Rhine-II] (rev 78)
Subsystem: VIA Technologies, Inc. VT6102 [Rhine II] Embeded Ethernet Controller on VT8235
Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping+ SERR- FastB2B- DisINTx-
Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx-
Latency: 32 (750ns min, 2000ns max), Cache Line Size: 32 bytes
Interrupt: pin A routed to IRQ 10
Region 0: I/O ports at cc00 [size=256]
Region 1: Memory at fdffe000 (32-bit, non-prefetchable) [size=256]
Capabilities: [40] Power Management version 2
Flags: PMEClk- DSI- D1+ D2+ AuxCurrent=0mA PME(D0+,D1+,D2+,D3hot+,D3cold+)
Status: D0 PME-Enable- DSel=0 DScale=0 PME-
Kernel driver in use: via-rhine
Kernel modules: via-rhine

dmesg | grep 00:12.0


ACPI: PCI Interrupt 0000:00:12.0[A] -> Link [LNKA] -> GSI 10 (level, low) -> IRQ 10
ACPI: PCI interrupt for device 0000:00:12.0 disabled
ACPI: PCI Interrupt 0000:00:12.0[A] -> Link [LNKA] -> GSI 10 (level, low) -> IRQ 10
ACPI: PCI interrupt for device 0000:00:12.0 disabled
ACPI: PCI Interrupt 0000:00:12.0[A] -> Link [LNKA] -> GSI 10 (level, low) -> IRQ 10
ACPI: PCI interrupt for device 0000:00:12.0 disabled
ACPI: PCI Interrupt 0000:00:12.0[A] -> Link [LNKA] -> GSI 10 (level, low) -> IRQ 10

syburgh
2008-11-08, 15:23
Re service network stop, etc.. I'm not sure what you're getting at here. I didn't expect that ethtool would work with the network stopped and ifconfig not showing a eth0 device...but it does work. I'm afraid I need a little more direction as to what to try.

In the pm-utils hook script (15network) have it run "/sbin/service network stop" on suspend (where you call enable_wol) and "/sbin/service network start" on resume (where you call "service network restart"). You can comment out the existing calls (enable_wol and restart_network)

gharris999
2008-11-08, 15:38
Gotcha. No, that didn't make a difference either.

dmesg after suspend/resume:


PM: Syncing filesystems ... done.
PM: Preparing system for mem sleep
Freezing user space processes ... (elapsed 0.00 seconds) done.
Freezing remaining freezable tasks ... (elapsed 0.00 seconds) done.
PM: Entering mem sleep
Suspending console(s)
sd 2:0:0:0: [sdb] Synchronizing SCSI cache
sd 2:0:0:0: [sdb] Stopping disk
sd 0:0:0:0: [sda] Synchronizing SCSI cache
sd 0:0:0:0: [sda] Stopping disk
serial 00:08: disabled
ACPI: PCI interrupt for device 0000:00:10.4 disabled
ACPI: PCI interrupt for device 0000:00:10.3 disabled
ACPI: PCI interrupt for device 0000:00:10.2 disabled
ACPI: PCI interrupt for device 0000:00:10.1 disabled
ACPI: PCI interrupt for device 0000:00:10.0 disabled
ACPI: PCI interrupt for device 0000:00:0f.1 disabled
ACPI: PCI interrupt for device 0000:00:0f.0 disabled
ACPI: Preparing to enter system sleep state S3
Disabling non-boot CPUs ...
Back to C!
PCI: Setting latency timer of device 0000:00:01.0 to 64
ACPI: PCI Interrupt 0000:00:0f.0[B] -> Link [LNKB] -> GSI 11 (level, low) -> IRQ 11
ACPI: PCI Interrupt 0000:00:0f.1[A] -> Link [LNKA] -> GSI 10 (level, low) -> IRQ 10
ACPI: PCI Interrupt 0000:00:10.0[A] -> Link [LNKA] -> GSI 10 (level, low) -> IRQ 10
usb usb2: root hub lost power or was reset
ACPI: PCI Interrupt 0000:00:10.1[A] -> Link [LNKA] -> GSI 10 (level, low) -> IRQ 10
usb usb3: root hub lost power or was reset
ACPI: PCI Interrupt 0000:00:10.2[B] -> Link [LNKB] -> GSI 11 (level, low) -> IRQ 11
usb usb4: root hub lost power or was reset
ACPI: PCI Interrupt 0000:00:10.3[B] -> Link [LNKB] -> GSI 11 (level, low) -> IRQ 11
usb usb5: root hub lost power or was reset
ACPI: PCI Interrupt 0000:00:10.4[C] -> Link [LNKC] -> GSI 11 (level, low) -> IRQ 11
serial 00:08: activated
sd 0:0:0:0: [sda] Starting disk
ata1.00: ACPI cmd ef/03:0c:00:00:00:a0 filtered out
ata1.00: ACPI cmd ef/03:45:00:00:00:a0 filtered out
ata1.00: configured for UDMA/100
sd 0:0:0:0: [sda] 39070080 512-byte hardware sectors (20004 MB)
sd 0:0:0:0: [sda] Write Protect is off
sd 0:0:0:0: [sda] Mode Sense: 00 3a 00 00
sd 0:0:0:0: [sda] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
sd 2:0:0:0: [sdb] Starting disk
ata3: link is slow to respond, please be patient (ready=0)
ata3.00: NODEV after polling detection
ata3.00: revalidation failed (errno=-2)
ata3: failed to recover some devices, retrying in 5 secs
ata3.00: configured for UDMA/133
sd 2:0:0:0: [sdb] 1953525168 512-byte hardware sectors (1000205 MB)
sd 2:0:0:0: [sdb] Write Protect is off
sd 2:0:0:0: [sdb] Mode Sense: 00 3a 00 00
sd 2:0:0:0: [sdb] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
PM: Finishing wakeup.
Restarting tasks ... done.
via-rhine.c:v1.10-LK1.4.3 2007-03-06 Written by Donald Becker
ACPI: PCI Interrupt 0000:00:12.0[A] -> Link [LNKA] -> GSI 10 (level, low) -> IRQ 10
eth0: VIA Rhine II at 0xfdffe000, 00:40:63:ed:3b:66, IRQ 10.
eth0: MII PHY found at address 1, status 0x786d advertising 05e1 Link c5e1.
eth0: link up, 100Mbps, full-duplex, lpa 0xC5E1
eth0: no IPv6 routers present

..and right after resume via power switch (still no wol), ethtool eth0:


Settings for eth0:
Supported ports: [ TP MII ]
Supported link modes: 10baseT/Half 10baseT/Full
100baseT/Half 100baseT/Full
Supports auto-negotiation: Yes
Advertised link modes: 10baseT/Half 10baseT/Full
100baseT/Half 100baseT/Full
Advertised auto-negotiation: Yes
Speed: 100Mb/s
Duplex: Full
Port: MII
PHYAD: 1
Transceiver: internal
Auto-negotiation: on
Supports Wake-on: pumbg
Wake-on: g
Current message level: 0x00000001 (1)
Link detected: yes

syburgh
2008-11-08, 15:50
Tried leaving the module loaded and stop/start the network service?

I don't think the halt script would remove it on shutdown. Maybe this is important for via_rhine.

gharris999
2008-11-08, 15:50
This really sucks. I've got three systems built around these via cn boards and I'll be really disappointed if S3 is unusable on them. I bought a Jetway J7F4K1G2E board based on VIA CN700 + 8237RP, but it doesn't seem to support S3 at all, at least there is no mention of it in the BIOS and pm-suspend reports that suspend isn't supported.

Finally, I do have a MSI 945GM1 MS-7265, rev 3 board which is a lot closer to the atom board in the Wind PC. Hopefully, S3 will work on that. I'm waiting on a used core 2 duo mobile cpu from ebay before I can get that one going.

vrobin
2008-11-08, 16:05
syburgh: I spent most of yesterday working through the permutations of your excellent wiki, but to no avail. This is what I've tried so far:

1). Created /etc/pm/config.d/modules ..containing:


SUSPEND_MODULES="uhci_hcd mii via_rhine"
RESUME_MODULES="via_rhine uhci_hcd"


At this point, my hair is lying in tufts on the floor and my cranium is sporting new bald spots.

Any help you folks can give me in tracking this down would be greatly appreciated. Thanks.


The fun of it is that the problem I talked you about was about the same network adapter:

(extract from my dmesg)
via-rhine.c:v1.10-LK1.4.3-d10 2007-03-06 Written by Donald Becker
via-rhine: Broken BIOS detected, avoid_D3 enabled.
ACPI: PCI Interrupt 0000:00:12.0[A] -> Link [LNKA] -> GSI 10 (level, low) -> IRQ 10
RVTEST: wolstat 00
eth%d: Reset succeeded.
eth0: VIA Rhine II at 0xfdffe000, 00:40:63:e8:38:ef, IRQ 10.
eth0: MII PHY found at address 1, status 0x786d advertising 05e1 Link 45e1.
eth0: No D3 power state at shutdown.

If I were you, I would try to replace the linux default driver with the driver provided by via. Or you could write via-rhine maintainer about the problem...

syburgh
2008-11-08, 16:06
Sounds unfortunate. Maybe pm-hibernate works? I found Intel D201GLY2 supported S4 (hibernate) but not S3 (suspend).

gharris999
2008-11-08, 16:28
Ok, tried removing /etc/pm/config.d/modules and kept the stop/start network. No difference.

Interestingly, if I bypass pm-suspend and push into S3 via:

echo -n "mem" > /sys/power/state

..WOL still doesn't work, but this is what gets reported in /var/log/messages:


Nov 8 16:24:35 squeezenas kernel: PM: Syncing filesystems ... done.
Nov 8 16:25:15 squeezenas kernel: Freezing user space processes ... (elapsed 0.01 seconds) done.
Nov 8 16:25:15 squeezenas kernel: Freezing remaining freezable tasks ... (elapsed 0.00 seconds) done.
Nov 8 16:25:15 squeezenas kernel: Suspending console(s)
Nov 8 16:25:15 squeezenas kernel: sd 2:0:0:0: [sdb] Synchronizing SCSI cache
Nov 8 16:25:15 squeezenas kernel: sd 2:0:0:0: [sdb] Stopping disk
Nov 8 16:25:15 squeezenas kernel: sd 0:0:0:0: [sda] Synchronizing SCSI cache
Nov 8 16:25:15 squeezenas kernel: sd 0:0:0:0: [sda] Stopping disk
Nov 8 16:25:15 squeezenas kernel: parport_pc 00:09: disabled
Nov 8 16:25:15 squeezenas kernel: serial 00:08: disabled
Nov 8 16:25:15 squeezenas kernel: ACPI: PCI interrupt for device 0000:00:10.4 disabled
Nov 8 16:25:15 squeezenas kernel: ACPI: PCI interrupt for device 0000:00:10.3 disabled
Nov 8 16:25:15 squeezenas kernel: ACPI: PCI interrupt for device 0000:00:10.2 disabled
Nov 8 16:25:15 squeezenas kernel: ACPI: PCI interrupt for device 0000:00:10.1 disabled
Nov 8 16:25:15 squeezenas kernel: ACPI: PCI interrupt for device 0000:00:10.0 disabled
Nov 8 16:25:15 squeezenas kernel: ACPI: PCI interrupt for device 0000:00:0f.1 disabled
Nov 8 16:25:15 squeezenas kernel: ACPI: PCI interrupt for device 0000:00:0f.0 disabled
Nov 8 16:25:15 squeezenas kernel: ACPI: Preparing to enter system sleep state S3
Nov 8 16:25:15 squeezenas kernel: Disabling non-boot CPUs ...
Nov 8 16:25:15 squeezenas kernel: ACPI: PCI Interrupt 0000:00:0f.0[B] -> Link [LNKB] -> GSI 11 (level, low) -> IRQ 11
Nov 8 16:25:15 squeezenas kernel: ACPI: PCI Interrupt 0000:00:0f.1[A] -> Link [LNKA] -> GSI 10 (level, low) -> IRQ 10
Nov 8 16:25:15 squeezenas kernel: ACPI: PCI Interrupt 0000:00:10.0[A] -> Link [LNKA] -> GSI 10 (level, low) -> IRQ 10
Nov 8 16:25:15 squeezenas kernel: usb usb2: root hub lost power or was reset
Nov 8 16:25:15 squeezenas kernel: ACPI: PCI Interrupt 0000:00:10.1[A] -> Link [LNKA] -> GSI 10 (level, low) -> IRQ 10
Nov 8 16:25:15 squeezenas kernel: usb usb3: root hub lost power or was reset
Nov 8 16:25:15 squeezenas kernel: ACPI: PCI Interrupt 0000:00:10.2[B] -> Link [LNKB] -> GSI 11 (level, low) -> IRQ 11
Nov 8 16:25:15 squeezenas kernel: usb usb4: root hub lost power or was reset
Nov 8 16:25:15 squeezenas kernel: ACPI: PCI Interrupt 0000:00:10.3[B] -> Link [LNKB] -> GSI 11 (level, low) -> IRQ 11
Nov 8 16:25:15 squeezenas kernel: usb usb5: root hub lost power or was reset
Nov 8 16:25:15 squeezenas kernel: ACPI: PCI Interrupt 0000:00:10.4[C] -> Link [LNKC] -> GSI 11 (level, low) -> IRQ 11
Nov 8 16:25:15 squeezenas kernel: via-rhine: Woke system up. Reason: Magic packet.
Nov 8 16:25:15 squeezenas kernel: eth0: link up, 100Mbps, full-duplex, lpa 0xC5E1
Nov 8 16:25:15 squeezenas kernel: serial 00:08: activated
Nov 8 16:25:15 squeezenas kernel: parport_pc 00:09: activated
Nov 8 16:25:15 squeezenas kernel: sd 0:0:0:0: [sda] Starting disk
Nov 8 16:25:15 squeezenas kernel: ata1.00: ACPI cmd ef/03:0c:00:00:00:a0 filtered out
Nov 8 16:25:15 squeezenas kernel: ata1.00: ACPI cmd ef/03:45:00:00:00:a0 filtered out
Nov 8 16:25:15 squeezenas kernel: ata1.00: configured for UDMA/100
Nov 8 16:25:15 squeezenas kernel: sd 0:0:0:0: [sda] 39070080 512-byte hardware sectors (20004 MB)
Nov 8 16:25:15 squeezenas kernel: sd 0:0:0:0: [sda] Write Protect is off
Nov 8 16:25:15 squeezenas kernel: sd 0:0:0:0: [sda] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
Nov 8 16:25:15 squeezenas kernel: sd 2:0:0:0: [sdb] Starting disk
Nov 8 16:25:15 squeezenas kernel: ata3: link is slow to respond, please be patient (ready=0)
Nov 8 16:25:15 squeezenas kernel: ata3.00: configured for UDMA/133
Nov 8 16:25:15 squeezenas kernel: sd 2:0:0:0: [sdb] 1953525168 512-byte hardware sectors (1000205 MB)
Nov 8 16:25:15 squeezenas kernel: sd 2:0:0:0: [sdb] Write Protect is off
Nov 8 16:25:15 squeezenas kernel: sd 2:0:0:0: [sdb] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
Nov 8 16:25:15 squeezenas kernel: Restarting tasks ... done.

There's your via-rhine wakeup message...but the system didn't wake!

gharris999
2008-11-08, 17:23
If I were you, I would try to replace the linux default driver with the driver provided by via. Or you could write via-rhine maintainer about the problem...
My problem at the moment is that I'm down to using a 1/2 speed dial-up internet connection and I don't have the kernel source installed on this machine. So, I can't compile the via driver at this time. I have to wait two weeks before my satellite internet comes back on-line. One of the many advantages of living in the sticks.

I guess I should stop obsessing about this problem and get back to ripping music, eh?

If, eventually, I do get the via driver to compile, install and work, I'll report back here.

Thanks for all the help.

syburgh
2008-11-08, 17:40
One last thought: http://linux.derkeiler.com/Mailing-Lists/Fedora/2007-02/msg01577.html

vrobin
2008-11-09, 04:34
About the "avoid D3", I don't know about your precise motherboard, but on mine ( CN-10000EG ) the problem wasn't there at all as this avoid D3 was needed just for a given NIC not relative to epia MB.

If you're facing the same problem as I did, you're probably in a dead end tweaking modules shutdown/restore.

To sum up, after some tweaks, I managed to wake up by WOL only once every two sleep states (I think the first WOL after full reboot was working, but not the second). The problem seemed related to either internal registry of the chipset or the way it works with ACPI layer... in one word, it's an impossible mess!

I wrote the maintainer, tried some patching, did a lot of specific code debugging in the kernel (by adding a kprintf every two lines) but I wasn't lucky and after 3 weeks, I stopped trying.

All I remember is the the via driver worked correctly the first time:
http://www.viaarena.com/default.aspx?PageID=420&OSID=25&CatID=2590&SubCatID=180
(looking in viaarena, you also can find binaries).

Good luck,
Robin

vrobin
2008-11-09, 04:37
I've just found this page:
http://kerneltrap.org/mailarchive/linux-netdev/2008/10/4/3508494

I haven't read it entirely, but it seems interesting at first sight...

Robin

syburgh
2008-11-09, 05:54
Wow, what a mess. Atom boards are pretty inexpensive these days...

vrobin
2008-11-09, 06:01
Wow, what a mess. Atom boards are pretty inexpensive these days...

The problem (the one I faced at least) is not intel atom vs via c7, but rather with a broken driver and that could happen with atom too.

Despite this, I was looking at atom MB, I just couldn't find fanless atom boards... are you aware of such products?

Robin

syburgh
2008-11-09, 11:16
Don't know of any fanless Atom boards, but I replaced a fanless Core Solo system (AOpen DE-945FL) SC system with an actively cooled Atom (MSI Wind PC) and have been quite happy with the result. The disk was the loudest aspect of both systems.

Atom boards seem to be pretty recent designs with good feature support (considering the "mature" state of the chipset). It's unfortunate that ACPI and WOL are so inconsistently implemented, even after so many years in deployment.

gharris999
2008-11-09, 12:27
vrobin: thanks for all that. I've gotten to where you got: the system, in S3, will WOL only every other time. WTF, man. Obviously, something is getting toggled, but what, and by whom?

The info at kernel trap you pointed me to is interesting. But it doesn't seem to work for this case. His instruction, modified to:

echo -n enabled > "/sys/bus/pci/devices/0000:00:12.0/power/wakeup"

..just results in a:

-bash: echo: write error: Invalid argument


I also didn't manage to find a binary driver for the VT6102 Rhine-II, rev 78 for Fedora 9, so I'm back to waiting until I can install the kernel source and compile the driver.

Again, thanks to you both for the hand-holding.

PS: syburgh: this experience has made the Wind PC look more and more interesting to me. My current SC servers have one 2.5" PATA system drive and one or two 3.5" SATA audio (& backup) drives. I take it that there are no IDE ports on the Wind PC's mobo and that I should look at using CF for a system drive instead. Is that true? If so, is there a particular CF module you would recommend? Could one fit the CF plus two 3.5" SATA drives into the Wind PC's chassis? What is the motherboard that comes in the Wind PC?

Sorry to pester you with these questions, but it looks like you're the expert on this system.

Edit: now that I've taken a closer look at your wiki at http://wiki.syburgh.com/audio0-local, it looks as though you're not using a CF as a boot device.

vrobin
2008-11-09, 14:02
vrobin: thanks for all that. I've gotten to where you got: the system, in S3, will WOL only every other time. WTF, man. Obviously, something is getting toggled, but what, and by whom?

The info at kernel trap you pointed me to is interesting. But it doesn't seem to work for this case. His instruction, modified to:

echo -n enabled > "/sys/bus/pci/devices/0000:00:12.0/power/wakeup"

..just results in a:

-bash: echo: write error: Invalid argument


I also didn't manage to find a binary driver for the VT6102 Rhine-II, rev 78 for Fedora 9, so I'm back to waiting until I can install the kernel source and compile the driver.


The post in kerneltrap is rather recent, it may be a command new to kernels 2.6.27 or upper (I'm only 2.6.21).
There's a very small chance that binary driver suits your exact distro as these pages aren't update too often, but by any luck, some binary module could have worked for you... the worst you're risking is some good old kernel panic.

If the problem you have is the same as mine (it seems very probable), you can count on the rhinefet driver to work. I was just displeased not to use an official kernel driver, that's why I kept it. As the system is noiseless and not very power consummating, I find more convenient to keep it on 24/7. It's running for 18 months now, and not a single trouble to report.

There is another thing that can help you... try another nic. During my debug tests, I plugged another NIC in the PCI expansion slot and the WOL worked perfectly with this new NIC board. So I can assure you that the problem is with the linux rhine driver.

@syburgh: I always had bad experiences with via chipsets, and the thin epia base customers don't encourage strong support for these hardware... intel broader base is a good point. I hope they'll do some fanless mini itx board for the time I'll have to replace my epia.

gharris999
2008-11-09, 19:32
One other thing that I've noticed with these Via EPIA CN boards: if I set spin-down on my media disk (1T Seagate ST31000340AS) via:

/sbin/hdparm -S120 /dev/sdb

..in rc.local, the drive never spins back up and the system hangs.

I don't know the quirks of hdparm nor of this seagate drive. But I do know that this spin-down works just fine with my old intel chipset based SC server.

syburgh
2008-11-10, 06:03
PS: syburgh: this experience has made the Wind PC look more and more interesting to me. My current SC servers have one 2.5" PATA system drive and one or two 3.5" SATA audio (& backup) drives. I take it that there are no IDE ports on the Wind PC's mobo and that I should look at using CF for a system drive instead. Is that true? If so, is there a particular CF module you would recommend? Could one fit the CF plus two 3.5" SATA drives into the Wind PC's chassis? What is the motherboard that comes in the Wind PC?

Sorry to pester you with these questions, but it looks like you're the expert on this system.

Edit: now that I've taken a closer look at your wiki at http://wiki.syburgh.com/audio0-local, it looks as though you're not using a CF as a boot device.

You're correct-- there are no IDE connectors on the board. The kernel IDE driver finds ports: the master is connected to the CF slot and the other has no connector at all. The Atom boards from Intel do have an IDE connector, so that might be a consideration.

I do have a CF module installed (Transcend 266x 2GB), it is presently unused (intended to host FreeNAS or something similar if they ever support this hardware, ACPI S3, and WOL).

2x3.5in SATA disks would work fine, though you will need ears to mount the second disk in the 5.25in bay. Power and SATA signal cables are pre-routed so it should be pretty straightforward.

There are some good internal photos of the system on Flickr (they're not mine, but mine looks the same): http://flickr.com/photos/steve_price82/2827191996

florca
2008-11-10, 12:27
Despite this, I was looking at atom MB, I just couldn't find fanless atom boards... are you aware of such products?

MSI have just introduced a fanless Atom board, but I think it's aimed at the embedded / industrial market and looks pretty pricey ($175 was one price I saw).

Their press release is here: http://global.msi.com.tw/index.php?func=newsdesc&news_no=674 and there's a comparative review of it (under its old name) here: http://www.mini-itx.com/reviews/atoms/default.asp?page=1

Very tempted, but it would be good to see some competitive products and / or a dual-core version....

Brgds
Phil

gharris999
2008-12-01, 21:00
Just a follow up on my wol problems with the Via EPIA CN1000G board:

I finally got the Via Rhine driver source ver 5.15, with file dates from July of 2008. The driver compiled and installed without error against the current Fedora 10 kernel: 2.6.27.5-117.fc10.i686. The driver seems to work fine, network traffic wise. But wol doesn't work at all!!! The system, with this driver installed, is seemingly deaf to all wol attempts.

Ethtool eth0 reports the the driver/nic supports wake-on: puag. Ethtool can set the nic to any and all of those modes. However, when I attempt to disable wol, ethtool reports:


# ethtool -s eth0 wol d
Cannot set new wake-on-lan settings: Bad address
not setting wolUninstalling the rhinefet driver and restoring the kernel's stock via-rhine driver restores what limited wol this board seems to have, i.e. wol works from halt, but not from S3.

What a pita. I'll be buying one of those fan-less atom boards florca mentioned when they become available later this month. And I guess I'll have 3 Via boards for sale if the atom works out.

epoch1970
2008-12-02, 04:53
Just in case, because I've seen this black magick happening. On my asus core solo board I've had transient reliability problems with reboot or shutdown (stuck at "Will now restart" / hanging at reboot), executed by the system or by the watchdog (the software C part *and* the iTCO hardware part).
The problem could happen after at least one suspend-to-ram/wake-up cycle.

I think I finally came to the bottom of the problem. I got rid of pm-utils completely, and call directly "s2ram -f -a 3" to suspend to ram. Now my system seems to reboot and powercycle as it should.
While I had S3/WOL working reliably with "s2ram -f", I noticed there was something not quite right using vbetools.
Prior to the 1st suspend, "vbetools vbemode get" would return a value, and after wake-up the value was different.
Adding the "--acpi-both" (= -a3) option to s2ram allowed for vbemode to return the same as before sleep. And now it looks like I reboot better.

The machine is headless, has no X installed, and I use the kernel boot option "vga=0". The board has an Intel 945-somesuch IGP chip.

I'm probably off-topic, but who knows :-)

gharris999
2008-12-02, 09:35
I'm probably off-topic, but who knows :-)
No, no, not at all. Thanks for this. I'll give this a try.

gharris999
2009-01-12, 11:03
MSI have just introduced a fanless Atom board, but I think it's aimed at the embedded / industrial market and looks pretty pricey ($175 was one price I saw).

Their press release is here: http://global.msi.com.tw/index.php?func=newsdesc&news_no=674 and there's a comparative review of it (under its old name) here: http://www.mini-itx.com/reviews/atoms/default.asp?page=1

Very tempted, but it would be good to see some competitive products and / or a dual-core version....

Brgds
Phil
I bit the bullet and bought one of these boards. Testing it with Fedora 10 and Ubuntu 8.10, so far it looks as though it doesn't suffer from any of the defects that the EPIDA CN boards have in terms of S3 sleep and WOL-ability. The N270 atom cpu, with hyper-threading enabled, doesn't seem to provide a dramatic performance improvement over the Via board. But I haven't done any sort of real benchmarking here.

If it wasn't so ludicrously expensive, I'd recommend it as a good replacement for the CN10000E and the like.

See: http://www.logicsupply.com/products/ms_9830

PS: for Ubuntu, I had to download Intel's e1000e driver code and compile it in order to get the NICs recognized. Also, the board came from Logic Supply without any manual and there's no manual on the MSI website either. Logic Supply, after a phone call to their tech support number, was kind enough to email one to me, though.

Shellness
2009-11-03, 07:27
I am a techno-novice and after two years of procrastination I am going to build a server along the lines of that in http://www.ulverston.myzen.co.uk/mini-itx/index.htm. Before I jump I would welcome some advice from the forum:

1. Motherboard type/manufacturer – are there any recommendations as to what to aim for, to the uninitiated the choice is bewildering (I am presuming that board specifications have advanced since the guide was written).

2. Power – I will go for an SSD for the boot disc to keep the noise and power down. I want the whole assembly to consume as little power as possible so wonder what best to specify in terms of components, including hard discs. Additionally, it is unclear from the article what power the 60W power supply pulls when the unit is dormant or is just serving internet radio. All steers welcomed.

3. Back-up – I want to use the unit to back-up the PCs in the house and to act as a back-up to all of the music & photo files that will be on the primary hard disc, i.e. I expect to have two large hard discs which act as a back-up to each other (I see this as the only practical way of ensuring my music/photo files are safe from a disc failure). Again, all steers welcomed, noting that any discussions on ways of keeping my data duplicated welcomed (RAID? – please remember my opening comment about being a techno-novice).

For the record, today I have just one Squeezebox3. I expect to add to this a Squeezebox Duet once I have got the server working.

Thanks.

hellesangel
2009-11-04, 01:35
That's quite an undertaking for a techno novice, I suspect you'll learn a bit along the way... The golden piece of advice: Go slowly and carefully!

First, motherboard - the choice is indeed bewildering but then if you make a list of what you actually _need_ (stressed, need) then this usually reduces the selection quite a bit. Before buying enter your favourite board manufacturer and name into Google alongside terms like 'stability problem' or 'system crash' etc. to see what others are posting on forums like this.

For audio serving and backups you _need_ a relatively modest CPU/RAM, a core 2duo is tonnes, check out VIA cpus too, gigabit LAN is not necessary, make sure you have enough SATA connectors (you talk about building a RAID array), don't worry about 'hardware' RAID support on the board (see below), and VGA on board is great if you're using this primarily as a server, and check out the power control features the board may have. Boot from LAN, USB, SD card etc is useful.

Power is a fun area, coupled with noise. I've lost track a bit of current developments but current disks are pretty quiet, i run a server with 5x Samsung Eco-green HD103SI disks at 5400rpm and they're almost inaudible. Remember the more stuff you pack into a small case the bigger the cooling problem gets - running big fans slower makes less noise for the same air movement than smaller fans faster.

The backup idea is an odd one, somewhat at odds with your other requirements for this box. You say you want to boot from SSD to avoid a hard disk, then pack two in anyway for backup? It's a pretty clever mainboard that can shut down these disks entirely when they're not being used so that means noise and heat from them permanently. Advanced power control like this would be is the first thing a board's manufacturer screws up so check it actually works before buying - even dedicated NAS devices often don't fully shut down unused disks because of driver screwups.

Also remember two points about backups: First, backups are best made, verified, then taken off-line to ensure they don't get corrupted along with your original. They usually form a snapshot of your data at a particular instant, then held safe - a hard disk that is permanently accessible is not safe. Second, RAID arrays give fault tolerance, they are _not_ a backup! Always have a second copy, perhaps in a second RAID array, but never think 'oh I've mirrored my disk using RAID so I'm safe'.

About installing a software RAID array - this is pretty easy with most modern Linux distributions using mdadm (Google that), and this is an excellent tool - but this is not for novices. Neither is installing a hardware RAID array. If you do either then build your array up, fill it with test data, then go through a few tests to simulate a disk failure and ensure you can rebuild the array before you fill it with your critical data. And keep a backup, as I said above. I say this from experience gained the hard way...

Hope this helps.

Shellness
2009-11-04, 04:16
Thanks for the comprehensive reply, and belated thanks to Michael for the solid guide. I have some iterative queries, the biggest first.

To clarify the back-up/RAID query perhaps there is danger of nomenclatural confusion. What I want is:
1. To be able to copy data from household PCs over WiFi to a storage device. In the event of a PC crash I would not have lost data
2. To be able to maintain a current copy of non-PC data. Specifically this is >1,200 ripped CDs (that are served to my Squeezeboxes) plus other odds and sods that I do not need to keep on PCs (e.g. five years of photos, old tax records, employment records, etc)

For item (1) I was presuming to use one hard disc where I would copy files over WiFi when I wanted to “back-up” (using my loose terminology) a PC. Ideally this would be automated using something like Syncback (as recommended by Michael’s guide)

For item (2) I was presuming that what I needed to do was to run two hard discs which mirrored each other, ideally automatically and without my intervention.

I presumed that to satisfy both (1) and (2) that I needed a pair of discs. In addition, I presumed to use an SSD to boot the ITX in order to minimise energy use, noise, etc (this also avoids having data stored on a boot device and, since I use Squeeze to listen to a lot of internet radio rather than just to serve FLACs & podcasts, avoids having a hard disc twitching away unnecessarily).

I note your comments about checking the power management to ensure that the motherboard does actually power-down the hard discs and stop them twitching. I also note your comments about software RAID. Perhaps what I should have asked was: given my requirements what do I need to do to provide the ability to (1) hold a copy of PC data and (2) duplicate all server data on two distinct discs?

Concerning the choice of motherboard, I don’t know who manufactures them so the idea of Googling my favourite manufacturer is a non-starter. Given my pup status perhaps an exchange with people at mini-itx.com on what they recommend is in order? I was presuming the board to be supplied with the CPU so that this choice was taken from me.

On discs, do I gather that the lower the rpm the lower the potential noise, heat and energy consumption? I had always presumed that the bigger the number the better the disc on the basis of data access speeds.

Lastly, I was not explicit in my posting. I wish to avoid the use of a fan altogether, as Michael did. Does this suggest that I am best to limit myself to 2.5” hard discs to try to keep the power and heat down? I was considering 3.5” given my expectations for storage capacity over time.

As you can see, given my novice status, “to presume” is high up my list of repeated verbs. I am looking forward to the project and, after two years of chewing the end of my pencil, I’m ready to start. So if you or anyone else has anything to add to the earlier response I would appreciate the advice so that I can hone down my choices and get on with the build.

Thank you.

hellesangel
2009-11-04, 07:04
Ah, I see what you mean with 'backup' - sort of a security copy, that's fair enough then. Just be aware of the limitations and possible drawbacks and risks of this approach.

Two hard disks that mirror each other without your intervention is called RAID1, and mdadm can set that up easily enough. Just choose two disks of similar size, ideally identical units, and a few commands later you will have a fully mirrored setup. This assumes your server is running Linux - and Ubuntu seems popular, I chose openSUSE a while back at random and just stuck with it. Either way it's a learning curve to climb in itself.

There are some main boards that have the CPU mounted but generally you have a socket of a particular type and have to choose a CPU that fits. This is not a big deal if you have to do it, ask again if you need this advice!

About powering down the disks - I'm sceptical this will work, that you'll find a main board that supports it, look carefully at the options and Google for user comments. A 2.5" disk makes a lot less noise but they're expensive and limited in storage. And also remember that a disk spinning constantly makes less noise than one that periodically clunks into park position, then clunks out of it again.

syburgh
2009-11-04, 07:46
I've some experience with these sorts of systems. Some of my perspectives:

It's a far more difficult DIY project to build a system with passive cooling than a system with low/limited airflow. Even with a very efficient system (e.g. Atom, Via) you will still encounter high temperatures that can reduce the life of your mechanical hard disks (assuming they are internal, external drives tend to have even worse cooling)
Your interest in spinning down the hard drives will run counter to use of software RAID (even mirroring). It is possible to have both software RAID and spin-down (at least with Linux), but requires an in-depth understanding of the mechanics of the Linux RAID subsystem (md), caching, your filesystem of choice, etc. I would not recommend using the RAID "feature" on any desktop motherboard (in Linux parlance this is known as "fakeraid". Google the term for details).
Photos, records, and your audio library could be more of a write rarely (mostly once) and read occasionally sort of scenario. This is not a situation where the administrative complexity or inflexibility of RAID is clearly offset by one of RAID's benefits (e.g. performance, automated recovery, or degraded availability).
It could be both easier and more robust for you to invest time automating things with syncback, rsync, or robocopy (this keeps the disks independent, so they can/should spin-down when idle) than figuring out how to have both RAID and spin-down. If the performance of a single spindle proves insufficient you can always migrate to software RAID
If you are worried about recovery time/inconvenience it could be worth figuring out how to keep as much of your data as possible mounted read-only (remount rw when you need to write to it). This offers very good protection against power outages
I also use flash for the OS. It's worth considering an IDE (or CF-IDE), SATA (or CF-SATA), or USB header flash module (e.g. from Transcend, Logic Supply, etc.) as they tend to be less expensive than an entire 2.5 or 3.5 in flash drive and have a small enough capacity that you won't be tempted to store data there in future years. Modules don't require a 2.5 or 3.5 in bay, so you might be able to use a more compact case if you go this route.

Your project sounds fun-- good luck!

Shellness
2009-11-04, 10:05
Thank you for the reply.

Your assessment of the data held on the discs is a fair one in that they change infrequently and, largely, get added to and not deleted (music, photos, etc). What I would like is to automate as much as possible the mirroring between the two hard drives so that adding (say) a music folder to one automatically gets replicated on the other, and that I do this simply by dragging and dropping the file over WiFi from my PC. (Currently the malarkey I have to go through to achieve this from my PC onto two external hard drives which are stored away after use is such that it discourages frequent copy and “back-up”.) Likewise I wish to automate the back-up/security-copy of PCs so that no-one in the house has to think about hunting the external hard drives out, etc, etc, etc – back-ups here are done monthly at best and this concerns me.

I will check your suggestions of syncback, rsync and robocopy for the above.

Meanwhile, concerning the use of a different flash for the boot process, it looks as though this has the potential to free up a SATA socket which looks like a handy feature should I ever need a third slot. I presume that pointing the CPU at the USB drive to find the boot code is simple?

maggior
2009-11-04, 12:29
I also use flash for the OS. It's worth considering an IDE (or CF-IDE), SATA (or CF-SATA), or USB header flash module (e.g. from Transcend, Logic Supply, etc.) as they tend to be less expensive than an entire 2.5 or 3.5 in flash drive and have a small enough capacity that you won't be tempted to store data there in future years. Modules don't require a 2.5 or 3.5 in bay, so you might be able to use a more compact case if you go this route.


I too have flash for my OS drive. I don't have a super power efficient server; my goal was to save drive bay space and drive connections. I also like separating my data and OS.

In my main server, I'm using an 8GB PNY USB flash key running OpenSuSE 11.0. I've been running this server for about a year trouble free!

I just configured a second server using an 8GB 233x CompactFlash card installed in a CF-IDE adapter. I just set this up last night and it works great. That machine is running OpenSuSE 11.1.

The plus of either of these solutions is that they will be cheaper than an SSD.

The USB stick is the cheapest route. I got my PNY 8GB USB stick for $25 a year ago. Speed-wise, you are limited by the 12 MB/s bandwidth of USB, plus the USB interface will eat some of your CPU. For running all of the squeezeboxes in my house, it is more than sufficient.

The CompactFlash solution is the best of everything IMO. My 233x 8GB flash card and CF-IDE adapter cost me ~$45. You get better throughput without using CPU bandwidth. This mattered on this machine since it is older. Here, you can't cheap out on the speed of the card. A 233x or 266x should be good enough, but don't get an 80x or something like that.

In both cases, I installed the OS as though it were a hard drive. I did not do anything special to minimize disc writes.


If you read about this type of solution on line, you will read about the life of flash and the fact that a particular cell can only be written to a finite number of times. This is certainly true...however, modern flash devices (both CF and USB keys) implement something called wear leveling (http://en.wikipedia.org/wiki/Wear_leveling). This prevents the same cell from being hammered and spreads the writes across all of the cells over time. So, if you have space for the wear levelling to work (i.e. you don't have the device 99% full), the device will last probably long past the life of the machine it is in.

With my 8GB flash key, I did some calculations based on it being half full and having constant log file updates (which is really a worst case senario - the drive is only 1/4 full and log file updates do NOT occur constantly) and I determined my drive would last 25+ years before it failed.

Good luck with your project - it sounds like fun and you will learn a lot.

maggior
2009-11-04, 12:34
Meanwhile, concerning the use of a different flash for the boot process, it looks as though this has the potential to free up a SATA socket which looks like a handy feature should I ever need a third slot. I presume that pointing the CPU at the USB drive to find the boot code is simple?

Yes, that's easy. You just need to make sure that the machine is capable of booting from a USB device. As far as I know, all modern BIOSs support this, but it's something to be on the lookout for.

With the IDE-CF or SATA-CF route, that wouldn't be an issue since it would look just like a hard disc.

To avoid confusion, when I install on a setup like this, I disconnect all drives except for the flash drive I'm installing on. After the fact, it's easy to mount the data drives. This will ensure that the OS installs entirely on the flash device.

(Wow, my 1000th post :-).

raven22
2009-11-05, 02:00
Why not make it yourself easy.
I bought a cheap nas like the DNS-323. It is 2 bay and there is an active forum and a simple software hack called funplug which opens the box and runs for example rsync and make a backup of one disk to the other automatically overnight. It is all documented very well, even for non linux techies. You could run SBS on the DNS but it is very slow so i recommend buying a Sheevaplug computer to run SBS.
With this you have a cheap, good performance system which uses in total only 10 watt idle (it allows your disks to spin down).