what are the odds it can survive this?

[gumby327]

I tend to use refurbishment hardware so problems are expected.  Given the screenshot who thinks replacing these drives with the ones in the mail tomorrow will magically heal it?

[trurl]

7 minutes ago, gumby327 said:

refurbishment hardware so problems are expected

Even if parity were good, parity has none of your data. Parity is just an extra bit that allows a missing bit to be calculated from all the other bits. To reliably rebuild every bit of a missing or disabled disk, it must be able to reliably read every bit of parity plus every bit of all other disks. You never want unreliable disks in your server.

 

attach diagnostics to your NEXT post in this thread.

 

[gumby327]

it sounds like I did not lose any data then.  I will continue my data loading, it is not done yet.  As for reliable, everything is backed up.

 

[trurl]

6 minutes ago, gumby327 said:

it sounds like I did not lose any data then.

Don't know where you got that idea. Certainly not from anything I said.

 

There are some things about your screenshot that are somewhat ambiguous, such as 2 disks with X when you only have single parity.

 

That is the reason I asked for diagnostics so I could try to better understand your situation.

 

Definitely you should not continue to transfer data to the system as it is now.

[Aaron Arnold]

you may also want to take a look at the drive in unraid it has a list of different values and see what type of error it is if it's a UDMC (Not sitting at my server that might not be the right name exactly), in which case it could just be a cable issue, which a new drive would not fix.  As trurl said a diagnostic file would have told us this.  

[trurl]

11 minutes ago, trurl said:

such as 2 disks with X when you only have single parity.

My mistake, you have dual parity so that clears up the screenshot. But you also have 2 disabled disks, so no parity protection currently, and the disabled disks will have to be rebuilt.

 

Really you are wasting your time transferring data to this unstable system as it is now. You need to get things working well. Quite possibly it is just bad connections that can be easily fixed, but we can't tell without...

 

Diagnostics

[gumby327]

Let me venture a little further out into my response here.  Those drives were hugely over worked and on my bad list for some time.  I knew two were for sure goners and one was iffy.  They were throwing every possible trait of a bad drive when they were in Windows.  I decided to torture them by putting them in an array again, (for the 3rd time).  They were a science experiment.  Well, they failed as predicted and I plan on going though my process of replacing them to see if the new drives come on line and how much data was lost in the end.  Based on their hours of operation they (64 meg of cache) were on 24/7 in a server farm for 11 years before they failed.  For me in my servers ... I have been using them a couple of years.  The ones I normally get are 128 meg of cache and they tend to be 5 to 7 years old and have tons of time left on them.  The ones I just bought were the 128's, but the price was $50 each so they will be a treat for sure.  Treat as in they will most likely have some hiccup.  They probably have ridiculous hours on them for that price point.  It is a media (movie) server so when it fails I pull out my backup server and run my robocopy to put back the damaged and drive on.

 

Here is my take on life in computing, if nothing ever failed I would never learn anything and life would be boring.

 

[trurl]

Why bother with parity then? It isn't required and sounds like you don't rely on it for anything (or at least won't be able to).

 

You could just use those 2 parity disks as extra storage. And write speed would be much better without realtime parity updates.

[gumby327]

it was an experiment to see if Unraid's flavor of it was advantaged over my hardware one, or Windows Storage Pools ... I think I am interested in giving that up now, but I wonder if it would be a matter of starting the build from scratch.  I am getting in a new controller and moving some hardware around later this week so I may do that, if I can figure out how to make Unraid do that.

 

[gumby327]

To be honest, I would like to mount a NTFS-3g 12TB disk locally to speed up the transfer, but the possibility of losing my master data makes it not worth it.

 

[trurl]

1 hour ago, gumby327 said:

how to make Unraid do that

Do what exactly?

[gumby327]

I stopped the transfer I was doing stopped the array and pulled off the bad Parity disk.  That was easy, but changing raid level is not clear to me other than a full rebuild.

 

[gumby327]

since we are talking a lot I thought I would place my diagnostics out there so you can see how bad it really is.  it is bad.

tower-diagnostics-20211214-2020.zip

[trurl]

17 minutes ago, gumby327 said:

changing raid level

Unraid IS NOT RAID. There is no raid level in the parity array. There is just Unraid. You can have btrfs raid levels in additional pools such as "cache". Often SSDs are used there for speed.

 

https://wiki.unraid.net/Manual/Overview

 

[gumby327]

I think, and this is my perception ... unRAID when you say it that way is beautifully simple.  Here I was looking for the hidden trap door to drop through for other things, when no, it is unRAID.  So, you have no choice to not use the parity portion of the array then?  And if there isn't that is fine I think.  I feel this fits my needs exactly in that I have a TV server that pumps in new recordings and I can spin those into the raid cache and out to the array on a move.  But the entire array is a reader.  It serves the 7 TV's in my house.  Nothing more.

[trurl]

1 minute ago, gumby327 said:

no choice to not use the parity portion of the array then?

 

1 hour ago, trurl said:

It isn't required

 

[gumby327]

Good, so I can simply take the drive out of that slot and leave it empty then

 

[trurl]

SMART attributes for the disks that were assigned as parity and disk4 don't look good. Looks like they became unassigned disks after disconnecting.

SMART attributes for all other disks looks OK though.

 

 

[trurl]

Can't tell whether any disks mount since the diagnostics are with the array stopped.

[trurl]

1 minute ago, trurl said:

Can't tell whether any disks mount since the diagnostics are with the array stopped.

Not entirely true, syslog shows them mounting the last time the array was started. Even emulated disk4 mounted, so you could conceivably rebuild it though I would replace it for rebuild.

[gumby327]

the three disks that are questionable, two failed, and 1 with relocation of sectors are all headed to the recycling center.  Based on this exercise I have strong confidence that the unRAID solution fits me fine.  It made it so I could have multiple fails and still worked.

Edited December 15, 2021 by gumby327

[gumby327]

My update of what happened.  I installed the SAS (was SATA) hard drive controller and booted it up with the drives pulled I deemed bad (3x) removed by SN from the identification screen in unRAID.  Installed the three new drives and it came up with the error that I had too many drives out of place.  So I went over to  tools new config and clicked it.  Rebooted for good measure and the screen showed me all the normal data is about to get destroyed alerts.  I said yup, go ahead and getter done.  It came up right away with 70% of my data present, all of it was unprotected, and Parity-Sync and rebuild was afoot for the damaged contents.

 

That is exactly what I expected to occur and was well within my parameters for this server.  That and my write speeds are now double what they were.  One day to built Parity is now 6 hours.  Two more drives started showing warnings ... so, back to Amazon.  I am wondering if maybe this time I go with low end desktop new drives....  I think they would have more life in them time/cost ...

[trurl]

12 hours ago, gumby327 said:

low end desktop new drives

That is what I do. If you set things up correctly (dockers/VMs not on the array) the array disks will spin down when not in use.

 

Since I have ITX case I have to upgrade disks to get more capacity, and repurpose the older drives long before they actually fail.