• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Largest, most baddest file server I ever built

A

AtomicFire

[H]ard|Gawd
Joined
Dec 7, 2001
Messages
1,214
IMG_1630.jpg

IMG_1631.jpg

IMG_1632.jpg

IMG_1633.jpg

IMG_1634.jpg

IMG_1636.jpg

IMG_1637.jpg

IMG_1638.jpg

IMG_1639.jpg

IMG_1641.jpg

IMG_1642.jpg

IMG_1643.jpg

IMG_1644.jpg

IMG_1645.jpg

IMG_1664.jpg

IMG_1665.jpg

IMG_1666.jpg

8xSATA Hot swap hard drives :)

IMG_1667.jpg

IMG_1668.jpg

Hot swap fans :)

IMG_1669.jpg


1.4 TB of highly redundant storage. Built for a local radio station to archive all their music, designed to be on 24/7/365 The budget didn't allow for redundant power supplies, but that will be added on later.

Motherboard: Supermicro Super P4SPA+ w/o SCSI option
RAID Controller: Areca ARC-1120 PCI-X
Hard Drives: 8xSeagate 200gb 7200.9 SATA NCQ + 2 spare
System Drive: 1xSeagate 80gb 7200.7 SATA
CPU: P4 2.8E
Case: Supermicro SC743T-650

Hard drives get to a max of 31C in a room ambient of 24.4C

:D :)
 
drizzt81 said:
only one stick of RAM? that is not redudant ;)
Click to expand...
despite the smiley, QFT. I've had many a server die for this reason alone.
 
svet-am said:
despite the smiley, QFT. I've had many a server die for this reason alone.
Click to expand...
I'd somehow expect that even with multiple sticks of RAM if one goes bad the server will go down. Am I incorrect on that? Two sticks certainly would make rebooting quicker, since the crappy one need to be yanked, no waiting for parts. Same with the lack of dual-cpus :)

then again, it looks like your client was on a 'budget'.

ps.: we need specs :D
 
drizzt81 said:
I'd somehow expect that even with multiple sticks of RAM if one goes bad the server will go down. Am I incorrect on that? Two sticks certainly would make rebooting quicker, since the crappy one need to be yanked, no waiting for parts. Same with the lack of dual-cpus :)

then again, it looks like your client was on a 'budget'.

ps.: we need specs :D
Click to expand...

You're right, the client was on a budget. One stick of ram for now because the actual order of two sticks of 1gb ECC was lost somewhere, we're working with Newegg to sort that out. They also threw the hard drives into the box without any support, just peanuts so one of the two spare hard drives needs to be sent back. We eventually decided that SMP was un-needed because of the cost, and ebcause everywhere I read if one CPU died, the system will still crash.

Specs will be posted soon, gathering parts right now and finishing an install of Fedora on it.
 
Yummy. I love backplanes.

Atomic - I couldn't see the drives, maybe I didn't look close enough. What kind of drives did you use? How has your experience with the Areca been so far? I'm kind of in the market right now, and I'm interested in an Areca. Did you decide to go with RAID 6? Just curious. Sorry for all the questions.
 
Mmmmmmm mmm.. mm mmmm m mm mmm mmMMM!MMMMMMMMM! :cool:

Just be wary that the system may be slightly unstable if a drive goes bad until you activel replace it. Dunno if its just that way with raid5 or if raid6 solves it... but drives arn't automatically removed that fail smart iirc.

These things are DAMN fast though!
 
drizzt81 said:
I'd somehow expect that even with multiple sticks of RAM if one goes bad the server will go down. Am I incorrect on that?
Click to expand...
Not if you're using properly specc'ed ECC RAM (which I will assume was used in a mission-critical box like this). With ECC, if a bank of memory starts to fail the CRC check, then the firmware knows not to attempt writes into that bank and will most likely just disable that whole channel (stick). In some cases, I've seen firmware slick enough to shut down access to only the offending bank. In the absolute worst case (not properly ECC specc'ed, but ECC RAM nonetheless), then the CRC would constantly fail and memory operations would just become very slow because it would attempt an operation and the operation would fail and the firmware would have to keep retrying the operation until it found a bank that works. Slow, yes -- but it would technically keep the server operational until an admin took notice.

If you're using standard off-the-shelf SDRAM, then a memory failure is probably just going to bring the machine to its knees. Essentially, an operation would be attempted, would fail, and then the BIOS would go 'Oh Shit!' and proceed to lock up the machine. If it doesn't lock up the machine, then you'd see issues with data integrity and the like.
 
svet-am said:
Not if you're using properly specc'ed ECC RAM (which I will assume was used in a mission-critical box like this). With ECC, if a bank of memory starts to fail the CRC check, then the firmware knows not to attempt writes into that bank and will most likely just disable that whole channel (stick). In some cases, I've seen firmware slick enough to shut down access to only the offending bank. In the absolute worst case (not properly ECC specc'ed, but ECC RAM nonetheless), then the CRC would constantly fail and memory operations would just become very slow because it would attempt an operation and the operation would fail and the firmware would have to keep retrying the operation until it found a bank that works. Slow, yes -- but it would technically keep the server operational until an admin took notice.

If you're using standard off-the-shelf SDRAM, then a memory failure is probably just going to bring the machine to its knees. Essentially, an operation would be attempted, would fail, and then the BIOS would go 'Oh Shit!' and proceed to lock up the machine. If it doesn't lock up the machine, then you'd see issues with data integrity and the like.
Click to expand...
Useful info!

I see Supermicro and Areca. Two brands I am slowly beginning to love. Cool...
 
I do have to say that one the thing that bothers me from all of the pics is that not one single person is wearing an anti-static strap.

<rant>

ISO 9001 certifications is a big deal, espescially when dealing with other peoples' gear. I'm a hardware design engineer (I design PCBs for a small company). If any of our folk were to ever step into the lab and handle our gear without an anti-static strap, I'd shoot them in their frickin' heads. MOS devices are very sensitive to static discharge. And, in today's computer world it's not a "Zap! It's Dead!" thing. Damage from static electricity can (And usually is) cumulative in nature. i.e., a single zap does nothing at the moment, but it will begin to degrade the lifespan of the device over time. This is espescially true of gear (most expansion cards and hard disks) that are repeatedly installed and removed over their life span.

Please, espescially if you're a computer professional (or plan to be one after college), get in the habit of wearing the strap. It may look ghey, but it will save you and your employer a ton of money. It costs a lot of money to become ISO9001 certified and a *single* violation and a single infraction will cause your company (and every single company higher than yours in the supply chain) to instantly lose that certification. That's how serious this issue is...

</rant>
 
So you are saying, if you for example work in a firm assembling computers and you touch a graphics card and it zaps and gets destroyed by static electricity, your company and every company above you loose their iso certification?
 
nice setup.
What model case is that and how much was it?
Also, how warm do the 8 drives in the cages get?
 
cv643d said:
So you are saying, if you for example work in a firm assembling computers and you touch a graphics card and it zaps and gets destroyed by static electricity, your company and every company above you loose their iso certification?
Click to expand...
In the SUPPLY chain. For example, I work at a company where we design PCBs. Now assume that we get our power transistors from company Y in China. We use Company Y transistors exclusively in our designs.

Next, assume that all companies in the supply chain are ISO 9001 certified. If some dumbass at Company Y over in china decides not to wear his anti-static device and gets *caught*, suddenly the certification board revokes their certification. Then, they look at the folks above them in the supply chain (that would be us, since we use their part). Suddenly, WE have lost our certification.... damn! The rationale for this is that the certification boards HAVE to be able to certify to consumers and government agencies that a device will be electrically 'clean' from inception to delivery. Without having a bottom-to-top certification dependency, they cannot make this assurance.

Yes, this is very real and we deal with it all of the time in our industry. We have to be extra careful because we're just doing PCB design and construction. We're still not at the top of the food chain (that would be a company like Dell or Gateway -- system ISVs).

So, if you work at a local mom and pop store assembling computers with off-the-shelf components, then YOU are the top of the food chain. However, if your mom and pop is ISO 9001 certified (not required for a store like this, but a nice selling point to consumers) then yes, you could easily lose your certification over an infraction like that.

Note that the reason this is different for a local mom and pop is because they are the last folks before handoff to the consumer. If you're anywhere in the middle, then you've got to take this seriously.
 
Motherboard: Supermicro Super P4SPA+ w/o SCSI option
RAID Controller: Areca ARC-1120 PCI-X
Hard Drives: 8xSeagate 200gb 7200.9 SATA NCQ + 2 spare
System Drive: 1xSeagate 80gb 7200.7 SATA
CPU: P4 2.8E
Case: Supermicro SC743T-650

Hard drives get to a max of 31C in a room ambient of 24.4C

About the ISO 9001 cert and wrist bands - yes I know I should have them, but this time I didn't have my strap with me. It was constructed on a wooden table, and I made sure I grounded myself before handling the components.
 
AtomicFire said:
Motherboard: Supermicro Super P4SPA+ w/o SCSI option
RAID Controller: Areca ARC-1120 PCI-X
Hard Drives: 8xSeagate 200gb 7200.9 SATA NCQ + 2 spare
System Drive: 1xSeagate 80gb 7200.7 SATA
CPU: P4 2.8E
Case: Supermicro SC743T-650

Hard drives get to a max of 31C in a room ambient of 24.4C

About the ISO 9001 cert and wrist bands - yes I know I should have them, but this time I didn't have my strap with me. It was constructed on a wooden table, and I made sure I grounded myself before handling the components.
Click to expand...

Supermicro Super P4SPA+ does not look like the motherboard you used, has no 64bit pci slots

EDIT: everything is different, thats not the mobo
 
AMD_Gamer said:
Supermicro Super P4SPA+ does not look like the motherboard you used, has no 64bit pci slots

EDIT: everything is different, thats not the mobo
Click to expand...
Agreed. That looks more like the P4SCi (which doesn't have a SCSI option...).

 
You must log in or register to reply here.
Back
Top