JonathanM

Typically they would copy the files to a new drive, but if you tell them you need bit level cloning for parity recovery they would attempt that if asked, obviously that would require the same size drive.

In that case, for best results you need to send the drives to a professional recovery service. You can easily do more harm than good by attempting this yourself.

That part of your scenario is not good. When power comes back, you want to be in control of the subsequent start, not automatically come back up. Many times when power is restored, it isn't on for very long before it goes out again for a period of time. The recharge time is often 10 or 20 times longer than the outage, so a second power cut would result in a server crash. As far as brand recommendations, typically APC is the first choice, as those are typically supported in unraid out of the box. There is a plugin for NUT available, which has a much broader list of supported models. https://networkupstools.org/stable-hcl.html First thing to find out is your full load power draw, and the amount of time it takes to shut down with no advance warning, as in all VM's and containers up and doing their thing.

VMWare != KVM

Does it work if you do this?

Did you mean 8088? Unless you changed the listening port in qB, I think you want to map 8088 to host and 8080 to container. Assuming you are using a bridge.

Before you proceed any further, please clarify your priorities. Do you want Data that was on drives Working drives for use in array You really can't have both unless you are extremely lucky and skillful, in equal parts.

Unraid uses https://en.wikipedia.org/wiki/Kernel-based_Virtual_Machine as its VM hypervisor, so if it's doable in KVM, it's probably doable in Unraid.

IMHO, that's not an accurate statement. The file system itself is ok, it's just that it's not forgiving of poor hardware or mismanagement.

Keep in mind that the motherboard and CPU must support it as well, doing 1 video passthrough is easy, almost any hardware will work, doing 2 is more demanding, typically only server or gaming hardware support it, doing 3 or more requires very specific hardware, best to find someone doing it successfully already and copy their setup exactly.

I don't think that will work, as the card doesn't split those 8 ports into separate IOMMU items. I'm not aware of any way currently to split a single display card into multiple KVM OS's, even if the card has multiple ports. The LTT videos all have one card per VM AFAIK. I remember reading somewhere that it might be possible in the future with different hardware, but I'm not aware of anything running right now.

Are you sure this specific plugin has been reworked to use with 2.X deluge? Plugins written for 1.X won't work with 2.X

Sometimes it works that way, more often once they start to lose capacity they drop off much faster instead of degrading linearly. The most common failure mode is that they still appear to hold some charge, but as soon as they get a load, they drop to almost 0. That's why typically as soon as they start degrading it's time to replace. Not for computers with good quality PSU's. The always on UPS's are needed when the connected equipment can't tolerate even a single missed cycle in the power wave. Switching UPS's drop the power for a very small but definite period of time while they switch over. Most electronics have filters that can cover that gap without a problem. Non-switching are always running the load off the batteries, and always charging the batteries. Under normal use the batteries aren't actually drawn down or charged much, as the charger balances to the load, but when the power goes out nothing actually changed as far as the output is concerned, just the batteries start being drawn down because the charger is off. Fun fact, the models I use specify a higher output power than input power, because they can briefly push the output harder than the charger can refill the batteries. Good for startup surges, they are one of the few UPS's that are rated for laser printers. Probably the same. The issue with power interruptions causing data corruption isn't about the hardware specifically, it's a matter of timing, and what exactly is in process and not fully completed when the rug is pulled out. If the server is idle for a period of time before the power is removed, the chances of corruption are pretty low. If you have a bunch of files being written, chances of corruption is almost guaranteed. Medical equipment usually specifies the higher quality units.

The UPS itself can last many years longer, the batteries need to be replaced periodically. Oversizing does two things, allows longer runtime, and consume more electricity during normal non-backup running. Most do a periodic test that simulates a power failure and starts drawing power from the batteries. If the batteries lose voltage faster than normal, the UPS signals a battery failure. Many UPS's also track the calculated runtime based on how much capacity the batteries had when new, and the tested capacity, and signal a failure when the battery is significantly weaker. Under full load is the only figure that matters, as shutting down the running system puts it at full load for the duration of the shutdown. Since you can't divine the future, there is no way to know how long the power outage will be, so you must plan to always shut down the system during an outage. That means to properly size a UPS, you find the full load power draw, and the maximum time it takes to do an unplanned shutdown, with all services and VM's being utilized. You then take that power and double the time, and look on the manufacturers spec sheet to find the appropriate UPS. If your system takes 100W at full load, and it requires 5 minutes to shut down, then you need to find a UPS capable of supplying 100W for 10 minutes. This is the minimum size required, and assumes you will configure the server to shut down almost immediately when the power is cut, waiting 1 minute at most. Also keep in mind that your network equipment should also be on the UPS if you plan on communicating with your server after the power goes out and before it's shut down. Do you keep a full current backup of everything on your server? If so, then you can probably skip a UPS.

Here you go. https://lithium-ion-ups.com/

Sure, but you will have to reconfigure and reboot.

Note. You only get ONE chance to get this right. The pins will break if bent more than a couple times. First hand experience talking here.

Automobile batteries are normally used in 2 ways, 1, to start the car, 2, to smooth out the power supplied by the alternator. If your car's spark plugs are running off of the battery, you are going to have a bad time in a few minutes. If things are working properly, starting the car takes a VERY tiny percent of the batteries capacity. If it's really cold out, and the car is in a poor state of tune, you may end up using a good portion of the battery before it starts, but you won't do that very often before you are replacing the battery. And yes, 99% of UPS batteries are SLA (sealed lead acid)

Are you sure FF is an error? On at least some boards, FF is the "all OK". Your monitor may not support the signal being sent. Do you have access to a different one?

The issue with the power cables isn't the continuous draw, that's perfectly fine. The way unraid works with a group of drives in the parity array, it will spin up all the drives at once if there is an error detected. The instantaneous draw of all the drives spinning up at once could draw enough current to cause the voltage to sag for just a split second. So, just when unraid needs clean data, you could be causing one or more of the drives to hiccup as the power does a quick brown out. You could keep the clean look, and split that segment of 10 into 5 and 5, fed from above and below with two different sets of wires from the PSU. I'm assuming since you are doing custom cable crimping that you are using soldered or crimp and soldered joins to the original PSU feeds? I would definitely eliminate any slip fit adapter connections you can, those are a bear to get right, and will cause high resistance heat building connections if done wrong. As far as data cabling causing issues, it's probably not going to show up right away, because at the moment everything is right and tight. It's after many heat and vibration cycles that any tension on the connectors can cause alignment issues to start happening, typically throwing CRC errors as the drive has to resend data when the first transmission is partially garbled by electronic noise.

Same behaviour with no CPU at all?

They are VERY similar to automobile batteries. They typically last between 3-5 years, don't like extreme temperatures, don't like extreme discharges, don't like repeated deep discharges, can fail for no apparent reason sometimes. Configure your equipment to finish shutting down before the batteries drain below 50%, keep them at human comfortable temps, and you have the best chance of keeping them for as long as possible. Many name brand UPS's try to make you think the batteries are special, I have yet to find one that uses truly proprietary batteries, they all just add plastic bits and wire jumpers with strong double stick tape to generic batteries. If you have a local source for alarm panel batteries, they are generally the same.

Ahh, I see. I googled the specs of a GIGABYTE W480 VISION, and found 6 SATA ports on the spec sheet. If I would have googled GIGABYTE W480 VISION W, I wouldn't have been confused. Amazing how 1 letter changes everything. 🙂

@phea Your machine is super clean, looks amazing. Please don't feel attacked by all the criticizing posts, we all really do admire a clean build. We are trying to give you some ideas for improving the long term stability of the system, so you don't start having strange issues that can be difficult to pin down. You obviously put a LOT of time and effort into the build, we want you to have a long and successful run with it. Quick question, I can't see where the last 4 SATA cables are connected? The motherboard only shows 6.