(2017-01-05, 10:56)Deepjungle Wrote: Hello,
are there some news about the problems with large partitions > 8.4TB ?
Greetings
Deepjungle
Not yet.
I do have a question though - a while ago you posted the following:
(2016-12-31, 12:31)Deepjungle Wrote: I did a fresh install on a second raspi 3 with another RAID System. Same case but with 4x4TB Seagate drives. With 1022 it is working after upgrading to 1031 in this case it did not get detected.
1022 dmesg log
http://sprunge.us/KSJO
1031 dmesg log
http://sprunge.us/MWYT
Subsequently you identified there is some sort of 8.4TB limit with the latest kernel (not sure what that is about, yet).
However, with your "working" #1022 system (4.8.2 kernel), the 4x4TB drives are only being recognised as having a capacity of 9TB (8.19TiB):
Code:
[ 0.000000] Linux version 4.8.2 ([email protected]) (gcc version 5.4.0 (GCC) ) #1 SMP Sat Oct 22 21:05:20 BST 2016
...
[ 9.840137] sd 0:0:0:0: [sda] 17581277184 512-byte logical blocks: (9.00 TB/8.19 TiB)
whereas with the #1031 system (4.8.6 kernel) the - I'm assuming - same disks have a capacity of 12TB (10.9TiB):
Code:
[ 0.000000] Linux version 4.8.6 ([email protected]) (gcc version 5.4.0 (GCC) ) #1 SMP Mon Oct 31 21:22:52 GMT 2016
...
[ 9.733719] sd 0:0:0:0: [sda] 23441768448 512-byte logical blocks: (12.0 TB/10.9 TiB)
Can you confirm that the drives were exactly the same in this test, and that only the RPi build is being changed?
If so, and assuming there is some sort of 8.4TB-ish limit somewhere (kernel, filesystem, USB/SATA bridge hardware), then perhaps the older 4.8.2 kernel is working because it is under-reporting (or under-recognising?) the total disk capacity - god knows if this might have then led to corruption of some sort once you started filling the disks.
Perhaps the more recent kernel (4.8.4 since #1023) has corrected that bug, and refuses to mount the drives for some reason (corrupt?) There's no indication in the #1022 and #1031 dmesg logs of the disks being mounted - can you run "journalctl -a | pastebinit" after the #1031 (or #latest) system has been up for about 2 minutes (or once you connect the drive).
Also, you said you tested with NTFS and could only mount up to 8,388.239 MB (I assume you mean GB - 8.388TB). Have you tested with ext4 as the file system, to see if the failure to mount is a limitation of NTFS (or the way this NTFS file system is presented to the kernel)? The disk size limit for NTFS with 4KB clusters is 16TB, but it would be worth trying a different filesystem to see if there is any different behaviour.