I am running ZFS filesystem version 4 and storage pool version 15 on a FreeBSD 
8.2-Release-amd64 kernel. I have a single 12TB pool based on a 3ware 9650 
controller with 8 seagate ST2000DL003 drives in a raid-5 configuration managed 
by the controller. 

I recently had a connector problem on a disk in the array while running a 
performance test that was writing a 1TB pattern file to the array. When the 
raid controller started reporting errors I stopped the test and re-seated the 
connector on the drive. After running a verify on the raid, I tried to read the 
partial pattern file and ZFS produced copious amounts of checksum error 
messages on the system console. So, I rm'ed the file, and got even more 
checksum errors interspersed with several I/O error 86 messages. Since the rm, 
ls no longer shows the file, but I did a scrub just to be sure the bogus file 
was gone, and got tons of checksum and i/o 86 errors. At the end, zpool status 

phoenix# zpool status -v zfsPool 
pool: zfsPool 
state: ONLINE 
status: One or more devices has experienced an error resulting in data 
corruption. Applications may be affected. 
action: Restore the file in question if possible. Otherwise restore the 
entire pool from backup. 
see: http://www.sun.com/msg/ZFS-8000-8A 
scrub: scrub completed after 3h40m with 6353 errors on Fri Jun 22 08:36:36 2012 

zfsPool ONLINE 0 0 6.20K 
da0 ONLINE 0 0 12.4K 

errors: Permanent errors have been detected in the following files: 


I have tried "zpool clear"/reboot/"zpool scrub" several times now, and get a 
similar set of errors and results. 

My question is - How do I get rid of this file? It is no longer linked to a 
directory entry, and there shouldn't be anybody with it open since I have 
rebooted several times. yet, zfs still tells me there's a broken file and I 
should replace it. It is most likely the pattern test file that I deleted, so I 
don't need it and I don't want to recover it. i would just like to get rid of 
it and get my filesystem clean again without resorting to starting over. 


zfs-discuss mailing list

Reply via email to