@Pete-S said in Blind swap / automatic rebuild on linux:

@DustinB3403 said in Blind swap / automatic rebuild on linux:

@Pete-S said in Blind swap / automatic rebuild on linux:

@travisdh1 said in Blind swap / automatic rebuild on linux:

@Pete-S said in Blind swap / automatic rebuild on linux:

@travisdh1 said in Blind swap / automatic rebuild on linux:

@Pete-S said in Blind swap / automatic rebuild on linux:

I often see that the argument for using hardware raid is to be able to initiate an automatic rebuild by just swapping a faulty drive for a new one.
A lot of people assume that software raid can't do that. But that's incorrect.

Software raid on linux (as in md managed by mdadm) can do the exact the same thing.

It's under policy and partition policy in mdadm.conf. You'll find on the man mdadm.conf page.
The spare-same-slot option would be the one that works the same way as hardware controllers usually do.

I haven't used it myself since I prefer to initiate the rebuild myself. But I wonder if you guys have used it?

I don't think blind swap is about automatic rebuild, that's a given no matter what software/hardware RAID is running. It's more about seeing the light is red instead of green on drive 6, so you know that is the one to replace.

The only example of not having that available, that I can think of, is https://www.45drives.com/

I don't know man.
A typical SMB would have no monitoring and any server would be stuck in a closet somewhere. Nobody would notice any red lights until several months later or until something breaks and then they'd have no clue what to do about it, wouldn't know who to call and wouldn't have any idea if the server even has warranty (it never has). A spare drive wouldn't be available unless it was an old discarded drive left on the shelf from the last time something was replaced.

While probably true, that doesn't really have anything to do with blind swap.

I'm just saying those that have their server park under control doesn't really need any LEDs. And those that really needs it, doesn't look at it.

But it would actually be a small thing to make a script that would indicate faulty drives. You look at /proc/mdstat and any drive showing a _ instead of U is lit up on the drive bay. It's controlled by SGPIO or SES. That's how the raid controller does it.

I thought MD was already capable of performing this. . .

I don't think so but I could be wrong.

I mean you could run raid 1 on a pair of sd cards. Since that md works on any type of block device or partition there is no guarantee that there are any drive bay lights or anything of that nature. But it's possible that there is an option for it.

Has a quick look and it looks like the ledmon package monitors md arrays and set LEDs accordingly.
So yes, software raid can indicate what drive has an error directly on the chassis with some additional software.

@StorageNinja said in RAID rebuild times 16TB drive:

@scottalanmiller said in RAID rebuild times 16TB drive:

Its a system, not an IO, bottleneck typically. Especially with RAID 6. Its math that runs on a single thread.

Distributed storage systems with per object raid FTW here. If I have every VMDK running it's own rebuild process (vSAN) or every individual LUN/CPG (how Compellent or 3PAR do it) then a given drive failing is a giant party across all of the drives in the cluster/system. (Also how the fancy erasure code array systems run this).

Yeah, that's RAIN and that basically solves everything 🙂

@scottalanmiller said in Dell MD1220 RAID 5 Rebuild Question:

@Jimmy9008 said in Dell MD1220 RAID 5 Rebuild Question:

@scottalanmiller said in Dell MD1220 RAID 5 Rebuild Question:

@travisdh1 said in Dell MD1220 RAID 5 Rebuild Question:

That makes a little more sense. They've negotiated a deal with Microsoft from the sounds of it.

No, way, way more likely they just figured out how hard it is to get caught.

I totally doubt anything like that is going on here. I have been told our annual budget for licensing which we pay is in the region of £600,000. Either way, not my problem 🙂

Wow. But, sounds like only that high because they don't know what they are using, lol. The only way to get licensing truly down is to know what you use. Someone is both ignoring what is used, but also encouraging unlimited use. Both things set MS up to just keep making it more and more expensive. It's a trick, sounds easy, but makes one lazy licensing person encourage not keeping licensing lean - basically giving Microsoft the power to charge anything that they want down the road.

Yeah, not your problem, but definitely a symptom of management issues and a lack of clear thinking. If they are truly paying their bills, my guess is a licensing "specialist" who has created their own job and knows if MS isn't used heavily, their job would go away, so is doing stuff to encourage you to lock in MS so that that specialist can't be eliminated. Basically creating their own job.

Yeah, I don't disagree with anything you said. It's just not my issue. I still get to buy shiny new toys 🙂