We have 2 ocfs2 filesystems on our 6 node cluster running OEL5.10 that were
created before indexed-dirs were active by default.
We were having lag in accessing sub directories that contained 10s of thousands
of files, so thought adding the indexes would help, and it did for D/Q, we ran
into this in our PROD environment.
[root at sahp9465 ~]#cat /etc/redhat-release
Red Hat Enterprise Linux Server release 5.10 (Tikanga)
[root at sahp9465 ~]#uname -r
2.6.32-400.33.3.el5uek
[root at sahp9465 ~]#rpm -qa |grep ocfs2
ocfs2-tools-1.8.0-11.el5
adding the indexes to the first filesystem went fine, but adding to the 2nd
resulted in errors
tunefs.ocfs2: Install directory trailer failed while enable indexed-dirs
tunefs.ocfs2: Install directory trailer failed while building indexed trees
The FS was fsck.ocfs2 prior to the attempt to add indexed-dirs and came back
clean.
I then fsck'd again and removed the feature as it said it was enabled, and I
didn't want corruption similar to as seen in :
https://oss.oracle.com/pipermail/ocfs2-users/2012-May/005749.html
[root at HOST ~]#fsck.ocfs2 -y /dev/mapper/oapr1p1
fsck.ocfs2 1.8.0
Checking OCFS2 filesystem in /dev/mapper/oapr1p1:
Label: oapr1p1
UUID: C257DA6F9C2E487D9AFD2602064AC2B3
Number of blocks: 131070310
Block size: 4096
Number of clusters: 131070310
Cluster size: 4096
Number of slots: 16
/dev/mapper/oapr1p1 is clean. It will be checked after 20 additional mounts.
[root at HOST ~]#tunefs.ocfs2 -Q "Features enabled = %H\n"
/dev/mapper/oapr1p1
Features enabled = sparse inline-data xattr
[root at HOST ~]#time tunefs.ocfs2 -vv --fs-features=indexed-dirs
/dev/mapper/oapr1p1
Enabling feature "indexed_dirs"
Opening device "/dev/mapper/oapr1p1"
Asking for 4096 blocks of I/O cache
Got 4096 blocks
Asking for 2048 blocks of I/O cache
Got 2048 blocks
Checking for dirty journals
Verifying the global allocator
Device "/dev/mapper/oapr1p1" opened
Opening device "/dev/mapper/oapr1p1"
Verifying the global allocator
Device "/dev/mapper/oapr1p1" opened
Opening device "/dev/mapper/oapr1p1"
Verifying the global allocator
Device "/dev/mapper/oapr1p1" opened
tunefs.ocfs2: Install directory trailer failed while enable indexed-dirs
tunefs.ocfs2: Install directory trailer failed while building indexed trees
Closing device "/dev/mapper/oapr1p1"
Device closed
Closing device "/dev/mapper/oapr1p1"
Device closed
Closing device "/dev/mapper/oapr1p1"
Verifying the global allocator
Device closed
[root at HOST ~]#time tunefs.ocfs2 -vv --fs-features=indexed-dirs
/dev/mapper/oapr1p1
Enabling feature "indexed_dirs"
Opening device "/dev/mapper/oapr1p1"
Asking for 4096 blocks of I/O cache
Got 4096 blocks
Asking for 2048 blocks of I/O cache
Got 2048 blocks
Checking for dirty journals
Verifying the global allocator
Device "/dev/mapper/oapr1p1" opened
Opening device "/dev/mapper/oapr1p1"
Verifying the global allocator
Device "/dev/mapper/oapr1p1" opened
Opening device "/dev/mapper/oapr1p1"
Verifying the global allocator
Device "/dev/mapper/oapr1p1" opened
Directory indexing feature is already enabled; nothing to enable
Closing device "/dev/mapper/oapr1p1"
Device closed
Closing device "/dev/mapper/oapr1p1"
Device closed
Closing device "/dev/mapper/oapr1p1"
Verifying the global allocator
Device closed
(reverse-i-search)`fsc': fsck.ocfs2 -y /dev/mapper/oapr1p1
[root at HOST ~]#fsck.ocfs2 -y /dev/mapper/oapr1p1
fsck.ocfs2 1.8.0
Checking OCFS2 filesystem in /dev/mapper/oapr1p1:
Label: oapr1p1
UUID: C257DA6F9C2E487D9AFD2602064AC2B3
Number of blocks: 131070310
Block size: 4096
Number of clusters: 131070310
Cluster size: 4096
Number of slots: 16
/dev/mapper/oapr1p1 is clean. It will be checked after 20 additional mounts.
[root at HOST ~]#
Display all 3129 possibilities? (y or n)
[root at HOST ~]#tunefs.ocfs2 -Q "Features enabled = %H\n"
/dev/mapper/oapr1p1
Features enabled = sparse inline-data xattr indexed-dirs
[root at HOST ~]#time tunefs.ocfs2 -vv --fs-features=noindexed-dirs
/dev/mapper/oapr1p1
Disabling feature "indexed_dirs"
Opening device "/dev/mapper/oapr1p1"
Asking for 4096 blocks of I/O cache
Got 4096 blocks
Asking for 2048 blocks of I/O cache
Got 2048 blocks
Checking for dirty journals
Verifying the global allocator
Device "/dev/mapper/oapr1p1" opened
Opening device "/dev/mapper/oapr1p1"
Verifying the global allocator
Device "/dev/mapper/oapr1p1" opened
Opening device "/dev/mapper/oapr1p1"
Verifying the global allocator
Device "/dev/mapper/oapr1p1" opened
We have 5 indexed directories to truncate.
"5" from "5" indexed directories are truncated.Closing
device "/dev/mapper/oapr1p1"
Device closed
Closing device "/dev/mapper/oapr1p1"
Device closed
Closing device "/dev/mapper/oapr1p1"
Verifying the global allocator
Device closed
real 21m43.166s
user 0m11.923s
sys 0m19.696s
[root at HOST ~]#tunefs.ocfs2 -Q "Features enabled = %H\n"
/dev/mapper/oapr1p1
Features enabled = sparse inline-data xattr
Any suggestions would be greatly appreciated.
Thanks,
JD
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
http://oss.oracle.com/pipermail/ocfs2-users/attachments/20140807/98153287/attachment.html
