Keep getting read errors on Parity build with new drives

[dkaholz]

Hey all,

Building an Unraid server was my first foray into Linux so i apologize if at any point during this i sound confused.

for the most part i have gotten my array set up as a plex server and it runs fine with no parity, no issues or errors.

When i try to build parity, after a few minutes i immediately start getting read errors. I thought it might be because of a bad drive, one had mechanical issues found on it and i replaced it, but i keep getting read errors on parity drives and they become disabled while building parity. Ive swapped out a bunch of drives and at this point only 1 or 2 of the eight drives arent brand new out of the box.

 

Ive attached my diagnostics zip and any help would be greatly appreciated but i am all out of ideas.

tower-diagnostics-20220620-1309(1).zip

[Frank1940]

You might start by running a 24 hour memtst which is one of the boot options.

 

The parity disk and disk 5 have missing smart reports.  This is probably because they are offline...

 

Since this is a new build and you (apparently) have had a lot of drive issues, I would suggest that you have a good look at the power supply.  Ideally, it should have a single 12V rail with a current capacity of 24amperes.  (16Amperes might work if the drives are the 'green' type.) 

[dkaholz]

Parity drive is currently disable but i pulled it out of the box this morning. And disk 5 was the one with mechanical issues.

[Frank1940]

offline = disabled

 

I am a bit confused.  How much data have you actually put on this server?  (The reason for the question is that you have ~28TB of storage space available across seven disks.  And this is a 'new' server.   Are you actually going to need this much storage in the near future? One disk has 48,000 poweron hours and another has 65,000 hours.   I can understand ( for financial reasons) to want to use these disks but if you don't need the their storage space for a year or two, why put the hours on them...)

[dkaholz]

It's that big because it was a plex server I was building for the last couple of years.  It's basically full of data and I've just now gotten the time to start a new server

[JorgeB]

Two of the SATA ports are set to IDE, go into the BIOS and change them to AHCI/SATA, usually ports 5/6, this is a known issue with these AMD chipsets and it's where parity is connected.

[dkaholz]

Memtest came back fine, i changed the AMD settings in BIOS to be ahci/sata after that i tried to start the server and run the parity build and everything hung up and my disks 1 and 2 gave me "unmountable - wrong file system" and the other one was just stuck on "mounting" these disks have the right file system on them and are brand new no issues until now

[JorgeB]

Post new diags.

[dkaholz]

Wow that turned into a nightmare real quick, the server became unresponsive so i had to reboot, after that i was able to start the array but every read from Disk 2 had an error, boot up showed SMART errors on the drive so i hooked it up to Parted Magic and ran a health test, showed no signs of anything wrong, grabbed a new sata cable because that might be the only thing i hadnt changed out and started the server back up. Now it is building parity but im sure it will error out, i can post new diags but my understanding is after a reboot they are gone? Or is that wrong? Posted them just to be safe

tower-diagnostics-20220621-0941.zip

[JorgeB]

4 minutes ago, dkaholz said:

Now it is building parity but im sure it will error out,

Everything looks go so far, all disks are mounting and there are no read errors, if there are errors during the sync grab new diags before rebooting.

[dkaholz]

There was an 8tb seagate barracuda drive i had in the beginning of all of this that i thought might be causing issues, it was plugged in when this all started, it isnt plugged in now but i dont think its enough to narrow down the issue just yet

[dkaholz]

Parity stopped because of a read error and went to the screen below, diags attached

 

tower-diagnostics-20220621-1030.zip

[JorgeB]

00:14.1 IDE interface [0101]: Advanced Micro Devices, Inc. [AMD] FCH IDE Controller [1022:780c]
    Subsystem: ASRock Incorporation FCH IDE Controller [1849:780c]
    Kernel driver in use: pata_atiixp
    Kernel modules: pata_atiixp

 

Controller is still in IDE mode.

[dkaholz]

Damnit i must not have changed the correct setting, i think i got it, i rebooted and parity is building, i downloaded diags again, i tried to look through them but dont see a mention of IDE on the logs dunno if i was looking in the right files though

tower-diagnostics-20220621-1108.zip

[JorgeB]

Now it's OK.

[dkaholz]

Parity ran longer than it has but still errored out

tower-diagnostics-20220621-1434.zip

[JorgeB]

Try replacing both cables, power and SATA, if you don't have new ones swap both cables with another disk and try again.

[dkaholz]

I actually replaced the sata cable yesterday and the power cable is different from previous drives I've tried

[JorgeB]

Then it's likely a disk problem, despite the healthy looking SMART.

[dkaholz]

I guess I can let it build parity off the disks one by one and see if I can find the culprit that way because except for 2 of these disks they are all brand new.  And the two disks passed health checks

Edited June 22, 2022 by dkaholz

[trurl]

Only disks 5, 6 have had extended self tests.

 

And those weren't recent.

[dkaholz]

They shouldve been done a few days ago, otherwise every drive is brand new. I rebooted the server because the parity drive wont even let me select it anymore any my other 8TB shows up as having a smart error, im going to return it, somehow that is 2 drives out of the box that are apparently bad....