• Please review our updated Terms and Rules here

Bought a dell poweredge at the swap meet and got scammed I guess, most expensive thing i bought, how can I fix it?

Well there is two PSU's each 1,400w so that's close to 3kw I am unable to get 240v so I guess I will just sell it
That's not how dual PSUs work in servers like that. They are there for redundancy, not power output, so the server will never use more than the power a single PSU offers, in fact you can pull one out completely and the system would work fine, though it would be complaining about a failed PSU.
 
That's not how dual PSUs work in servers like that. They are there for redundancy, not power output, so the server will never use more than the power a single PSU offers, in fact you can pull one out completely and the system would work fine, though it would be complaining about a failed PSU.
Oh I assumed it was one PSU for each server
 
Dell only released two power supplies for that chassis, a 1100W unit (XVKF0) and 1400W (RN0HH). The former is a universal 120/240v unit, while the latter is 208/240v. The possibility of someone swapping to a lower power PSU is extremely remote due to the configuration of the server. The only possibility of damage would be plugging the 1400W unit into 120v, because the power supply would be trying to draw excessive current on startup, which may blow the fuse or damage the SMPS circuitry.

If OP wanted to, they could buy the 1100W units and swap them out, but I would recommend pulling two of the CPUs and the associated RAM to avoid overloading it. While 460W is well below the PSUs rating, you have to account for transient current spikes, which may trip the lower power PSU.

Here is an Ebay listing for a 1100W unit: https://www.ebay.com/itm/114018234468
Could I take one of the nodes out, buy only one smaller PSU put that in and would that get the power under control? The server can run with only one node and one PSU right? How much power would that use I'm trying to keep it reasonably under control
 
What is that you are trying to achieve? Like what is your end goal with this hardware?

I have to be honest with you, it sounds like you are way outside your depth on this, but also that this hardware is just not going to be worth your time, though that really depends on your goals and use case. You bought a hyperscale enterprise server, those are really not designed to run outside of a data center and not in massive clusters.

This is like like buying a formula 1 car and expecting your local automotive parts place to help you get it up and running. The basic automotive principals apply in very general terms, but under the hood an F1 car is going to be completely different to that of a Civic.
 
That's not how dual PSUs work in servers like that. They are there for redundancy, not power output, so the server will never use more than the power a single PSU offers, in fact you can pull one out completely and the system would work fine, though it would be complaining about a failed PSU.

Yes, dual PSUs in servers can work like that. They can be configured to be redundant or do load balancing, where both PSUs are used at the same time. It depends on how the server has power implemented. It's usually configured in the firmware setup on the server board, or via a management interface.

Could I take one of the nodes out, buy only one smaller PSU put that in and would that get the power under control? The server can run with only one node and one PSU right? How much power would that use I'm trying to keep it reasonably under control

If you have a single 1100W PSU (assuming the server can operate with just a single unit), you'll need to pull two of the CPUs and any RAM installed in the sockets for those particular CPUs. You'll need to consult the manual on what valid CPU configurations are available, because generally specific sockets need to be populated in order. If there's any silk screen text near the CPU sockets, it may tell you which sockets are which. If you have sockets 0, 1, 2 and 3, you want to keep CPUs in sockets 0 and 1 and pull 2 and 3. Be very careful doing so, these CPUs are large and the sockets are fragile. I'd recommend reinstalling the heatsink over the empty socket to prevent something falling inside the socket and destroying it.

With just two CPUs, you'll have 32 "cores". The Bulldozer architecture was wonky and didn't follow standard x86 design. Each "core" had two integer units and one floating point unit shared between them. This meant that tasks that did integer only workloads were mostly unconstrained, but anything that did floating point work would be severely constrained. This is why Microsoft and Linux eventually modified the thread dispatcher in the kernel to stagger heavy loads between modules and backfill the second integer unit with an idle background task to avoid two resource intensive processes from slowing each other down trying to use the same resources.
 
What is that you are trying to achieve? Like what is your end goal with this hardware?

I have to be honest with you, it sounds like you are way outside your depth on this, but also that this hardware is just not going to be worth your time, though that really depends on your goals and use case. You bought a hyperscale enterprise server, those are really not designed to run outside of a data center and not in massive clusters.

This is like like buying a formula 1 car and expecting your local automotive parts place to help you get it up and running. The basic automotive principals apply in very general terms, but under the hood an F1 car is going to be completely different to that of a Civic.
Yeah I thought I was getting a much less powerful server, should've realized when it weighed 82lbs, Ive never done server stuff before I was going to use it as a server, and upgrade a computer I use for that purpose, but I realize I can't really do that it's way to impractical. I then just wanted to play with it, (since you know I haven't used server stuff before) but I think your right it's not worth my time I am just going to try to sell it on marketplace. I paid $70, unsure of its worth maybe try $150?
 
Certainly worth a try. It might be good for someone doing a home lab, but not needing to have it powered on all of the time. (That’s how I run my power hungry enterprise equipment, I leave them off until I need to test something).

That server has quad cpu, each with 16 cores, so it might be pretty useful in a virtualization setup, but it doesn’t officially support anything past esx 5.5 and the cpu doesn’t support past 7

I say list it and see if you get any takers.
 
Yeah I thought I was getting a much less powerful server, should've realized when it weighed 82lbs, Ive never done server stuff before I was going to use it as a server, and upgrade a computer I use for that purpose, but I realize I can't really do that it's way to impractical. I then just wanted to play with it, (since you know I haven't used server stuff before) but I think your right it's not worth my time I am just going to try to sell it on marketplace. I paid $70, unsure of its worth maybe try $150?

If you really want to get a more practical server to work with and learn, then I'd suggest looking into something like a Dell R320 or R330 server. It's real server hardware, but much more like regular computer hardware and way less of a beast than the server you have. The R330 will cost more, but it'll use less power and be quiet enough to run on a desk while playing with it.
 
Dell only released two power supplies for that chassis, a 1100W unit (XVKF0) and 1400W (RN0HH). The former is a universal 120/240v unit, while the latter is 208/240v. The possibility of someone swapping to a lower power PSU is extremely remote due to the configuration of the server. The only possibility of damage would be plugging the 1400W unit into 120v, because the power supply would be trying to draw excessive current on startup, which may blow the fuse or damage the SMPS circuitry.

If OP wanted to, they could buy the 1100W units and swap them out, but I would recommend pulling two of the CPUs and the associated RAM to avoid overloading it. While 460W is well below the PSUs rating, you have to account for transient current spikes, which may trip the lower power PSU.

Here is an Ebay listing for a 1100W unit: https://www.ebay.com/itm/11401823

Dell only released two power supplies for that chassis, a 1100W unit (XVKF0) and 1400W (RN0HH). The former is a universal 120/240v unit, while the latter is 208/240v. The possibility of someone swapping to a lower power PSU is extremely remote due to the configuration of the server. The only possibility of damage would be plugging the 1400W unit into 120v, because the power supply would be trying to draw excessive current on startup, which may blow the fuse or damage the SMPS circuitry.

If OP wanted to, they could buy the 1100W units and swap them out, but I would recommend pulling two of the CPUs and the associated RAM to avoid overloading it. While 460W is well below the PSUs rating, you have to account for transient current spikes, which may trip the lower power PSU.

Here is an Ebay listing for a 1100W unit: https://www.ebay.com/itm/114018234468
Just want to make sure before I do anything, can I run the server with one 1100w PSU just with some CPUS out? And can I run it with just one node and one 1100w PSU?
 
Just want to make sure before I do anything, can I run the server with one 1100w PSU just with some CPUS out? And can I run it with just one node and one 1100w PSU?

The server should theoretically run on a single PSU, since they're designed to be redundant. But, the server may complain on boot or have an alarm if it can't detect the second PSU. The firmware setup should have options to control how the power supplies are configured.

Also is there any way to control the fan speed?

Dell servers have fan control settings in the firmware setup. Different servers have different fan configuration available.
 
Also is there any way to control the fan speed?
With a single active PSU the fans will run at turbojet speed, if it's like most other PowerEdges. To get the fans slower you need both PSUs powered and software running on the operating system you're running; the iDRAC (if so equipped) controls the fans.
 
Sometimes the higher wattage supplies are required for the high end configurations.

Also, some Dell poweredge will be OK with one PSU, if they were powered up initially with just one. But that was in the older days when dual PSU was an extra paid for option.

You can also get a Y cable and just run both PSU off of the same outlet. Assuming that your configuration supports the lower wattage PSUs.
 
Sometimes the higher wattage supplies are required for the high end configurations.

Also, some Dell poweredge will be OK with one PSU, if they were powered up initially with just one. But that was in the older days when dual PSU was an extra paid for option.

You can also get a Y cable and just run both PSU off of the same outlet. Assuming that your configuration supports the lower wattage PSUs.
It should be fine with the lower end PSU's if I take some cpus out apparently I just don't want to buy 2 PSU's if I don't have to
 
With a single active PSU the fans will run at turbojet speed, if it's like most other PowerEdges. To get the fans slower you need both PSUs powered and software running on the operating system you're running; the iDRAC (if so equipped) controls the fans.

You don't need to use iDRAC in the OS to control the fans, you can do it in the firmware setup at boot. If he's lucky, that server will allow a speed setting for the fans, rather than the older three options that just gave you low/medium/high.

I have a bunch of Precision Tower 5810s and PowerEdge T350 pedestal servers that I have set to 35% fan speed. It keeps them from annoyingly ramping up to full speed and back down again because their fan curves are garbage. They basically let the system overheat and go "oh crap i need to cool down!" over and over again.

Also have a few older T3600 and one T5610 that just have the low/medium/high fan speed override. Low is too low and medium is too fast, but still better than full speed.
 
You don't need to use iDRAC in the OS to control the fans, you can do it in the firmware setup at boot. If he's lucky, that server will allow a speed setting for the fans, rather than the older three options that just gave you low/medium/high.

I have a bunch of Precision Tower 5810s and PowerEdge T350 pedestal servers that I have set to 35% fan speed. It keeps them from annoyingly ramping up to full speed and back down again because their fan curves are garbage. They basically let the system overheat and go "oh crap i need to cool down!" over and over again.

Also have a few older T3600 and one T5610 that just have the low/medium/high fan speed override. Low is too low and medium is too fast, but still better than full speed.
Im probably going to end up just selling the ram and CPU'S and scrap the rest
 
Unfortunately, you won't get much for the CPUs and RAM. G34 socket CPUs are worthless due to the market being flooded with them from scrapped servers. The RAM is equally worthless since it's ECC Registered DDR3. That stuff will only work in servers or high end workstations.

I looked up Opteron 6276 on a few places, and they're $15-30 a piece. The RAM is about the same. Accounting for marketplace fees, you won't get much for them.

The motherboard itself has the most value, because those are generally scrapped and fewer of them are available. Since they're so large and proprietary, they're not economical to ship.

I would just try offering the entire server locally for some cost on FB marketplace or whatever other local market you may have around.
 
Back
Top