[Ocfs2-announce] OCFS2 1.4.7-1 and OCFS2 Tools 1.4.4-1 released

[Sunil Mushran]

All,

We are pleased to announce the release of OCFS2 1.4.7-1 and OCFS2 Tools
1.4.4-1 for Oracle's and Red Hat's Enterprise Linux 5 Update 2 and higher.

Oracle's Unbreakable Linux Network users who are subscribing to the "OCFS2
1.4 packages for Enterprise Linux 5" channel can upgrade to this release
by running up2date.
http://oss.oracle.com/pipermail/el-errata/2010-April/001438.html
http://oss.oracle.com/pipermail/el-errata/2010-April/001439.html

Red Hat's Enterprise Linux 5 users can download and install the relevant
file system and tools packages from oss.oracle.com.
http://oss.oracle.com/projects/ocfs2/files/RedHat/RHEL5/
http://oss.oracle.com/projects/ocfs2-tools/files/RedHat/RHEL5/

COMPATIBILITY

This release is fully compatible with earlier releases of OCFS2 1.4.
Users can upgrade their nodes to the new version in a rolling manner.

This release is on-disk compatible with OCFS2 1.2.x. Users can install
the software and mount the older volumes as-is. However, a rolling upgrade
from 1.2 to 1.4 will not work.

RECOMMENDATION

This is just to remind users to add the "noatime" mount option to the
mounts that hold the Oracle datafiles, redologs, archivelogs, voting file,
etc. This is for OCFS2 1.4 only.

WHAT'S CHANGED

This release includes mostly bug fixes.

The one new feature we've added is not much of one. It allows users
to change the fence method from the default of machine reset to panic.
This was requested by some developers who are interested in the vmcore
dump that is generated when a machine panics. So unless you want the
same, our recommendation would be for you to leave the fence method as-is.
Do note that the fence method of a node can be toggled between "reset"
and "panic" at anytime.

To view the current fence method, do:
	# cat /sys/kernel/config/cluster/CLUSTER/fence_method 
	reset

To change to panic, do:
	# echo panic > /sys/kernel/config/CLUSTER/cacl10/fence_method 
	# cat /sys/kernel/config/cluster/CLUSTER/fence_method 
	panic

The bug fixes can be classified under three groupings. The first group
involves cluster locking. Specifically in the area of downconverting
cluster locks. The links below explain two of the more interesting
problems. Our thanks to David Teigland of Red Hat for helping us fix
these problems.
http://oss.oracle.com/git/?p=ocfs2-1.4.git;a=commit;h=e8ef96c444326e4262fd371729e7beebda1af4d1
http://oss.oracle.com/git/?p=ocfs2-1.4.git;a=commit;h=39febfd5ee7948c018b667e0b909886e1cfa1235

The second group of bug fixes concern NFS support. This release fixes
a nfsd lockup issue and a stale inode read problem. Again, the links
below describe the problems in detail.
http://oss.oracle.com/git/?p=ocfs2-1.4.git;a=commit;h=2d561d3636c80af24063a74ae8c817661c574d78
http://oss.oracle.com/git/?p=ocfs2-1.4.git;a=commit;h=aa20775d1e7feba9b22e761fa9b69bd5c3f043bd

The last group of fixes concerns users encountering erroneous out-of-space
errors. Our analysis found that the errors were triggered because the file
system could not grow the extent block allocator because of free space
fragmentation. The extent block allocator houses the extent blocks that are
used when an inode needs more than approx 250 extents to describe a file.
So the way this plays out is that, in the early going, when free space is
contiguous, the inodes rarely use the extent blocks. They start getting
used just when the free space is fragmented enough that the extent allocator
cannot be grown.

The sad part is that the space required by this allocator is typically very
small. So small that there was no reason we could not allocate it up front.
In this release, the format tool, mkfs.ocfs2, reserves upto 0.3% of the volume
for this allocator. Furthermore, if the file system finds that this allocator
cannot be grown, it now can steal free blocks from another slot's allocator.
The first fix will help newly formatted volumes. The second fix will also
help existing volumes.

The final fix for this problem will be provided in the next patch update
(1.4.8). In it, we will allow the block allocators (inode and extent) to be
grown even when a 4MB contiguous chunk is not available. Users will be
able to activate this feature (discontiguous block groups) on existing
volumes. This feature is currently in testing.

BUGS FIXED

ossbz#970  Unfair postponement of local lock requests (Livelock)
ossbz#1175 BUG in dlm_free_dead_locks() (Oops during dlm recovery)
ossbz#1178 BUG in ocfs2_prepare_downconvert() (Oops during downconvert)
ossbz#1189 Free space trouble in a ocfs2 partition (ext_alloc cannot be grown)
ossbz#1202 dlm_add_lock_to_array:1192 ERROR: mismatched lvbs! (lock migration)

FEEDBACK

Please do not hesitate to email us at the ocfs2-users at oss.oracle.com
with any questions and comments.

The OCFS2 Team

OCFS2: http://oss.oracle.com/projects/ocfs2
TOOLS: http://oss.oracle.com/projects/ocfs2-tools
DOCS : http://oss.oracle.com/projects/ocfs2/documentation/