Strange behaviour while building parity

[KRiSX]

Hi all, I'm currently looking to build/transition to unRAID from my existing Windows DrivePool setup and had something very weird happen during parity build. Due to the large amount of data I'm dealing with I'm having to be quite careful doing this transition, so right now my unRAID setup consists of 2 x 8tb's for parity, 1 x 1tb and 1 x 750gb SSD's for cache and a single 1tb hdd just to get the ball rolling (will be removing/replacing as soon as I free up some of my other drives - DrivePool's "put everything everywhere" approach is making it harder than I'd like.

 

Anyway, yesterday I started the lengthy parity build process. My server has about 30 drives in it all up, 24 of which are used for my main storage (including the 2 I'm now dedicating to parity). I have been hoping to take advantage of the disk spin-down to save on power as I've traditionally been running all drives spun up for years on my existing setup, so I had spin-down set to 1 hour.

 

During the build process I checked back from time to time and did see the drives not in the array spin-down and thought to myself "awesome, that works exactly as I'd expect", until I noticed the main Parity drive was spun-down! This didn't seem right, but I left it running. I also noticed all the drives would spin back up from time to time despite not being accessed or part of the array. It seems that SMART is triggering them to spin-up based on what I see in the system log, more on this later.

 

8+ hours later the build finished and said all is good, i thought to myself that is highly unlikely, but ok if its wrong verify will fix it or maybe I'll just run a verify to be sure. At this point I hadn't added my cache drives, so I shut down the array to prepare those drives and add them.

 

Upon restarting the array I was extremely shocked to see another 1tb drive I'd been playing with (so it was formatted in xfs but not in any array) had taken place of the main 8tb Parity drive! The array started and said all was ok when it clearly wasn't, it was after midnight at this point and I was about ready to ditch the whole thing, but I'm very keen to make the switch. So I shut down the array, formatted the 2 x 8tb's, performed a new config and started the build again - but this time, I set drive spin-down to Never. Going on nearly 10 hours with another hour and a half or so to go (slower than last time), parity is nearly finished building, however I am quite concerned that this will happen again so will need to do some testing before I go trusting my data to unRAID.

 

I have a couple of theories as to what happened/what caused it but thought I'd get it all down and see what the experts think. My main theory is that due to the majority of my drives being connected via an Adaptec 6805 and Intel SAS Expander, the spin-down options simply don't work or play nice with the system. The drives are all individual JBOD and unRAID sees them just fine, but as mentioned earlier the drives on the Adaptec kept spinning back up for SMART checks but some other random disks I have connected to the motherboard SATA ports did not, they stayed spun down as expected. This to me says I shouldn't be using spin-down on the Adaptec perhaps?

 

Another thing I've noticed is that after a reboot of the whole server, drives pop into the Historical Devices list when literally nothing has changed with them. I believe this is part of the Unassigned Devices plugin, but it does concern me because why would it think the drives have changed or been removed when they haven't? (I did see an update for Unassigned Devices late last night that mentioned drives being identified incorrectly or something along those lines - so perhaps thats fixed now, but thought I'd mention it anyway)

 

Sorry for the long post, but I've been performing a lot of testing before taking the leap with this and I'm starting to wonder if I should continue or not.

 

Appreciate and help or advice

Thanks

Edited January 24, 2022 by KRiSX

[KRiSX]

Update, Parity is built and I've restarted a couple times, stopped and started the array, even spun down disks (which annoyingly about 14 started up shortly after for some reason)... can't get it to break again... very odd, but seems stable now at least *crosses fingers*

Edited January 25, 2022 by KRiSX

[Vr2Io]

4 hours ago, KRiSX said:

8+ hours later the build finished and said all is good, i thought to myself that is highly unlikely

 

4 hours ago, KRiSX said:

another 1tb drive I'd been playing with (so it was formatted in xfs but not in any array) had taken place of the main 8tb Parity drive!

Yes, those abnormal.

 

4 hours ago, KRiSX said:

Adaptec 6805 and Intel SAS Expander

Issue may relate, but I haven't 6805. Does SMART data could be show ? Pls check the log have abnormal disk disconnect / connect / detect again etc. Some hardware combination may not work well once disk spindown.

 

4 hours ago, KRiSX said:

the spin-down options simply don't work or play nice with the system.

It also true, but at least if no disk activity in array, it spin-down to save power in my experiences.

 

 

 

 

Edited January 25, 2022 by Vr2Io

[trurl]

attach diagnostics to your NEXT post in this thread

[KRiSX]

On 1/25/2022 at 2:28 PM, trurl said:

attach diagnostics to your NEXT post in this thread

 

thanks, sorry for the delay I've been busy clearing off disks in DrivePool so have only just had a chance to boot unRAID again. All appears to be ok at the moment, but I'm not sure if I should use spin-down or not at this stage.

newbehemoth-diagnostics-20220128-1200.zip