402blownstroker
[H]ard|DCer of the Month - Nov. 2012
- Joined
- Jan 5, 2006
- Messages
- 3,257
Currently I have a zfs on linux setup. Over the last two months I have seen a couple of stability issues. I am currently scrubbing the pool nightly and capacity is around 450GB of 3TB. The pool is made up as follows:
1) Transferring large files, 10+ GB, causes the system to hang from 30 seconds to 10+ minutes. The machine does respond to pings, but it can not be accessed: slogin, nfs, or console( it is non responsive ).
2) I moved one of the SSD cache drives to a log drive to improve write performance and things went to complete hell. Basically any type of pool access would cause kernel panics. This would happen with local or remote access, read or write operations. I moved the SSD back from log to cache and things appear to be work again.
In both cases zpool status reported no issues, problems, corrupted data, etc. Are these issues with ZFS on linux? Would native solaris perform better and be more stable?
I am also most things about going back to good olde RAID5 or RAID50.
- (4) 1TB WD Black Drives in RAIDZ
- (2) 128GB SSD as cache drives
- 9 Filesystems all with sharenfs=on set
1) Transferring large files, 10+ GB, causes the system to hang from 30 seconds to 10+ minutes. The machine does respond to pings, but it can not be accessed: slogin, nfs, or console( it is non responsive ).
2) I moved one of the SSD cache drives to a log drive to improve write performance and things went to complete hell. Basically any type of pool access would cause kernel panics. This would happen with local or remote access, read or write operations. I moved the SSD back from log to cache and things appear to be work again.
In both cases zpool status reported no issues, problems, corrupted data, etc. Are these issues with ZFS on linux? Would native solaris perform better and be more stable?
I am also most things about going back to good olde RAID5 or RAID50.