Bug 14696

Summary: EXT4-fs error (device dm-3): ext4_mb_release_inode_pa: free 777, pa_free 778
Product: File System Reporter: Krzysztof Oledzki (ole)
Component: ext4Assignee: fs_ext4 (fs_ext4)
Status: CLOSED OBSOLETE    
Severity: high CC: alan
Priority: P1    
Hardware: All   
OS: Linux   
Kernel Version: 2.6.30.8 Subsystem:
Regression: No Bisected commit-id:

Description Krzysztof Oledzki 2009-11-26 11:32:10 UTC
Two hours ago my database server stopped working. I turned out to be a problem with ext4:

pa ffff88063643bd00: logic 1024, phys. 2677760, len 1024
EXT4-fs error (device dm-3): ext4_mb_release_inode_pa: free 777, pa_free 778
Aborting journal on device dm-3:8.
Remounting filesystem read-only

The problem was fixed by:

# fsck /dev/VG0/sql
fsck from util-linux-ng 2.16.1
e2fsck 1.41.9 (22-Aug-2009)
/dev/mapper/VG0-sql_tmp: recovering journal
/dev/mapper/VG0-sql_tmp contains a file system with errors, check forced.
Pass 1: Checking inodes, blocks, and sizes
Deleted inode 66 has zero dtime.  Fix<y>? yes

Inodes that were part of a corrupted orphan linked list found.  Fix<y>? yes

Inode 67 was part of the orphaned inode list.  FIXED.
Inode 68 was part of the orphaned inode list.  FIXED.
Inode 69 was part of the orphaned inode list.  FIXED.
Inode 2975 was part of the orphaned inode list.  FIXED.
Deleted inode 3731 has zero dtime.  Fix<y>? yes

Inode 3732 was part of the orphaned inode list.  FIXED.
Inode 3733 was part of the orphaned inode list.  FIXED.
Inode 3734 was part of the orphaned inode list.  FIXED.
Inode 3735 was part of the orphaned inode list.  FIXED.
Pass 2: Checking directory structure
Pass 3: Checking directory connectivity
Pass 4: Checking reference counts
Pass 5: Checking group summary information
Block bitmap differences:  -2678117
Fix<y>? yes

Inode bitmap differences:  -(66--69) -2975 -(3731--3735)
Fix<y>? yes

Free inodes count wrong for group #0 (4457, counted=4467).
Fix<y>? yes

Free inodes count wrong (2093400, counted=2093410).
Fix<y>? yes


/dev/mapper/VG0-sql_tmp: ***** FILE SYSTEM WAS MODIFIED *****
/dev/mapper/VG0-sql_tmp: 3742/2097152 files (19.4% non-contiguous), 963461/8388608 blocks

Before I started the fsck I had made a copy of that fs, so I'm still able to provide a dump if necessary.
Comment 1 Krzysztof Oledzki 2009-12-13 07:23:28 UTC
My fs has just crashed again:

pa ffff88063643b410: logic 1024, phys. 8079360, len 1024
EXT4-fs error (device dm-3): ext4_mb_release_inode_pa: free 1000, pa_free 1001
Aborting journal on device dm-3:8.
Remounting filesystem read-only

# fsck -f /var/sql
fsck from util-linux-ng 2.16.1
e2fsck 1.41.9 (22-Aug-2009)
/dev/mapper/VG0-sql: recovering journal
Clearing orphaned inode 2975 (uid=60, gid=60, mode=0100600, size=0)
Clearing orphaned inode 69 (uid=60, gid=60, mode=0100600, size=0)
Clearing orphaned inode 68 (uid=60, gid=60, mode=0100600, size=0)
Clearing orphaned inode 67 (uid=60, gid=60, mode=0100600, size=20)
Clearing orphaned inode 66 (uid=60, gid=60, mode=0100600, size=0)
Pass 1: Checking inodes, blocks, and sizes
Pass 2: Checking directory structure
Pass 3: Checking directory connectivity
Pass 4: Checking reference counts
Pass 5: Checking group summary information
Block bitmap differences:  -8080068
Fix<y>? yes


/dev/mapper/VG0-sql: ***** FILE SYSTEM WAS MODIFIED *****
/dev/mapper/VG0-sql: 6099/2097152 files (22.8% non-contiguous), 1069002/8388608 blocks
Comment 2 Alan 2012-06-18 12:45:12 UTC
Closing as obsolete - ext4 has changed extensively since 2.6.30

If this is still seen with modern kernels please update the bug