Large Disk Failure Help

[JorgeB]

4 minutes ago, pwm said:

If it was my server, I would start with a read-only parity scan just to test-read all drives and see what happens.

Yes, but since he didn't have parity IMO doing a read check with the data disks only or adding parity disk(s) and doing a parity sync would be the same operation/risk, if there are errors parity woun't be built, same with read check only, if there are no errors and he plans to add parity any way might as well add it now.

[pwm]

2 minutes ago, johnnie.black said:

Yes, but since he didn't have parity IMO doing a read check with the data disks only or adding parity disk(s) and doing a parity sync would be the same operation/risk, if there are errors parity woun't be built, same with read check only, if there are no errors and he plans to add parity any way might as well add it now.

Too many threads - I forgot there was no parity drive. So no parity data to be careful with.

 

Then it would be meaningful to try to build parity to a new drive, while giving each data drive a full read-through. If one of the data drives only survives a single read-through, then the system would at least have the required redundancy to recreate the contents of that drive. And if the parity build fails then it was because one or more data drives was already in too bad condition, in which case it's time to try data recovery on the drives who gives read errors.

[tential]

17 hours ago, jonathanm said:

No.

 

Now that you know each drive spins up on its own, CAREFULLY trace and unplug and replug each and every connection, both power and SATA. Also unplug and replug any PCIe cards.

 

After you've pretty much reassembled your server from scratch, then start it up and see how many drives are missing, if any.

 

Get diagnostics after unraid boots. I'm on the fence about whether or not you should attempt to start the array before obtaining diagnostics, maybe @johnnie.black has an opinion on whether starting the array would be beneficial at this stage of diagnosis / recovery. I'm operating under the "first do no harm" principle, and don't want anything to write to the drives yet until we know more about their health and the overall health of the server.

tower-diagnostics-20180115-2213.zip

 

All 10 show up now.

Not sure what's going on, whata does my diagnostics look like? 

I'm not sure what I should do here should I attempt to migrate everything to new harddrives or what?

 

I'm happy, excited, and scared to get my hopes up.

 

I don't need to try to save any specific files so there is that.

 

Edit:   When I'm adding the parity drive for that part, do I add both at the same time? 

Edited January 15, 2018 by tential

[JorgeB]

Start the array, if all looks good stop and add parity, you can add both at the same time.

[tential]

Just now, johnnie.black said:

Start the array, if all looks good stop and add parity, you can add both at the same time.

 

What should I be looking for to see if "all looks good"?

 

I'll prep some drives for parity!

[pwm]

39 minutes ago, tential said:

I'm not sure what I should do here should I attempt to migrate everything to new harddrives or what?

You haven't shown any log data that indicates any errors with the drives.

So no indication that you would need to migrate to new drives.

 

If you had dropped the computer from a table, then I would not trust the drives anymore.
But most probably, you bumped the computer hard enough that some connector came loose.

 

Kicking a computer when in use, you may bump the head inside the HDD.

The most probable outcome is that nothing happens.

If the HDD is busy writing, then it can produce a failed write - newer HDD has a g-sensor that detects impulses to let the drive know that it can't trust that write.

If extremely unlucky, the head could manage to hit the surface. But the air pressure is quite high so it isn't very easy for the head to press itself through that air cushion.

But it would really take quite a lot of bad luck to damage one drive - and even more to damage multiple drives.

[JorgeB]

Just now, tential said:

What should I be looking for to see if "all looks good"?

They mount and there are no read errors.

[tential]

Ok, I can't add the parity drive for 2 days until the new controller cards come in.  I forgot I had no Sata ports left

 

I'm guessing I should wait til then before starting the array?

Edited January 15, 2018 by tential

[JorgeB]

You can still start it and check all disks mount without errors

[tential]

They all mount, should I start a Check now?

[JorgeB]

I would wait for parity, at least you'd be protected after the first compleat read, but most likely your'd be fine either way, if they do a complete read now without errors most likely will do another in a week.

[tential]

I'm in no rush, I'll wait!

[JorgeB]

2 minutes ago, tential said:

I'm in no rush

 

Clearly

 

On 12/01/2018 at 9:59 PM, tential said:

It's been a year, and I'm trying to figure out which drives are safe and which ones need to be trashed

 

Might as well wait another week.

[tential]

Yup.

It's 2 day shipping, so not too long.  I wish I had realized when I ordered the cables.  I mean I did... but it didn't click that I needed cables AND expansion slots.

 

If these all do work.... I have 6 extra HDDs(The 10 I had bought because I thought this server was dead) with no bays now lol.  Oops....guess I'll need an unraid server #2 soon if this all goes well.

Edited January 15, 2018 by tential

[JorgeB]

Forgot to mention earlier, before doing the parity sync change the onboard SATA controller from IDE to AHCI and before or after update to latest stable, you're on a very old release.

[tential]

1 hour ago, johnnie.black said:

Forgot to mention earlier, before doing the parity sync change the onboard SATA controller from IDE to AHCI and before or after update to latest stable, you're on a very old release.

Errr do I do that change in the mobo settings right?

I swear I've worked with computers before!

 

And ya, I'll make sure to update too. 

[JorgeB]

In the bios, usually in the storage section.

[tential]

Ok, got my card installed, everything boots fine, I changed the sata controller just now, and now I'm updating.  Then I can add the parity drives which will check all my drives?

Edit: Default is AHCI... why do I have to change it to a default that I never changed it off from?!  Weird!\

Added tower-diagnostics-20180118-0023.zip

just to be safe incase it is needed

Edited January 17, 2018 by tential

[tential]

So I am doing the parity check... this takes awhile!

Whole thing looksl ike it takes 25 hours normally but because I used 8 TB drives for parity and my biggest drive is 5TB, I think it should b done at the 5TB mark right?

 

Anyway....

 

So I've been researching how to use dockers and I found the community apps.... I feel like that should be included in unraid?  Lol, anyway, this negates the need for my other small server.  I can turn that into media consumption device now.  I knew unraid had a lot of functionality, but now I don't even have to think about it for using it.  The SpaceInvader youtube videos have been quite helpful in getting a more thorough understanding too. 

[JorgeB]

5 minutes ago, tential said:

So I am doing the parity check... this takes awhile!

Assume you men parity sync, yes it takes a while.

 

6 minutes ago, tential said:

Whole thing looksl ike it takes 25 hours normally but because I used 8 TB drives for parity and my biggest drive is 5TB, I think it should b done at the 5TB mark right?

No, it will sync the 8TB, though it should speed up once you get past 5TB, though doing this should improve the overall speed:

 

-make sure the bottom addon controller has only 2 disks, since that is a DMI slot is shared with the onboard SATA controller

-change your tunables from default to these (settings -> disk settings):

 

md_num_stripes=4096
md_sync_window=2048
md_sync_thresh=2000

 

[tential]

15 minutes ago, johnnie.black said:

Assume you men parity sync, yes it takes a while.

 

No, it will sync the 8TB, though it should speed up once you get past 5TB, though doing this should improve the overall speed:

 

-make sure the bottom addon controller has only 2 disks, since that is a DMI slot is shared with the onboard SATA controller

-change your tunables from default to these (settings -> disk settings):

 

md_num_stripes=4096
md_sync_window=2048
md_sync_thresh=2000

 

 

Yes the parity sync sorry!

 

Darn, well at least it will go faster!  I am excited to watch a movie today to celebrate.

 

Bottom addoncontroller card -

Errr I'm not sure what you mean here sorry.

Are you talking about my PCI E controller cards?

I have 3 total (3rd one coming today) and access to all my PCIE slots.  I installed the first 4 port one (bottom on my mobo) I had in a PCI 1 lane, but the next one I put in the x16 lane?  Does that matter at all?

 

Should I just flip the cards around so that hte 2 port card it in that "bottom" slot?  How can I figure out the designation to ensure we're talking about the same card?

 

How come these settings need to be changed from defaults and can I do it while the parity sync is running or do I have to wait til after?

 

I assumed that I really couldn't do much while the parity sync was running, so I've just been reading.

 

Edited January 18, 2018 by tential

[JorgeB]

2 minutes ago, tential said:

Bottom addoncontroller card -

Errr I'm not sure what you mean here sorry.

Are you talking about my PCI E controller cards?

 

OK, you have a 4 port Marvell controller and I though you were getting another one, I see you have an Asmedia 2 port controller.

 

The Marvell one would be faster if was the PCE x2 version, the 9235, instead of the 9215, but make sure one of them is in the first x16 PCIe slot, the one usually used for the GPU, as that's your only non DMI slot.

 

Performance would be best for example with an LSI 8 port controller on that slot, together with the onboard ports it would be good for 14 disks total.

[tential]

9 minutes ago, johnnie.black said:

 

OK, you have a 4 port Marvell controller and I though you were getting another one, I see you have an Asmedia 2 port controller.

 

The Marvell one would be faster if was the PCE x2 version, the 9235, instead of the 9215, but make sure one of them is in the first x16 PCIe slot, the one usually used for the GPU, as that's your only non DMI slot.

 

Performance would be best for example with an LSI 8 port controller on that slot, together with the onboard ports it would be good for 14 disks total.

 

Ok perfect, I understand you!

Well, I have 2 4 Ports so I'm going to be screwed someway I guess.  I didn't understand that about DMI slots (or that they existed) until now.  Good to know!

In the future if I ever upgraded this, what is this LSI 8 port controller?

I'm trying to find it and everything I see looks extremely expensive.  I think I remember looking this up, and deciding to take the performance hit for budget reasons if I'm not mistaken.

 

Will it help at all to use my other x16 slot?

Edited January 18, 2018 by tential

[JorgeB]

1 minute ago, tential said:

In the future if I ever upgraded this, what is this LSI 8 port controller?

Any LSI with a SAS2008/2308/3008 chipset in IT mode, e.g., 9201-8i, 9211-8i, 9207-8i, 9300-8i, etc and clones, they can be found used on ebay starting at around $50, and you'd be better off with an used LSI instead of those.

 

2 minutes ago, tential said:

Will it help at all to use my other x16 slot?

No, that's also a DMI slot, and only x4 electrically.

[tential]

Hmmm.... I guess I should just buy the controller card off ebay then and return everything else when it gets here.  I couldn't find it anywhere else until you mentioned.

Any preference as to the chipset I should get or are the models similar for unraid purposes?

 

I don't want to redo this again but looks like I'll get a decent performance jump if I do.  My speeds picked up from 80-90 to 150 once it got to the 8TB drives which are th ones on the 2 SATA port x16 slot spot.