Hello!
We run several systems of this configuration, here:
OS: Debian Sarge
Kernel: vanilla 2.6.16+ (+ Areca driver patch)
RAID: Areca ARC-1160 (Raid 6) or 2x 3ware 9500S-8 (Raid 5) + SW RAID 0
HDD: 15x WD or Seagate
The partition is about 3.3T and we use no extra options for mount.
xfs_info /path
meta-data=/path isize=256 agcount=32,
agsize=26855466 blks
= sectsz=512
data = bsize=4096 blocks=859374912,
imaxpct=25
= sunit=0 swidth=0 blks, unwritten=1
naming =version 2 bsize=4096
log =internal bsize=4096 blocks=32768, version=1
= sectsz=512 sunit=0 blks
realtime =none extsz=65536 blocks=0, rtextents=0
The xfs filesystems get corrupted from time to time. Syslog is being
filled up with errors, then (sample attached), and the system load goes
up (I am not sure, what exactly happens there). So,
- the kernel is filling up syslog and the filesystem is still mounted RW
- when the system is rebooted, it mounts the xfs, without forcing the
check (though there is non-zero value in the fs_passno column
in /etc/fstab)
- errors are logged again
- explicit run of xfs_repair finds and repairs several issues
Can anyone explain to me, what (and why is that :-) is going on with the
filesystem?
Why the corrupted fs is still used? In RW manner???
TIA,
ViNiL
xfslog
Description: Text document
pgpFSrn8Uj5vx.pgp
Description: PGP signature
|