The replacement for the OWC 32TB PCIe array arrived, and I sent the first one back to Illinois. I powered down the workstation, removed the external drives, popped the cover (so easy on the model), and installed the new array. I put the skin back on, connected the external drives, plugged everything into the UPS, and hit the p0wer button. The workstation came up, after the usual fairly long delay while it POSTed 512 GB of DRAM.
I opened SoftRAID. It couldn’t see the array.
I opened Disk Manager. It couldn’t see the new array either, but the wasn’t the worst thing: the OWC 32 TB Thunderbolt 3 array show as “Failed”. I rescanned the drives, and after an hour and a quarter, I was still looking at the hourglass. I disconnected the Thunderbolt 3 array so I could get on with the PCIe card installation.
No amount of fiddling around helped, so I called OWC customer support. They couldn’t help much with an error this fundamental. They suggested running their diagnostics. I downloaded the diagnostics program and it hung. I rebooted. The machine refused to shut down, so I forced a power cycle. The computer failed to restart, displaying a screen that did me no good because I couldn’t select either off the two options because of my Bluetooth keyboard and mouse. I power cycled again, and Windows came up.
When the computer came up after the reboot, the diagnostics program ran, and I emailed the results to OWC. I peeked at the files, and saw that the diagnostics recognized the array. While OWC examined the diagnostics report, I fired up SoftRAID again, and saw that it could now see the array.
But there was a problem:
I knew right away what was wrong. OWC, in their infinite wisdom, had shipped me an array preconfigured in such a fashion that it wouldn’t run under Windows. For shame. But that had happened to me before, so I knew what to do: delete all the partitions and volumes, and rebuild the array with SoftRAID.
I opened Disk Manager again, saw the eight SSD drives in the array each with 3 partitions, for a total of 24 partitions. I started to tear down the partitions with Disk Manager while I waited for the tech to come back on the line. He agreed that was the right thing to do, and stayed on the phone while I deleted all the partitions. Then SoftRAID saw 8 disks that it could make into an array. I told it to take all 8 drives and make a striped array. Formatting took about 5 minutes, but it was finally happy.
I took the failed OWC 4M2 Thunderbolt 3 array over to my Lenovo P16 laptop. I showed up as failed there, too, but the Device Manager could see the individual drives.
I left the 4M2 plugged in while I instructed GoodSync to transfer over about 16 TB of data from a 32 TB OWC Thunderbolt 4 array on the Lenovo computer. The transfer ran for about 4 hours and failed. When I looked at the Lenovo, the screen was dark and I couldn’t wake it up. I removed the 4M2 array and forced a power cycle. This was turning into a habit. The next GoodSync transfer ran for about 12 hours before it failed, but the failure wasn’t due to the P16 dying, but to SoftRAID loosing track of the disk array. A reboot fixed that, and the rest of the transfer went fine.
For you GoodSync users, the sweet spot for maximizing transfer speeds over a 2.5 GbE connection for 50-500 MB image files is 8 to 10 files in parallel, and the average speed is about 220 MB/s.
So what’s up with the failed array? More on that later.
Leave a Reply