From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.0.1) Gecko/20021003 Description of problem: We have a NexSAN ataboy RAID connected to our RH 8 server. One of the partitions (an EXT3 partition @ 961GB) reached 100% earlier this week. Since that time, data has been removed from the partition (over 30GB). when doing a df -k /export/imagery (the partition), df shows: [root@loki root]# df /export/imagery/ Filesystem 1K-blocks Used Available Use% Mounted on /dev/Imagery/lvol1 961211736 916595968 0 100% /export/imagery [root@loki root]# NOTE that used is not near the 961GB, yet it reads 'Available' at '0', and 'Use%' at '100%'. I can add and remove files locally as normal, but the df counters don't change. we are also having problems with NFS since this problem occured, where clients (in particular Solaris 8) cannot write files larger than a few bites to the partition. What I've Done:: I have unexported the partition and unmounted it. --- Then remounted it, and found the problem still there. I then rebooted the machine, and the problem still exists. I have attempted multiple times to fsck the partition, but it always tells me that the partition is clean. I have looked at tune2fs, and have been unable to find any options that might help my situation.. Please advise. Version-Release number of selected component (if applicable): How reproducible: Didn't try Steps to Reproduce: 1. umount partition 2. remount partition 3. problem still exists. 1. reboot machine 2. when partition is mounted, problem still exists. Additional info: I have put this bug Severity as High, because I fear that, as this is a filesystem problem, that it may lead to a partition data loss, or other extremely bad things. OUTPUT OF 'tune2fs -l /dev/Imagery/lvol1' [root@loki root]# tune2fs -l /dev/Imagery/lvol1 tune2fs 1.27 (8-Mar-2002) Filesystem volume name: <none> Last mounted on: <not available> Filesystem UUID: e85c895a-de26-4dc0-81c7-36296f42cc35 Filesystem magic number: 0xEF53 Filesystem revision #: 1 (dynamic) Filesystem features: has_journal filetype needs_recovery sparse_super large_file Filesystem state: clean Errors behavior: Continue Filesystem OS type: Linux Inode count: 122077184 Block count: 244133888 Reserved block count: 12206694 Free blocks: 11153942 Free inodes: 121931155 First block: 0 Block size: 4096 Fragment size: 4096 Blocks per group: 32768 Fragments per group: 32768 Inodes per group: 16384 Inode blocks per group: 512 Last mount time: Thu Sep 11 14:49:26 2003 Last write time: Thu Sep 11 14:49:26 2003 Mount count: 17 Maximum mount count: 37 Last checked: Tue Aug 26 09:40:42 2003 Check interval: 15552000 (6 months) Next check after: Sun Feb 22 08:40:42 2004 Reserved blocks uid: 0 (user root) Reserved blocks gid: 0 (group root) First inode: 11 Inode size: 128 Journal UUID: <none> Journal inode: 8 Journal device: 0x0000 First orphan inode: 0 [root@loki root]#
The % is % available to normal users. Since 5% is reserved for the superuser only, you still need to remove about 30GB (if I'm reading it right) for it to show as < 100% full. You can change the reserved percentage with tune2fs -m .