This is a complete "barebones" product.
As such ...
RAM is not included.
Use your choice of 1, 2, 3 or 4 GB, but 2 GB as 2 x 1 GB or 4 GB as 2 x 2
GB is recommended as this enables the dual channel function of the
Northbridge.
Hard drive is not included.
Use your choice of HD. I used 1 TB as that's what I had available, but 2
TB is also good and most probably 3 TB.
If you want internal RAID, then the optional 2 x 2.5" drive carrier is
required, and the drives would necessarily be limited to 750 GB each. Or 1
TB each, if you can find those.
Everything else is included in the product, including 19" rack mount ears
(standard) and slides (optional, but included with the product) or nothing
at all for desktop mounting (in which case 3M-type stick-on feet would be
required ... available at any hardware store).
This is the high-end Atom 330 server, the one with 2 x gigabit LANs and
the availability of an optional PCI-e "riser" card for external RAID or
"port multiplier" cases.
All you have to do is add RAM and a hard drive and in thirty minutes can
you have Server Lion running on your desktop or in your rack.
Software support is through my DSDTs and tonymacx86's UniBeast +
MultiBeast, but it is my DSDTs which give you the best MacOS X support (I
have not tried a non-DSDT installation, but that might work, too).
Use Lion and Server Lion applications from the Apple store.
Nothing else to buy.
I have previously posted the "Guide" to build one of these.
Software required includes an external instance of Snow Leopard or Lion
(either will work).
Software also required includes:
1) UniBeast (I used 1.0.0),
2) mach_kernel_atom.10.7.1 (this is the 10.7.1 kernel from the open
software repository which has been modified to support atom processors),
3) the DSDT for your desired RAM configuration (1, 2, 3 or 4 GB),
4) MultiBeast for Lion,
5) Lion.app (from the Apple store ... I don't recommend so-called
"torrents"), and, optionally,
6) Server.app (also from the Apple store; this turns an instance of Lion
into Server Lion).
For the techies out there, the only kexts in /Extra/Extensions is
fakesmc.kext and, optionally, IOAHCIBlockStorageInjector.kext.
The mach_kernel_atom.10.7.1 completely eliminates the need for all those
weird kexts which try to fool the system into believing it is something it
isn't.
The Supermicro Atom 330 will report itself as:
Hardware Overview:
Model Name: iMac
Model Identifier: iMac12,2
Processor Name: Intel Core 2 Duo <---- This is certainly a key factor
Processor Speed: 1.6 GHz <---- OK, its a SLOW Core 2 Duo, but it is H-T
Number of Processors: 1 <---- There are actually two processors, each
with one core
Total Number of Cores: 2 <---- Each core supports two threads, so four
total threads
L2 Cache (per Core): 512 KB
Memory: 4 GB
Bus Speed: 533 MHz
Boot ROM Version: MultiBeast.tonymacx86.com
SMC Version (system): 1.30f3
Did anyone get one of these?
How much dowry did you have to cough up?
LOL ... I would love to have had daughters. Hell, I would love to have had
sons.
Not to be, it seems.
--phubai
Thank you for your kind words.
I partitioned the 1 TB hard drive (no reason to buy a SATA III drive as
the mobo's SATA channels are SATA II) as follows:
1) Lion, 20 GB,
2) Lion Backup, also 20 GB, and
3) the entire remainder of the drive as Lion Data.
All were (and still are) HFS+ Journaled.
I did the install of Lion 10.7 using the UniBeast 1.0.0 technique, with
that flash drive being made on another system which had Snow Leopard. Of
course, the installation method was modified because of the need for
mach_kernel_atom.10.7.1.
I also applied the 10.7.2 Combo Update to Lion, using a DVD copy as the
source of the update. The update is HUGE, almost 1 GB.
Now, after Lion tested out in each and every respect, I did the following:
1) I cloned Lion to the Lion Backup partition, then
2) I applied the Server Lion update to the Lion Backup partition, and
3) I thoroughly tested the Server Lion system.
Only after all of the above, I finally cloned the Lion Backup partition
(which was really Server Lion) to the Lion partition, tested that and then
renamed all the partitions to Server Lion (from Lion).
I performed all of the above with the belief that 2 GB was the maximum
storage configuration.
However, after reading Intel's specification for the 945GC chip, I
realized that it was the only one of that family which supported 4 GB.
Of course, the NewEgg AND the Supermicro specifications both stated that
only 2 GB was supported.
Anyway, I pressed on and installed 2 x 2 GB = 4 GB, and the machine's BIOS
menus WOULD recognize 4 GB, but MacOS X would fail to boot.
It was then that I realized that the Supermicro's DSDT had hard-coded the
RAM size.
Several tests later, using Ubuntu, and I was able to calculate the values
for the DSDT which would enable any of 1 GB, 2 GB, 3 GB or 4 GB to be
installed in the Supermicro.
I am now permanently running 4 GB in my Supermicro, and I intend on
keeping it that way.
I will most likely be buying the PCI-e 8x "riser" card so that I can run
one or two eSATA drive cabinets, using "port multipliers".
In this way, I can have up to 8 x 3 GB = 24 GB of external storage.
http://www.tonymacx86.com/viewtopic.php?p=236227#p236227
... is the link to the complete scoop on the Supermicro Atom 330
Server's DSDT and links for downloading them (modified for MacOS X).
in reading the specs on the new egg site, am i right in assuming that
533 mhz memory is the best buy, since faster memory will be limited by
the system bus speed of 533 mhz?
thanks!
Eggs-cellent!
> in reading the specs on the new egg site, am i right in assuming that
> 533 mhz memory is the best buy, since faster memory will be limited by
> the system bus speed of 533 mhz?
The Intel Atoms, most, anyway, are designed around a 533 MHz front-side bus.
PC2-4200 is the specification of the two 1 GB sticks which I bought from a
P45 system for this project.
DDR2 800 MHz is the specification of the two 2 GB sticks which I borrowed.
By my reading of the specifications, this is PC2-6400.
No matter, PC2-4200 is 533-1/3 MHz and PC2-6400 is 800 MHz, and the 945GC
will tolerate sticks in any order and in any capacity.
You really should get two matched sticks, if possible, as this will ensure
dual-channel RAM support.
The PC2-4200 RAM was bought simply to "prove" the 1 GB, 2 x 1 GB = 2 GB
and 3 GB configurations, so I bought based upon price, including shipping.
Happily, the PC-4200 worked well with the PC-6400.
The Atom 330 is sort of tied to 533 MHz (really, 533-1/3), anyway.
Once your Supermicro is up on Lion or Server Lion, you should be happy
with it.
Two gigabit E-net LANs are provided so you could, theoretically, allocate
one for (external) Internet access, through a DSL or cable modem and one
for (internal) Intranet access, through a local router, and then use the
Supermicro for controlling access to your entire site.
Oops!
The PC-6400 (800 MHz) sticks were the ones I borrowed from a P45 for this
project.
The PC-4200 (533.33 MHz) sticks were the ones I bought from an eBay seller
...
http://www.ebay.com/itm/2GB-2x1GB-PC2-4200-DDR2-Desktop-RAM-/120815520970
... for this project.
I can see by his most recent listing that his price has increased by a
dollar or so.
No matter.
What he provided me was two "matched" sticks (identical in every respect),
and either or both worked perfectly in the subject Supermicro Atom 330
Server product.
Should you want 2 GB, then I would recommend this item.
But, as I have proved, 2 x 2 GB = 4 GB is fully supported by this product,
and that is what I am using, and that is what I would recommend.
I have NOT done any testing of the ICH7R beyond 1 TB drives, but I believe
it should be good to at least 2 TB, if not 3 TB.
Not withstanding the above comment, this Supermicro product will accept a
PCI-e "riser" card ...
http://www.newegg.com/Product/Product.aspx?Item=N82E16816101390
... which can accommodate a SATA RocketRAID JBOD/RAID enclosure and
"bundled" card (the card is SATA III but the enclosure is SATA I) ..
http://www.newegg.com/Product/Product.aspx?Item=N82E16816132034
... which is supported natively by MacOS X Lion, and which I am already
using on other systems (a Shuttle SH67 and aGigabyte P67).
The RocketRAID 622 card will support two external enclosures, each with up
to four drives, and that is what I have, and use, although not presently
on the Supermicro Atom 330 Server.
When NewEgg runs an EggBeater special on these subsystems, the price comes
down to $109.99 and the shipping becomes "free".
MacOS X Lion (and probably also Snow Leopard) will see the all the drives
as individual drives, should that be your choice, which is what I have
elected to do (all are MacOS X-compatible HFS+ journaled drives).
On Nov 20, 9:35 pm, peterh...@cruzio.com wrote:
> > PC2-4200 is the specification of the two 1 GB sticks which I bought from a
> > P45 system for this project.
>
> > DDR2 800 MHz is the specification of the two 2 GB sticks which I borrowed.
> > By my reading of the specifications, this is PC2-6400.
>
> Oops!
>
> The PC-6400 (800 MHz) sticks were the ones I borrowed from a P45 for this
> project.
>
> The PC-4200 (533.33 MHz) sticks were the ones I bought from an eBay seller
> ...
>
> http://www.ebay.com/itm/2GB-2x1GB-PC2-4200-DDR2-Desktop-RAM-/12081552...
Welcome!
The two dual eSATA bracket and cable kits which I ordered from HK arrived
this morning.
I ordered TWO as they are VERY inexpensive and I expected that perhaps one
might arrive have a kinked SATA cable, leading to possible intermittents.
This proved to be true. One of the two had a visibly kinked cable, but not
necessarily intermittent (I have not yet tested it, and I may not).
So, the one which was apparently good in every respect was installed in
the Supermicro Atom 330 Server, thereby giving me two eSATA connections,
which on the surface may seem to be laughable as I have not yet attached
the top cover, therefore the two related iSATAs, and an additional iSATA
are still accessible (I have not yet elected to go with the dual 2.5"
internal SATA drive carrier and with RAID, and I may not).
I am liking this system better and better.
For a firewall or a basic server, one really doesn't need LOTS of
computing power, but two 1.6 GHz cores, each of which has hyper-threading
and EM64T, and 2 GB of RAM and 1 TB of disk storage is nothing to laugh
at.
Then, given a clean and fully operational Server Lion install with ALL
software updates and Server tools, all the latest-and-greatest from Apple
Inc, what can one ask more of for a firewall and an e-mail server?
> I am liking this system better and better.
>
> For a firewall or a basic server, one really doesn't need LOTS of
> computing power, but two 1.6 GHz cores, each of which has hyper-threading
> and EM64T, and 2 GB of RAM and 1 TB of disk storage is nothing to laugh
> at.
>
> Then, given a clean and fully operational Server Lion install with ALL
> software updates and Server tools, all the latest-and-greatest from Apple
> Inc, what can one ask more of for a firewall and an e-mail server?
Exactly! It may be more than I'll ever need in fact. I'm impressed
with it's potential for sure.
I MEANT 4 GB, the maximum for the 945GC which this product has.
Other 945s, MOST, in fact, have a 2 GB maximum.
the bottom line is that after the install, booting from the HD
(without the usb stick inserted), and using verbose boot, the first
part of the boot (where there is a listing of the kexts being loaded)
completes, but then at the beginning of the second phase of the boot,
after 6-8 lines of text which flashes by too fast to read, the system
reboots.
i have tried to install on two different HD's, with identical results.
now to back up:
i followed Peter's guide (in the tonymac thread) to the letter.
i prepared the unibeast usb stick using my purchased copy of 10.7.2.
at the bios screen, i selected "default options", and then changed IDE
to AHCI and changed the boot order so USB booted first (i also
subsequently tried selecting "optimized defaults", made the same two
changes, with the same result).
booting from unibeast, i partitioned the SATA HD, one partition, GUID.
the install proceeded to normal completion.
upon reboot, i selected the Unibeast installer, and boot proceeded to
the normal "set up your mac" series of screens.
i placed the dsdt from the Unibeast USB on the desktop and ran
multibeast 4.1.0, userDSDT, system utilities, lnx2mac enet driver.
multibeast proceeded to normal completion. lnx2mac proceeded to
normal completion.
before restarting, i copied the atom kernel from unibeast USB to the
root of my HD. i replaced the boot.plist created by multibeast with
the same from Unibeast USB (plist edited per peter's instructions).
performed shutdown, removed unibeast USB, and restarted. as mentioned
at the beginning of this message, the first part of the boot (kext
loading) proceeded normally but system rebooted after about 6-8 lines
of the second phase of the boot.
performed shutdown, reinserted unibeast USB. upon restart, i chose
Unibeast, and system booted from the HD as expected.
i would like to inspect the boot log to see what those 6-8 lines say.
where can i find it?
any other suggestions?
thanks in advance!!
kem
On Nov 22, 2:00 pm, faithie999 <faithie...@hotmail.com> wrote:
> ok, i'm stumped.
>
> the bottom line is that after the install, booting from the HD
> (without the usb stick inserted), and using verbose boot, the first
> piart of the boot (where there is a listing of the kexts being loaded)
Something isn't right.
I remember that I had to pull the HD out of the Supermicro once or twice,
mount it in an eSATA/USB 2.0 dock and perform some magic on the
installation using another Lion system, most probably my P67.
I didn't have to do that with my P35, P45, P67 or H67 installations of
Lion, so methinks it has something to do with the kernel. OR, the boot
loader.
Additionally, originally I had copied the Atom kernel over to the USB
stick and to the HD as mach_kernel, and that proved to be troublesome.
The installation was a success, but Software Update... would not work.
Turns out that Software Update... checks mach_kernel for a match on the
particular update, and, for example, the update to 10.7.1 will be dimmed
(i.e., not selectable) if mach_kernel is really mach_kernel_atom.10.7.1.
So, I went back and renamed mach_kernel to mach_kernel_atom.10.7.1,
changed the appropriate plists, restored mach_kernel from the USB stick
and the updates went well.
10.7 [ AKA, 10.7.0 ] doesn't care if you are booting
mach_kernel_atom.10.7.1, and neither does 10.7.1 nor 10.7.2, but they DO
care if mach_kernel matches the candidate update.
Now that I think about it, I seem to recall that I had an issue with the
/Extra folder, and I pared it down to just fakesmc and
IOAHCIBlockStorageInjector, although the second one is not mandatory.
I am sorry you are having these difficulties.
While I endeavored to make my Guide as complete and as accurate as
possible, I DID trip over myself in the naming of the kernel and my
"mid-course corrections" in that regard may have made the Guide less
accurate than it could have and SHOULD have been.
With Lion, I have found that an unexpected reboot after all the kexts have
been listed has more to do with the boot loader itself than it does with
the installed system.
For the Supermicro Atom 330 Server project, I tried to do it tonymacx86's
way and use his latest Chimera and his preferred renaming of the plists,
and their contents.
I specifically recall that Kakewalk 4 was not satisfactory (it rebooted
exactly or almost exactly as you described), but that the
then-newly-released UniBeast 1 worked quite well for me.
It may well be that the easiest solution ... although not a general one
... for the Supermicro Server is to duplicate a known-good instance of
Lion, and then use showallfiles to allow manually modifying the duplicate
to accommodate the Atom 330.
In this specific respect, it is the /Extra folder which needs special
attention as that is where the DSDT and the plists reside. The kernel is,
of course, in the root of the file system.
Sorry to be so long-winded, but I am having difficulty in reconstructing
the comedy of errors (really, terrors) in building a fully working
Supermicro Atom 330 Server system on 10.7.2 with all updates and no
reliability or functionality issues, whatsoever.
2 GB, either as 2 x 1 GB (dual-channel) or as 1 x 2 GB (single-channel)
use the same DSDT, the one I originally released BEFORE I went hog-wild
and decided to "tickle the tail of the dragon" and try every possible RAM
configuration, including the two which BOTH NewEgg and Supermicro said
were impossible, Intel's 945GC specification not withstanding.
2 GB is the practical minimum for a Lion system, although 1 GB will work,
just not very well as the boot loader is thereby forced to reduce the size
of the RAM disk into which it loads the kexts.
It doesn't appear to have these issues with 2 GB, 3 GB or 4 GB.
i suspect a bootloader issue.
the process i am now using for a successful boot is to have the
unibeast USB installed, and BIOS set so that USB boots first.
then when the chameleon screen appears with unibeast on the left
(since it has boot priority) and my HD on the right, i click on the HD
to highlight it, then press enter to start the boot.
the HD activity light doesn't glow immediately, leading me to believe
there is something being accessed from the USB stick (i don't have an
activity light on that particular USB stick), and then after several
seconds, the HD activity light begins flashing, as the boot is taking
place and proceeds successfully.
it's not the end of the world to need to have the unibeast USB
installed to boot, since this box will be a server and will presumably
be turned on 24/7.
i must also point out, since i know you're a big fan of sleep (ha
ha!!) that with the addition of SleepEnabler_10_7_1.kext installed in
s/l/e, sleep works perfectly!!
just for fun, i will try a new install using my preferred method--
using rboot (from tonymac) and a vanilla lion 10.7.2 usb installer
stick prepared the "normal' way from my purchased lion 10.7.2
download. i'll report on that when i do it.
thanks!
THIS is where npci=0x2000 and the PCIRoot flags can be helpful.
i didn't want to always have a usb stick tied up as a boot helper, but
i didn't want to spend hours/days figuring out why my lion install
wouldn't boot from the HD without the assistance of the Unibeast usb
stick.
i partitioned my HD with a 6 gb first partition and the rest of the
capacity as the second partition.
i used my unibeast usb stick to install lion on the second partition.
then i removed the HD from the supermicro and attached it via usb to
my hackbook. i created a unibeast installer on the first (6gb)
partition.
then i installed the HD back into the supermicro. upon booting, when
the chameleon screen appears, i select my lion installation and it
boots properly.
now i will enter the world of learning about home servers and what
interesting things i can do with one!
well, i've thrown in the towel (sort of).
i didn't want to always have a usb stick tied up as a boot helper, but
i didn't want to spend hours/days figuring out why my lion install
wouldn't boot from the HD without the assistance of the Unibeast usb
stick.
i partitioned my HD with a 6 gb first partition and the rest of the
capacity as the second partition.
i used my unibeast usb stick to install lion on the second partition.
then i removed the HD from the supermicro and attached it via usb to
my hackbook. i created a unibeast installer on the first (6gb)
partition.
then i installed the HD back into the supermicro. upon booting, when
the chameleon screen appears, i select my lion installation and it
boots properly.
now i will enter the world of learning about home servers and what
interesting things i can do with one!
> i didn't want to always have a usb stick tied up as a boot helper, but
> i didn't want to spend hours/days figuring out why my lion install
> wouldn't boot from the HD without the assistance of the Unibeast usb
> stick.
You changed your BIOS to select the HD 1st instead of the USB?
You're probably using Chimera, but I'd recommend trying the latest
Chameleon using Chameleon Wizard as an installer.
> one thing I've learned is that the 1st partition should always be
> your boot partition.
I agree. And if you screw up and install a bootloader to two different
partitions it can be a problem.
My installation of Lion on my Supermicro was two 20 GB partitions followed
by the entire remainder of the drive as a common data partition. All were
HFS+ Journaled.
The first non-EFI partition should be your primary MacOS X boot partition.
The second and third non-EFI partitions may be Windows and FAT (shared
Windows/MacOS X data partition), respectively, should you want to try dual
booting.
The next partition should be your secondary MacOS X boot partition.
You may have additional MacOS X partitions, say, for testing.
The last partition should be your MacOS X-only data partition.
I think 30 GB is a better size than 20 GB for the two MacOS X partitions,
but that is 20-20 hindsight.
As I reported, I partitioned my Supermicro 1 TB hard drive as 20 GB, 20 GB
and the entire remainder as HFS+ journaled.
I installed Lion on the first partition.
I screwed-up on the naming of the kernel (reported in greater detail
earlier) so I elected to pull the HD and use an existing Lion
installation, most probably my P67, to fix it.
I MAY have installed the boot loader at this point, under the P67, but I'm
not sure. I was not taking notes at this point, but I should have been in
retrospect.
I reinstalled the HD in the Supermicro and it booted OK.
I CCC-ed the Lion boot partition to the second, backup partition.
I applied the 10.7.2 Combo Update from a DVD.
I installed Server Lion, probably from a DVD, but it is small enough, 26.5
MB, that I may have used a USB flash drive.
I tested the Server Lion update and it was fine.
I CCC-ed the Server Lion to the second, backup partition and checked it
out by booting it.
As everything checked out, I renamed the Lion and Lion Backup partitions
to be Server Lion and Server Lion Backup. I also renamed the Lion Data
partition to be Server Lion Data.
I am sure I had to pull the Supermicro's hard drive at least once in order
to fix the kernel name problem. I may have installed Chimera at that time,
or I may have done it at another time, requiring the hard drive to be
pulled a second time.
It was only after the Supermicro was working 1000 percent (sic) that I
screwed it down using those special, very small-headed black screws which
Supermicro provides for the purpose.
This, then, is a reconstruction, and is probably what I most likely did.
Update ...
The dual iSATA to eSATA cables arrived the other day from HK.
One was installed in the Supermicro with the SATA port allocation as follows:
PRT0 - 1 TB boot drive
PRT1 - reserved for possible internal RAID (would require the dual 2.5"
drive carrier accessory kit)
PRT2 - eSATA
PRT3 - eSATA
I have a "watch" set on the PCI-e 4x "riser" accessory, just in case
NewEgg comes down on the shipping charges. I would then use this "riser"
for a RocketRAID 622 PCI-e 1x JBOD or RAID card.
NewEgg now lists it as "out of stock".
But not "deactivated".
Meaning, it may come back in stock at a later date.
again, i decided to do this because after installing lion using
unibeast, and running multibeast after install, the thing wouldn't
boot. it crashed/reset just into the "second phase" of the boot
process (after loading the kexts). it would only boot if i first
booted using the unibeast usb, then selecting my lion install at the
chameleon screen
i will use chameleon wizard to experiment with other versions of the
bootloader.
when i have some more time, i will do another clean install, not using
multibeast, and create my own /extra folder, install chameleon, and
see what happens. however, today the priority is cooking!
happy thanksgiving to all
ken
happy thanksgiving to all
ken
again, i decided to do this because after installing lion using
unibeast, and running multibeast after install, the thing wouldn't
boot. it crashed/reset just into the "second phase" of the boot
process (after loading the kexts). it would only boot if i first
booted using the unibeast usb, then selecting my lion install at the
chameleon screen
i started with a clean install--using the unibeast usb stick prepared
with my purchased 10.7.2 download.
i formatted with a single partition (GUID). after installing, but
before rebooting, i ran multibeast (user dsdt, using the correct dsdt
from peter) copied the atom kernel to the root of the lion partition
and copied the plist (modified per peter's instructions) from the
installer stick to the lion extra folder.
booted verbose--the "first part" of the boot (where all the kexts are
loaded) proceeded normally. then when the "second" part begins (the
screen font changes to a slightly larger size), about 6 lines quickly
scroll by and the the system reboots.
is the boot information stored in a log somewhere, so i could find out
what those 6 lines say?
my decision to install unibeast on a small (first) partition and then
install lion to the second partition is merely a workaround till i
(we!) get this figured out. while it is klugey, it is working fine
for now.
is the boot information stored in a log somewhere, so i could find out
what those 6 lines say?
booted verbose--the "first part" of the boot (where all the kexts are
loaded) proceeded normally. then when the "second" part begins (the
screen font changes to a slightly larger size), about 6 lines quickly
scroll by and the the system reboots.
when i did the original install (single partition) i copied the
boot.plist from the unibeast stick to the lion extra folder, and then
the unsuccessful boot followed. maybe there's a hidden file in the
extra folder of the unibeast stick that isn't present in the lion
install. when i do another clean install on a different HD (after
dinner; i'm the cook today) i will copy the entire extra folder (which
contains the plist as edited per peter's instructions) to the new lion
install, rather than copying just the plist.
This is precisely on-point.
Recently, I mucked about on my EM64T, but SSE3 Shuttle P35 (the proc is a
Pentium 4 541, 3.2 GHz with H-T).
Partition 1 (assuming the EFI partition is 0) was Leopard SSE4
Partition 2 was Snow Leopard SSE3
Partition 3 was Leopard SSE4
Partition 4 was Snow Leopard SSE3
As long as I left Partition 1 alone, I could boot Snow Leopard SSE3 from
either Partition 2 or Partition 4
I got a little too cute and erased the system from Partition 1.
Now nothing would boot.
I pulled the hard drive and CCC-ed Partition 2 to Partition 1.
Reinstalled the hard drive and it still would not boot.
I pulled the hard drive again and installed Chimera 1.4.1 on Partition 1.
Reinstalled the hard drive and now any partition which has MacOS X
installed, which is now only SSE3 versions of Snow Leopard, will boot.
Alas, Partition 1 is "special" and it MUST have a bootable MacOS X system
on it otherwise none of the other partitions are likely to be bootable.
The best course of action (in my now very humble opinion) is this:
1) make Partition 1 twice as large as you want it to be,
2) make the entire remainder of the hard drive a data partition to be
shared amongst MacOS X partitions,
3) install MacOS X on Partition 1 and ensure that it is working properly,
4) go back and "resize" Partition 1 to about one-half of its size,
5) "add" Partition 2 using the area freed-up by the "resize" (the number
of the data partition will change to 3),
6) CCC Partition 1 to Partition 2.
Here are some SPECIAL conditions for dual booting with Widows 7.
1) make Partition 1 the size you want it to be (not twice the size),
2) make Partition 2 the size you want Windows 7 to be (NTFS),
3) make Partition 3 the size you want your shared Windows 7/MacOS X data
partition to be (FAT),
4) make Partition 4 the same size as Partition 1,
5) make Partition 5 the entire remainder of the drive and use this for
shared MacOS X data only.
The reason for this is Widows 7 and anything which Windows 7 will see MUST
have partition numbers which are below 5 and the actual Windows 7
partition takes two partition numbers, one is apparently for the swap
partition.
As MacOS X is usually the primary OS, anyway, and as the first partition
MUST be MacOS X for the boot loader to work properly, your backup MacOS X
partition(s) MUST be AFTER the Windows partitions.
If you try to be cute and place the shared Windows 7/MacOS X data
partition after your backup MacOS X partition, Windows will not see it as
it has a partition number which exceeds 5.
AND, in the case of the Supermicro Atom 330 Server, the kernel is a
special one.
It is named mach_kernel_atom.10.7.1
However, you can get into all kinds of trouble by renaming it mach_kernel,
the default kernel name for MacOS X.
Yes, the system might boot, but then again it might not, particularly
after attempting a Software Update...
Software Update... attempts to prevent this sort of trouble by matching
the kernel's data to the data within the update itself. No match and the
update will not be allowed.
As long as mach_kernel remains whatever version of mach_kernel matches the
version of the OS, it is perfectly fine to have a mach_kernel_atom.10.7.1
around, which Software Update... will never check, but which the bootstrap
loader will always go to as the kernel flags point to it.
I have exactly the same issue on my Pentium 4 541 system. In this case,
mach_kernel is still there, but the bootstrap loader always loads
mach_kernel_non-atom, which is a special kernel for SSE3 but not SSSE3
Intel processors, and then only for Snow Leopard.
It doesn't matter if the kernel is hidden, or not. Normally, mach_kernel
IS hidden.
However, you can get into all kinds of trouble by renaming it mach_kernel,
the default kernel name for MacOS X.
thanks
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN"
"http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
<key>Kernel</key>
<string>mach_kernel_atom.10.7.1</string>
<key>Kernel Flags</key>
<string>arch=i386 npci=0x2000 PCIRootUID=1</string>
<key>GraphicsEnabler</key>
<string>No</string>
<key>Instant Menu</key>
<string>Yes</string>
<key>Legacy Logo</key>
<string>Yes</string>
<key>EthernetBuiltIn</key>
<string>Yes</string>
<key>GenerateCStates</key>
<string>Yes</string>
<key>GeneratePStates</key>
<string>Yes</string>
</dict>
</plist>
And, for good measure:
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN"
"http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
<key>SMfamily</key>
<string>iMac</string>
<key>SMproductname</key>
<string>iMac12,2</string>
<key>SMboardproduct</key>
<string>xxxxxxxxxxxxxxxxxxxx</string>
<key>SMserial</key>
<string>xxxxxxxxxxxx</string>
<key>SMbiosversion</key>
<string>MultiBeast.tonymacx86.com</string>
</dict>
</plist>
copied and pasted both boot and smbios.plist, same result (reboot)
i thought i could learn something from the system.log or kernel.log
after the unsuccessful boot. i removed the HD, attached to my
hackbook, looked in /var/log/system.log and /var/log/kernel.log but
there were no entries for the most recent boot. not sure why.
i'm just about out of ideas. the last thing i can think of is to ask
you what bootloader and version you're using?
Whichever one was provided with the release of MultiBeast Lion Edition ...
it was 4.0.3.
Here's how my Supermicro BIOS is setup:
Boot Features --> HPET Enabled
IDE Configuration --> Configure as AHCI, SATA before PATA
Super IO --> serial ports are disabled
System Memory --> Populated size 4096 MB, Available size 3576 MB
Northbridge Configuration --> Video Function Select --> 128 MB
I used Chameleon Wizard to read what it thought was my boot configuration.
It said the boot loader was Chimera 1.5.4 r1394.
I thought I had deleted the PATA device from the DSDT, but I didn't.
For MacOS X, HPET and AHCI is pretty much mandatory.
I suppose video could be reduced to 64 MB.
Here's some additional information which could be helpful.
Using showallfiles.app and selecting SHOW, the following boot files are
found on the first Lion partition:
boot (not hidden, but probably should be)
mach_kernel (hidden, and presumed to be 11.2)
mach_kernel_atom.10.7.1 (not hidden and known to be 11.1 with changes for
atom)
The same is true of the second Lion partition, which was CCC-ed from the
first.
Either partition will boot as specified using the <- or -> keys on the
keyboard.
I suspect, but do not know for sure, that there is another boot file in
the EFI partition of the Supermicro's hard drive. I suspect that your
Supermicro hard drive may be missing that boot file.
I also suspect that your system, which is requiring the USB flash drive to
be present, is using the boot file which is in the EFI partition of the
flash drive as the first phase of the boot loading process, and only after
you select the target partition is control being transferred over to the
boot file on that hard drive partition.
http://www.newegg.com/Product/Product.aspx?Item=N82E16816101262 /
http://www.newegg.com/Product/Product.aspx?Item=N82E16816101262R
... has now been "deactivated", meaning it may NEVER return.
No matter. It is most probably available from others.
Sad, as this was perhaps the very lowest cost true server product
which was 100-percent compatible with Server Lion.
how would i go about inspecting and "fixing" the EFI partition?
thanks in advance
sounds like a promising lead.
how would i go about inspecting and "fixing" the EFI partition?
thanks in advance
sounds like a promising lead.
how would i go about inspecting and "fixing" the EFI partition?
thanks in advance
UpdateEFI2, a completely new utility, is available ...
http://www.insanelywind.com/forum/download/file.php?id=1619
... here.
i won't bore you with my umpteen unsuccessful troubleshooting
attempts.
i finally decided to copy the Extra folder in its entirety from my
Unibeast USB stick to my fresh Lion HD install, replacing the Extra
folder created by multibeast. it booted fine.
to determine what made the difference, i unhid the contents of the
relocated extra folder (unibeast sets it to Hidden when it prepares
the usb stick) and decided to start removing the extensions one at a
time. my first try was to remove the nullcpupowermanagement.kext.
and, that was the key! without it, the boot exhibited the same
troublesome behavior--crashed back to reboot just after the "second
phase" of booting started.
i don't know if this is optimal, but i think it will get the job done.
thanks for all the suggestions along the way!!
ken
thanks!
On Nov 26, 8:42 pm, peterh...@cruzio.com wrote:
These are SUPPOSED to "mount" the EFI partition as something which you can
inspect and/or modify.
I HAVE NOT been able to get either to do that.
Glad to hear that you've finally got your Supermicro up and booting
without needing special help.
UniBeast usually builds an /Extra folder with a lot more kexts than a real
mobo is likely to need.
Here is what UniBeast built for one of my installs (NOT the Supermicro):
/Extra/Extensions ...
AHCI_3rdParty_SATA
EvOreboot
FakeSMC
IONetworkingFamily
IOUSBFamily
IOUSBMassStorageClass
JMicron36xATA
NullCPUPowerManagement
/Extra ...
org.chameleon.Boot
smbios
/Themes
dsdt
Here is what I pared my Supermicro's /Extra folder down to:
/Extra/Extensions ...
FakeSMC
IOAHCIBlockStorageInjector
NullCPUPowerManagement
The remainder of the /Extra folder is pretty much as above.
Yes, is is a good idea to hand-build the contents of the /Extra folder of
UniBeast and then to copy it over to the installed system (UniBeast may
either copy over an abbreviated /Extra folder or it may fail to copy
anything within it).
For a system, such as the Supermicro Atom 300 Server, which needs a
special DSDT, you know you are going to have to add the dsdt.aml file
within /Extra and at least FakeSMC and NullCPUPowerManagement.
IOAHCIBlockStorageInjector is a kext which simply causes the hard drive
icons to be displayed as silver rather than as yellow.
In particular, for a system which needs a legacy kernel, you will have to
manually copy the kernel into the root of the UniBeast USB flash drive and
the installed file system, too. You MUST leave mach_kernel alone as it is
used to check during various phases of the update and/or installation. In
the beginnings of my experiences with the Supermicro, I renamed
mach_kernel_atom.10.7.1 to mach_kernel, and that got me into a bootable
system alright, but none of the Software Updates... would apply. Renaming
mach_kernel back to mach_kernel_atom.10.7.1 and copying over the real
mach_kernel, and updating org.chameleon.Boot to refer to
mach_kernel_atom.10.7.1 got the Software Updates... to apply correctly,
both the 10.7.2 Combo Update, which is just over a GB, and the
subsequently found updates, such as the Brother printer update, and
others.
I believe most of my above observations were discussed when you first
reported having problem with your Supermicro.
what caused me all the head-knocking-against-wall was that i used the
"userdsdt" option in multibeast 4.0.3, which only installs one kext:
fakesmc.kext.
"easydsdt" option installs many other kexts, including
nullcpupowermanagement.kext.
since i had a your dsdt i chose the userdsdt option, never giving
thought to kexts i might need to add.
do you think there might be a dsdt edit so that
nullcpupowermanagement.kext isn't required?
thanks for all your hard work in creating the dsdt's and guide, as
well as the tip on the availability of the supermicro on newegg.
ken
In my experience, not all mobos can get by without NullCPUPowerManagement.
My Gigabyte P67 can. It has a very late Award BIOS DSDT, modified by
tonymacx86 for MacOS X compatibilty, and further modified by me for
further improved MacOS X compatibility.
My Shuttle H67 cannot. It has a very late AMI BIOS DSDT, modified by me
for MacOS X compatibility.
My Shuttle H67 came just before my Supermicro, and the Supermicro also has
an AMI BIOS DSDT.
I suppose what happened was I utilized the "lessons learned" on several
previous AMI BIOS DSDTs (a number of ASRocks and several late Shuttles)
and came up with my AMI BIOS DSDT Hacking Guide, which I originally
published on tonymacx86's site, and after users had reported that they
were being "blocked" from downloading my Guide, I brought it over to H
Q-A.
It has not substantially changed, and I now pretty much do new DSDTs from
memory, without referencing the AMI BIOS DSDT Hacking Guide.
There are FAR FEWER things to worry about on an AMI BIOS DSDT as Award
made some fairly significant mistakes on their DSDTs, mistakes which
allowed Leopard to work, but Snow Leopard would always fail. And, of
course, if Snow Leopard fails, Lion will also fail.
(In fact, Leopard often can work without a DSDT at all ... MacOS X will
simply read out the BIOS's DSDT and use it as-is).
But, Snow Leopard (all versions, 0 through 8) and Lion (at least versions
0, 1 and 2, but version 3 will soon be coming out) require a squeaky-clean
DSDT, particularly where the RTC is concerned.
You will know your DSDT is wrong on Snow Leopard and Lion if you get a
"CMOS Has Been Reset" error when you shut-down MacOS X and you are forced
to accept the CMOS defaults, which WON'T allow MacOS X to boot OR you
elect to manually re-enter all those defaults, which WILL allow MacOS X to
boot.
This is because the length of a variable was set to 0x04 rather than 0x02,
and Snow Leopard and Lion are using the two bytes which follow the 0x02
for their own purposes as the amount of NV-RAM (namely, CMOS) is quite
limited on PCs and Snow Leopard and Lion need lots of NV-RAM to store its
parameters across a reset or a power-off. Probably across a sleep as well.
AMI did not make this mistake on its DSDTs!
But, AMI needs lots of help in other areas as it produces DSDTs which
cannot be compiled using our old-faithful DSDTSE DSDT compiler (the
"professional" DSDT tools which Intel and Microsoft license to its
motherboard manufacturer clients are much more forgiving).
Typical AMI BIOS DSDT errors include:
1) _T_0 and related errors, as a leading underscore indicates a "reserved
word",
2) the usual error that almost everyone makes on _WAK and also Local0,
Local0 where the first Local0 has to be replaced by Zero,
3) SBUS device is missing or named incorrectly,
4) LPCB is named incorrectly (this is the "low pin count bus" chip which
every PC mobo has and which maps all the old devices from the PC/AT days
into a single chip which can ride on the Southbridge and provide floppy
and UART/USART support with a minimum of pins),
5) and also procedures which return values for ALMOST every eventuality,
but which DO NOT return a default value (as the default will never be
encountered, anyway, it is a simple matter to parse the entire decision
tree of the offending procedure and stick in a Return (Zero) statement
which thereby establishes the default return value).
Anyway, I am getting way off the point, so I will try and wrap this up.
No, I have NOT found a way to eliminate NullCPUPowerManagement.
It is either required or it is not.
If it IS NOT required, then including it doesn't get you into trouble.
However, if it IS required, then failure to include it will get you a KP
or worse.
A final remark.
My Shuttle H67 is generally a good machine, notwithstanding the fact that
its two ASMedia USB 3.0 chips (two rear ports, two front panel ports) are
completely useless as there is no USB 3.0 driver, but these ports work
fine on Windows 7, so that's a small mercy of some kind, and also the fact
that I cannot get ALC8xxHDA to work on it for the life of me.
Also, it requires NullCPUPowerManagement, whereas my Gigabyte P67 does not.
But, then, the Gigabyte is an Award BIOS DSDT whereas the Shuttle H67 is
an AMI BIOS DSDT.
Go figure!
What geekbench scores are you getting from these machines?
-L
I have obtained a Lion/Server Lion textbook from Amazon and I am busily
perusing that book.
My goal is an e-mail server and a DNS, both using only Server Lion, which
I can use to replace my Internet service provider.
It's not that I am dissatisfied with them, far from it, but I want more
control. I will still use a static IP which is provided by them. What I
will be saving is a minimum of US$10.99 per month, a minimum of US$132,
for hosting and e-mail, plus I will obviously be able to provide other
services, too.
Also, I have paid-for, and have owned my own TLD (basically, my name,
which some poor fool let lapse, and which I snapped-up a few microseconds
after it had lapsed), for many years, and it seems now it the time to host
my own Internet presence.
The Supermicro Atom 330 Server, with 4 GB of RAM and 1 TB of storage,
should be more than I really need.
The 4 GB of RAM was the basic issue for me, and with luck I was able to
overcome the stated 2 GB (by both NewEgg and Supermicro) RAM limit.
I presently have 1 TB of disk storage, but I am tempted to change to 1.5 TB.
5 to 10 minutes, maximum.
It is slow because USB flash drives can be very slow.
I have some which boot very fast, some which boot very slow.
Once the hard drive has been created and made bootable, things go pretty
fast.
I really appreciate everyone's help and all of the posts have helped
in some way. I may not give up on multibeast, but I will spend a lot
of time trying to understand more of how it works and what it's doing.
Perhaps selecting easybeast and letting the others alone would be a
start.
--phubai
On Nov 27, 4:23 pm, faithie999 <faithie...@hotmail.com> wrote:
> i think i have figured it out (for my system, at least!)
>
Also, I didn't install the Lnx2Mac gigabit E-net driver as Peter
suggested. I had ethernet connectivity and thought I'd do it later if
necessary. What would be the advantage of using these drivers as
opposed to what I have?
Thanks all!
--phubai
On Nov 30, 9:05 am, phubai <rjb...@gmail.com> wrote:
> And this may be my method of choice as well. I don't have a warm
> feeling regarding multibeast right now, nor does it me I'm afraid.
>
Congrats!
> Also, I didn't install the Lnx2Mac gigabit E-net driver as Peter
> suggested. I had ethernet connectivity and thought I'd do it later if
> necessary. What would be the advantage of using these drivers as
> opposed to what I have?
The Lnx2Mac driver covers more Realtek LAN chips than the regular one does.
If your LAN is working, no need for complications.
I'd left it on all afternoon to see what would come of it, and made
the mistake of not changing my Energy Saver settings, so of course, it
went to sleep. Nothing I pressed, stared at (or rubbed in desperation)
would revive it but a hard reset. I was angry at myself for forgetting
that simple preference setting, but I was rewarded by the desktop
again when I rebooted. Based upon your experience with yours, I have a
good feeling about this little machine.
Next up...Lion Server purchase! Did you have to do anything special in
preparation? Or after the upgrade? I will check the Extra folder of
course to make sure the org.chameleon.Boot.plist includes everything
that it should, that mach_kernel_atom.10.7.1 is still in root
directory, and that my Extensions folder is still complete, and then
reboot into server. I've some reading to do as well, and hope you'll
let us know how your project is moving along. I don't want to pay
extra for a static ip address, or I'd try hosting my site locally as
well. Have fun with it, and thanks!
--phubai
On Nov 30, 1:22 pm, peterh...@cruzio.com wrote:
>
> Congrats!
Nothing special, except you will need to have a reliable Ethernet
connection as the application which installs Server Lion is only 27 MB and
lots of extras are automagically downloaded over the 'Net.
Also, you should download Server Admin Tools, a freebee from Apple, not
from the Apple Store.
Server Lion goes right on top of Lion, not replacing it as it did on
earlier server versions.
Therefore, the kernel and plist and all that stuff remain the same.
All you really have to do is make sure Lion is reliable and so also is
your 'Net connection.
You are absolutely correct, once you successfully apply Server Lion, it
will appear as such when it reboots.