To your third post, if you had pulled out one or both of the disks, that could have helped the "failover" to occur.
If a disk hangs or disappears, look at the enclosure. Are any disk lights stuck on? or blinking fast?
disks 3 and 7 are electrically isolated completely. When they "disappeared", this was in the SoftRAID app? Did they not appear at all? or were the disks showing with thermometers? or ? in the disk tiles?
Just saw your "fail over" update (thank you). Yes, I pulled the disk in Bay 3 specifically because I thought that would be wise so the RAID didn't run broken. But evidently it was really broken because the disk in Bay 7 had not shown. No lights were stuck on on blinking fast, I was there at the enclosure. Disk 7 did not appear at all. Not showing with thermometers. No disk titles. Just not showing. It was about an hour before I realized that disk/bay 7 wasn't showing. Now all disks show.
No saves since and just checked dates/times. I'll get the mirror rebuild started but the deeper question is how could it fail over if I pulled half the RAID when the other half was not even showing under Disks? And if so, do I have an issue with my enclosure? Right now I'll just get the rebuild started. It'll take all night (24 hours? 4TB on spinning HDDs)
Followed directions to the letter. Re-build has begun. Maybe it won't take as long as I thought. Showing about 6 hours on the counter. Not too bad.
When Disk 7 disappeared, I wonder if it was the primary. That makes the most sense, as then it would have failed over to disk3.
A disk should never "disappear" like you describe. Now we have to figure out, enclosure or disk.
What I would do is next shutdown, move 7 to 4. Now you have it in a different location. If this happens again, it will either be slot/tray 7, or disk4. and we will know what the cause is.
Troubleshooting can be a pain at times, as you need to be meticulous to fix the correct component.
If the volume was not on Time Machine, then more likely the Spotlight index was the cause. It can be damaged where it is on continuous indexing, and as a result, the volume does not unmount. (Ithis is under MacOS's control not SoftRAID)
One fix is exempt the volume in Spotlight privacy settings, then restart and remove the exemption. that essentially resets the index. (It may take some time to re-index), then hopefully unmounting will behave correctly.
Sometimes a third party app can be the cause also, anything that leave the volume marked open.
When Disk 7 disappeared, I wonder if it was the primary. That makes the most sense, as then it would have failed over to disk3.
I believe this to be the case. To keep things short, I wish I had taken a screen shot but am virtually certain that upon loading SoftRAID and seeing the error, I believe it flashed an alert and then asked if I wanted to make the secondary disk the primary. As Disk 7 had disappeared, that would indicate that the Disk 3 was in fact the secondary and that Disk 7 had been the primary.
A disk should never "disappear" like you describe. Now we have to figure out, enclosure or disk.
What I would do is next shutdown, move 7 to 4. Now you have it in a different location. If this happens again, it will either be slot/tray 7, or disk4. and we will know what the cause is.
To be clear, when I say "disappear," nothing disappears once shown in the Disk column. If the disk shows, then it seems to be stable. Every single time this has happened has been either powering the Flex 8 on, or plugging in the TB cable. That's the only time this has happened, and every single time it has happened has been under those conditions. Once things are up and running successfully, things seem to be okay. So far at least.
I'll move 7 to 4 after the re-build completes. Recall that Disk 4 was also not shown but it didn't contain a RAID volume so evidently I got away with it.
Now I wish I had kept records of the 3 or 4 times this happened before, since I don't know if disks not showing has occurred on any other bays. I think after the re-build I will also physically label the disk trays with my label-maker, and also "Add Disk Label" within SoftRAID itself.
If it's a disk issue, Disks 4 and 7 are Seagate EXOS shipped with the Flex.
If this happens again (not even sure what to hope for now; if it does I can troubleshoot more; if it doesn't, that would be good), I'll try to react calmly enough to inspect everything more closely in the moment (disks, bays, etc.).
The more I thought about our discussion about Disk 7 not showing, and then reminding that Disk 4 had also not shown this evening, I found our exchange below from a couple weeks ago now. I said:
the next new issue (which was a bit concerning until re-connecting to the MBP which proved no data loss), only 4 of the 8 disks even appeared in the disk menu of SoftRaid, or in the Finder. (Then quit SoftRaid and looked in Disk Utility). So half my disks weren't even recognized.
You replied:
Note: If only 4 disks in the Flex 8 show up, that is a Flex 8 hardware issue, I cannot think oif any bug/issue in MacOS that can cause only half of the drives to show up. The lik,ely thing is the enclosure "hung" on one of the PCI buses. if you see this again, look at the drives, are any lights frozen on? are the 4 drives all on either top or bottom? (the Flex8 has two PCI buses)
I'm not sure if that's good or bad. It seems less likely to me that 4 disks would become problematic at the same time on the same day. I'll keep you posted as things continue either way.
Rebuild just finished. 3:05 AM (yes I stayed up to monitor it). The volume still shows a "missing secondary disk" notification but it points to both disks now and each disk shows the proper designation. Disk 3 set to "primary." Disk 7 set to "secondary."
I'm going to bed while I still can instead of trying "Remove Missing Secondary Disk" at 3 am. That said, from everything I've read in the Help/Manual document and on the OWC website, that's what I should do to clear the counters, correct? "Remove Missing Secondary Disk" isn't going to break RAID 1/mirror up, right? I'm a little frazzled, and becoming a little paranoid. Yes, I'll be an expert when this is over. But being an expert won't matter if my business completely dies for not being able to get back to work in time.
I'm starting to really think there may be an issue with the enclosure. Will call OWC in light of day. I almost hit "Remove Missing Secondary Disk" but got cold feet once it asked for my password to continue. A couple hours before sunrise it just seemed better to go to bed.
Remove missing secondary disk just removes all disks from the partition map that are not connected. (SoftRAID app would not know if the Mirror has or does not have an intentionally disconnected disk, we support up to 16 disks in a mirror volume)
If it does not happen immediately, quit SoftRAID and repeat. It should take zero time after you enter your password.
@softraid-support Thanks, just did this. Now I will physically label the disk trays with a label-maker, and Add Disk Label in the SoftRAID application. Then switch Disk 3 to Disk 7.
Started a support request with Mac Sales/OWC last night. They just asked for a Tech Support file from the i9 running SoftRAID 7 Pro, which I have sent.
Hope I'm not off-base but am running out of reasons to believe there is not an issue with my Flex 8. Worst part is if I can't reliably repeat it for proof, then getting any help from warranty may be another uphill battle. At this point I just need resolution. And another cancellation request for my business yesterday afternoon didn't exactly make me feel any better.
Don't swap 3 with 7. Swap 4 with 7. I am concerned the enclosure may be failing also, this will help isolate that.
One other thing is if you only have 4 drives in and plug in the enclosure. Do all the 4 drives always mount/show up?)
that would be a check if this is a power supply issue in the Flex. unusual issues can be harder to debug, sorry.
@softraid-support Thanks, that helps since I was getting ready to swap but had not just yet. First I labeled everything to get really organized moving forward. What does swapping 4 with 7 do? Just curious since both 4 and 7 were missing when this happened yesterday.
For the power supply check, does it matter which 4 drives? Top row? Bottom row? Any combination? For starters I'll just swap 4 with 7 until you tell me to move onto removing 4. I do wish disks not showing was more repeatable for troubleshooting.
Re: power supply, there's voltage regulators in the Flex right? Because of all the guitar work and being serious deep into tube amp repair/tech work purely for my own vintage amps, I know for a fact that wall voltage in Nashville runs too hot. I monitor it fairly regularly. There's such a market for it that there's a company that relocated here from the east coast just for the guitar community and builds a voltage attenuator to plug vintage amps into. I've seen wall voltage as high as 125v AC down to 119 in the summer when everyone's using their air conditioners.
Re: power supply. I didn't fully understand this a couple months ago but now it makes much more sense. This is a quote from a different thread before the holidays. I had not yet setup the Studio.
"SoftRaid has been slow to show disks in the left column and volumes in the right column for some time now."
SoftRAID shows the disks in real time, as they respond to MacOS. So either the drives are taking more time to spin up (less likely), or the power supply is weaker and taking more time to spin up. Something like that.
Just swapped 4 with 7 in the enclosure. Of course as soon as I plugged back into the Studio and powered up, everything is normal and disks a responding quickly. Short of data-loss, I think I'd prefer a repeatable consistent issue over the "once in a while" state of anxiety, never knowing when it's going to happen next. Based on the past few weeks though, it probably will.
Are disks 3 and 4 electrically isolated? Disk 4 also did not show yesterday (but now is showing just fine). Is it the top row for one set of power and the bottom row for the other? Or is each bay isolated from the other?
The way these enclosures work is AB, CD, EF, GH are isolated from each set. So 3-4 are connected (using the same PCI lanes)

