Stripe size issue o...
 
Notifications
Clear all

Stripe size issue or Drobo incompatibility

(@dcschooley)
Active Member

Hi,

I have two new Thunderbay 4's I'm using to replace a really old Promise Pegasus and a Drobo 5D3. My laptop is a MacBook Pro M1 Max running Monterey 12.6 and the Thunderbays are set up with RAID5 and 16k stripes. I'm transferring files over to the Thunderbays usiing Carbon Copy Cloner. I occasionally get kernel panics when transferring files from the Drobo to the larger of the two Thunderbays. The Pegasus to the smaller Thunderbay was never a problem. The MacBook Pro reboots but doesn't provide any log messages that I can find. Is this likely to be a Drobo problem, although I've never seen it before, or do I need to redo everything with 64k stripes?

I'm using SoftRaid 6.3

Quote
Topic starter Posted : 21/09/2022 1:06 pm
(@softraid-support)
Member Admin

This does not seem like the M1 panic bug. There, the panics would happen every restart immediately.

If this is only happening with the Drobo, then it could be a failure sign there.

Hopefully with CCC, you can see the progress, and what I would do is skip the next file/folder it is trying to copy, to get the rest.(Assuming there is just a bad sector) That is the best way to get around faulty areas on a disk.

 

let me know if this helps.

 

ReplyQuote
Posted : 21/09/2022 4:20 pm
(@dcschooley)
Active Member

There's no way really to see which file might be the problem because I am transferring multiple TB of data. The first time or two it happened quickly, but it's also done it after an hour or so. The Mac will go through several restarts after the fault happens before successfully staying up. Basically the Mac reboots while in the middle of a large transfer and then there are a couple of log-in followed by panic cycles. Sometimes the reboot doesn't occur until I start CCC back up. This is kind of hard to explain...

1. CCC is copying large amounts of data from Drobo to TB

2. Mac reboots

3. I log back in and restart CCC but don't restart the copy. Mac reboots. Do this one or two more times.

4. Mac finally stays up. 

 

It does seem to be one particular volume on the Drobo that is a problem. I'll know for certain later this evening. It's interesting that a file system problem on the Drobo would cause a panic, but if it is, then it's good I'm replacing it it. My initial thought was that it was related to the devices interacting somehow or a driver conflict. 

It doesn't matter if the Thunderbays are daisy-chained with the Drobo or on a separate channel. 

ReplyQuote
Topic starter Posted : 21/09/2022 5:12 pm
(@dcschooley)
Active Member

I've attached two different panic logs from this morning. 

ReplyQuote
Topic starter Posted : 22/09/2022 1:20 pm
(@softraid-support)
Member Admin

@dcschooley 

these are "DART" io controller crashes. A bug that Apple will fix in Ventura next month. (It is not being fixed for Monterey)

All you can do for now is see if waiting 5 minutes after startup prevents the panic, or temporarily removing one disk at startup, adding it back after 5 minutes.

else you need to delete the volume, and create it with 64k stripe unit sizes, which almost always works.

This post was modified 2 months ago by SoftRAID Support
ReplyQuote
Posted : 22/09/2022 2:52 pm
(@softraid-support)
Member Admin

Its interesting that your panics are not immediate, but after copying data. Most users with DART crashes, it happens instantly.

 

Maybe you can try removing one of the RAID 5 disks. See if you can copy, then when done, push the disk in and let SoftRAID rebuild the volume?

ReplyQuote
Posted : 22/09/2022 3:03 pm
(@dcschooley)
Active Member

Yes, it's been weird. The smaller Thunderbay has 16TB and that one didn't have any problems when I copied data over from the Pegasus. I didn't add the larger 64TB Thunderbay until the Pegasus was almost done. As I mentioned earlier, the copy from the Drobo to the 64TB ThunderBay went for a while and then crashed, which is what made me think the Drobo was a contributor to the problem. I've rebuilt the RAIDs several times, and the Mac panicked immediately after one of the rebuilds. I think there were at least two other times where I got panics. The pattern was always the same, with an initial panic, and logging back in would cause a near immediate panic and this would repeat itself a couple of times and then the system would hold. The locations of the different devices on the Thunderbolt buses seems to make some difference, but it's hard to say. 

In any case, I've switched both RAIDs to 64k stripes and everything seems to be working. Would there be any benefit to changing them back to 16k after Ventura comes out?

Thank you for your help!

ReplyQuote
Topic starter Posted : 22/09/2022 5:55 pm
(@softraid-support)
Member Admin

@dcschooley 

I have been testing 16k vs. 64k and no there is no significant reason to change back.

ReplyQuote
Posted : 22/09/2022 10:55 pm
Share:
close
open