[Ocfs2-tools-commits] branch, dlmfs-stackglue, created. ocfs2-tools-1.4.0-418-gcb4a06d
svn-commits at oss.oracle.com
svn-commits at oss.oracle.com
Wed Feb 10 01:16:47 PST 2010
This is an automated email from the git hooks/post-receive script. It was
generated because a ref change was pushed to the repository containing
the project "Tools to manage the ocfs2 filesystem.".
The branch, dlmfs-stackglue has been created
at cb4a06ddd849d9aa3351a1355396af1c8a3e85bc (commit)
- Log -----------------------------------------------------------------
commit cb4a06ddd849d9aa3351a1355396af1c8a3e85bc
Author: Joel Becker <joel.becker at oracle.com>
Date: Tue Feb 9 19:11:39 2010 -0800
libocfs2: Detect stackglue and use it for all libo2dlm interaction.
Now all ocfs2 tools will use libo2dlm+dlmfs for any stack that supports
it.
Signed-off-by: Joel Becker <joel.becker at oracle.com>
commit 3a2bbe83af2c1bf82e895e7299144a9b0cf9ee56
Author: Joel Becker <joel.becker at oracle.com>
Date: Tue Feb 9 13:41:58 2010 -0800
o2cb.init: Load dlmfs for user stacks when supported
If dlmfs supports user stacks, load and mount it at startup.
This is gated by the 'stackglue' capability. ocfs2_dlmfs reports this
as a module parameter named 'capabilities'. If ocfs2_dlmfs is built-in
or loaded, this will appear at
/sys/module/ocfs2_dlmfs/parameter/capabilities. If it is not loaded, we
can find it via modinfo.
Signed-off-by: Joel Becker <joel.becker at oracle.com>
commit 7eca205bbd8bd671908d5031d483f5e7751c1d79
Author: Joel Becker <joel.becker at oracle.com>
Date: Tue Feb 9 19:01:48 2010 -0800
libo2dlm: Don't drop locks at unlock.
Currently, libo2dlm unlinks files from dlmfs when a lock is unlocked.
This drops the lock in the DLM, losing the contents of the LVB and
removing any lock caching. While o2dlm holds onto a lock for a few
seconds, no DLM guarantees this will happen. We don't want that!
Instead, we change o2dlm_unlock() to leave the lock file around in
dlmfs. This is a truly cached lock. The calling program can explicitly
drop the lock with o2dlm_drop_lock(). Otherwise, the lock is dropped
when the domain is shut down.
lvb_torture now passes with fsdlm.
Signed-off-by: Joel Becker <joel.becker at oracle.com>
commit 1304485ee07f0420c45296e8f03f1aa0df602a92
Author: Joel Becker <joel.becker at oracle.com>
Date: Tue Feb 9 18:41:41 2010 -0800
libo2dlm: Detect invalid LVBs
When a DLM does recovery, it may have to invalidate the contents of an
LVB or two. The DLMs provide a method to detect this. dlmfs has
learned to report this via a zero-length read(2). libo2dlm now notices
the zero length and returns an error to the caller.
Signed-off-by: Joel Becker <joel.becker at oracle.com>
commit 80eb587cc2c38a2cf4bf4b7b31bdaa31d21ee8c6
Author: Joel Becker <joel.becker at oracle.com>
Date: Wed Jan 27 18:53:26 2010 -0800
libo2dlm: Add support for BASTs
ocfs2_dlmfs can now notify userspace of BASTs via poll(2) or select(2).
We need a way for libo2dlm users to actually access this information.
This change introduces two new functions. The first,
o2dlm_lock_with_bast(), is exactly like o2dlm_lock() except that it
takes a BAST function and returns a file descriptor to poll on. When
that file descriptor signals data is ready, the program calls the second
new function, o2dlm_process_bast(). This will fire off the appropriate
BAST.
If dlmfs isn't BAST capable, we'll get an error.
Signed-off-by: Joel Becker <joel.becker at oracle.com>
commit e57c18dda66a035f6ca71d2cf77efc7435ff27c6
Author: Joel Becker <joel.becker at oracle.com>
Date: Tue Feb 9 01:38:19 2010 -0800
libo2dlm: Provide capability queries
ocfs2_dlmfs can now report added capabilities via a module parameter.
libo2dlm now can query them before trying to use them.
The two known capabilities at this time are 'bast' and 'stackglue'.
Signed-off-by: Joel Becker <joel.becker at oracle.com>
commit 7e430159bd897f4087fe7fe88815578d1861fa7f
Merge: dc773865302e7a9b1b0fb3e2dfd2198cad70f8b2 f0d61e9ab703e5290b0f6780886c0aaae70813b2
Author: Joel Becker <joel.becker at oracle.com>
Date: Wed Jan 27 14:14:30 2010 -0800
Merge branch 'master' of /build/jlbec/ocfs2-tools/upstream/
commit dc773865302e7a9b1b0fb3e2dfd2198cad70f8b2
Author: Joel Becker <joel.becker at oracle.com>
Date: Tue Nov 17 12:44:03 2009 -0800
ocfs2_journal_check_available_features(): Never allow FEATURE_COMPAT_CHECKSUM
ocfs2 will never use jbd2 checksums. If a jbd2 checksum fails, the
transaction is not replayed. This means that some parts of the
transaction may have checkpointed while other parts have not. This
inconsistent state requires fsck to repair. ocfs2, being clustered,
can't tolerate that situation. Instead, ocfs2 prefers to replay the
journal. This allows the cluster to remain running. Using the METAECC
feature allows ocfs2 to detect and isolate any corrupted blocks.
Signed-off-by: Joel Becker <joel.becker at oracle.com>
-----------------------------------------------------------------------
hooks/post-receive
--
Tools to manage the ocfs2 filesystem.
More information about the Ocfs2-tools-commits
mailing list