We cannot restart recovery. Once we begin to recover a node, keep the state
of the recovery intact and follow through, regardless of any other node
deaths that may occur.
Signed-off-by: Kurt Hackel <kurt.hackel@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
If the previous master of the recovery lock dies, let calc_usage take it
down completely and let the caller completely redo the dlmlock() call.
Otherwise, there will never be an opportunity to re-master the lockres and
recovery wont be able to progress.
Signed-off-by: Kurt Hackel <kurt.hackel@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
Use the existing structure for blocking migrations when ASTs are pending to
achieve the same result. If we can catch the assert before it goes on the
wire, just cancel it and let the migration continue.
Signed-off-by: Kurt Hackel <kurt.hackel@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
Now we never change the owner of a lock resource until unmount or node
death. This will be re-enabled once some issues in the algorithm used have
been resolved.
Signed-off-by: Kurt Hackel <kurt.hackel@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
In dlmlock_remote(), do not call purge_lockres until the lock resource
actually changes. otherwise, the mastery info on the lockres will go away
underneath the caller.
Signed-off-by: Kurt Hackel <kurt.hackel@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
When mastering non-recovery lock resources, additional time was frequently
needed to allow the disk heartbeat to catch up with the network timeout. the
recovery lock resource is time critical and avoids this path.
Signed-off-by: Kurt Hackel <kurt.hackel@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
Recovery will spin in dlm_pre_master_reco_lockres if we do not ignore
timed-out network responses from dead nodes.
Signed-off-by: Kurt Hackel <kurt.hackel@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
Change behavior of dlm_restart_lock_mastery() when a node goes down. Dump
all responses that have been collected and start over.
Signed-off-by: Kurt Hackel <kurt.hackel@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
This is an error on the sending side, so gracefully error out on the
receiving end.
Signed-off-by: Kurt Hackel <kurt.hackel@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
Immediately purge a lockress that the local node is not the master of.
Signed-off-by: Kurt Hackel <kurt.hackel@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
Makes it easier for the recovery process to deal with node death.
Signed-off-by: Kurt Hackel <kurt.hackel@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
Take a reference on lockres structures while they are on the recovery list.
Signed-off-by: Kurt Hackel <kurt.hackel@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
handle errors during lock assert master by either killing self or other node
Signed-off-by: Kurt Hackel <kurt.hackel@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
The check for an empty lvb should check the entire buffer not just the first
byte.
Signed-off-by: Kurt Hackel <kurt.hackel@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
Recovery may have happened and it may now be mastered locally.
Signed-off-by: Kurt Hackel <kurt.hackel@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
The OCFS2 DLM allocates a number of pages for a hash to lookup locks.
There was a bug where a PAGE_SIZE bigger than the hash size (eg, 64K
pages) would result in zero pages allocated.
Signed-off-by: Joel Becker <joel.becker@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
This allows us to have a hash table greater than a single page which greatly
improves dlm performance on some tests.
Signed-off-by: Daniel Phillips <phillips@google.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
Gains us a bit of performance on loads which heavily hit the lockres hash.
Patch suggested by Daniel Phillips <phillips@google.com>.
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
This patch converts the combination of list_del(A) and list_add(A, B) to
list_move(A, B) under fs/.
Cc: Ian Kent <raven@themaw.net>
Acked-by: Joel Becker <joel.becker@oracle.com>
Cc: Neil Brown <neilb@cse.unsw.edu.au>
Cc: Hans Reiser <reiserfs-dev@namesys.com>
Cc: Urban Widmark <urban@teststation.com>
Acked-by: David Howells <dhowells@redhat.com>
Acked-by: Mark Fasheh <mark.fasheh@oracle.com>
Signed-off-by: Akinobu Mita <mita@miraclelinux.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Extend the get_sb() filesystem operation to take an extra argument that
permits the VFS to pass in the target vfsmount that defines the mountpoint.
The filesystem is then required to manually set the superblock and root dentry
pointers. For most filesystems, this should be done with simple_set_mnt()
which will set the superblock pointer and then set the root dentry to the
superblock's s_root (as per the old default behaviour).
The get_sb() op now returns an integer as there's now no need to return the
superblock pointer.
This patch permits a superblock to be implicitly shared amongst several mount
points, such as can be done with NFS to avoid potential inode aliasing. In
such a case, simple_set_mnt() would not be called, and instead the mnt_root
and mnt_sb would be set directly.
The patch also makes the following changes:
(*) the get_sb_*() convenience functions in the core kernel now take a vfsmount
pointer argument and return an integer, so most filesystems have to change
very little.
(*) If one of the convenience function is not used, then get_sb() should
normally call simple_set_mnt() to instantiate the vfsmount. This will
always return 0, and so can be tail-called from get_sb().
(*) generic_shutdown_super() now calls shrink_dcache_sb() to clean up the
dcache upon superblock destruction rather than shrink_dcache_anon().
This is required because the superblock may now have multiple trees that
aren't actually bound to s_root, but that still need to be cleaned up. The
currently called functions assume that the whole tree is rooted at s_root,
and that anonymous dentries are not the roots of trees which results in
dentries being left unculled.
However, with the way NFS superblock sharing are currently set to be
implemented, these assumptions are violated: the root of the filesystem is
simply a dummy dentry and inode (the real inode for '/' may well be
inaccessible), and all the vfsmounts are rooted on anonymous[*] dentries
with child trees.
[*] Anonymous until discovered from another tree.
(*) The documentation has been adjusted, including the additional bit of
changing ext2_* into foo_* in the documentation.
[akpm@osdl.org: convert ipath_fs, do other stuff]
Signed-off-by: David Howells <dhowells@redhat.com>
Acked-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: Nathan Scott <nathans@sgi.com>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Remove the code which attempted to catch it via dlmunlock() return status -
this never happens there.
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
Don't BUG() user_dlm_unblock_lock() on the absence of the USER_LOCK_BLOCKED
flag - this turns out to be a valid case. Make some of the related BUG()
statements print more useful information.
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
when starting lock mastery (excepting the recovery lock) wait on any nodes
needing recovery. fix one instance where lock resources were left attached
to the recovery list after recovery completed. ensure that the node_down
code is run uniformly regardless of which node found the dead node first.
Signed-off-by: Kurt Hackel <kurt.hackel@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
fixes hangs in lock mastery related to refcounting on the mle structure
Signed-off-by: Kurt Hackel <kurt.hackel@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
Rewrap the overly long source code lines resulting from the previous
patch's addition of the slab cache flag SLAB_MEM_SPREAD. This patch
contains only formatting changes, and no function change.
Signed-off-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Mark file system inode and similar slab caches subject to SLAB_MEM_SPREAD
memory spreading.
If a slab cache is marked SLAB_MEM_SPREAD, then anytime that a task that's
in a cpuset with the 'memory_spread_slab' option enabled goes to allocate
from such a slab cache, the allocations are spread evenly over all the
memory nodes (task->mems_allowed) allowed to that task, instead of favoring
allocation on the node local to the current cpu.
The following inode and similar caches are marked SLAB_MEM_SPREAD:
file cache
==== =====
fs/adfs/super.c adfs_inode_cache
fs/affs/super.c affs_inode_cache
fs/befs/linuxvfs.c befs_inode_cache
fs/bfs/inode.c bfs_inode_cache
fs/block_dev.c bdev_cache
fs/cifs/cifsfs.c cifs_inode_cache
fs/coda/inode.c coda_inode_cache
fs/dquot.c dquot
fs/efs/super.c efs_inode_cache
fs/ext2/super.c ext2_inode_cache
fs/ext2/xattr.c (fs/mbcache.c) ext2_xattr
fs/ext3/super.c ext3_inode_cache
fs/ext3/xattr.c (fs/mbcache.c) ext3_xattr
fs/fat/cache.c fat_cache
fs/fat/inode.c fat_inode_cache
fs/freevxfs/vxfs_super.c vxfs_inode
fs/hpfs/super.c hpfs_inode_cache
fs/isofs/inode.c isofs_inode_cache
fs/jffs/inode-v23.c jffs_fm
fs/jffs2/super.c jffs2_i
fs/jfs/super.c jfs_ip
fs/minix/inode.c minix_inode_cache
fs/ncpfs/inode.c ncp_inode_cache
fs/nfs/direct.c nfs_direct_cache
fs/nfs/inode.c nfs_inode_cache
fs/ntfs/super.c ntfs_big_inode_cache_name
fs/ntfs/super.c ntfs_inode_cache
fs/ocfs2/dlm/dlmfs.c dlmfs_inode_cache
fs/ocfs2/super.c ocfs2_inode_cache
fs/proc/inode.c proc_inode_cache
fs/qnx4/inode.c qnx4_inode_cache
fs/reiserfs/super.c reiser_inode_cache
fs/romfs/inode.c romfs_inode_cache
fs/smbfs/inode.c smb_inode_cache
fs/sysv/inode.c sysv_inode_cache
fs/udf/super.c udf_inode_cache
fs/ufs/super.c ufs_inode_cache
net/socket.c sock_inode_cache
net/sunrpc/rpc_pipe.c rpc_inode_cache
The choice of which slab caches to so mark was quite simple. I marked
those already marked SLAB_RECLAIM_ACCOUNT, except for fs/xfs, dentry_cache,
inode_cache, and buffer_head, which were marked in a previous patch. Even
though SLAB_RECLAIM_ACCOUNT is for a different purpose, it marks the same
potentially large file system i/o related slab caches as we need for memory
spreading.
Given that the rule now becomes "wherever you would have used a
SLAB_RECLAIM_ACCOUNT slab cache flag before (usually the inode cache), use
the SLAB_MEM_SPREAD flag too", this should be easy enough to maintain.
Future file system writers will just copy one of the existing file system
slab cache setups and tend to get it right without thinking.
Signed-off-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Switch from list_head to hlist_head. Make the size of the hash dependent
upon the allocated area, rather than a constant.
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
* fix a hang in recovery that occurred in dlmlock_remote. the $RECOVERY
lock was never moved to the granted queue even after getting DLM_NORMAL
back from the master node.
Signed-off-by: Kurt Hackel <kurt.hackel@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
* add dlm_wait_for_node_death function to be used after receiving a network
error. this will wait for the given timeout to allow the heartbeat
callbacks to update the domain map. without this, some paths may spin
and consume enough cpu that the heartbeat gets starved and never updates.
Signed-off-by: Kurt Hackel <kurt.hackel@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
* fix a bug in dlm_convert_lock_handler where dlm_lockres_release_ast was
being called even if no ast was ever reserved
Signed-off-by: Kurt Hackel <kurt.hackel@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
* after successfully taking the $RECOVERY lock in EX mode, recheck to make
sure that recovery has not already begun or completed on another node
Signed-off-by: Kurt Hackel <kurt.hackel@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
fs/ocfs2/dlm/dlmrecovery.c does now use msleep(), and does therefore
need to #include <linux/delay.h> for getting the prototype of this
function.
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
* fix a hang which can occur during shutdown migration
* do not allow nodes to join during recovery
* when restarting lock mastery, do not ignore nodes which come up
* more than one node could become recovery master, fix this
* sleep to allow some time for heartbeat state to catch up to network
* extra debug info for bad recovery state problems
* make DLM_RECO_NODE_DATA_DONE a valid state for non-master recovery nodes
* prune all locks from dead nodes on $RECOVERY lock resources
* do NOT automatically add new nodes to mle nodemaps until they have properly
joined the domain
* make sure dlm_pick_recovery_master only exits when all nodes have synced
* properly handle dlmunlock errors in dlm_pick_recovery_master
* do not propagate network errors in dlm_send_begin_reco_message
* dead nodes were not being put in the recovery map sometimes, fix this
* dlmunlock was failing to clear the unlock actions on DLM_DENIED
Signed-off-by: Kurt Hackel <kurt.hackel@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
Including <asm/signal.h> results in compilation failure on ia64 due to
not including <linux/compiler.h>
Including <linux/signal.h> corrects the problem.
Please apply.
Signed-off-by: Jeff Mahoney <jeffm@suse.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
- cluster/sys.c: make needlessly global code static
- dlm/: "extern" declarations for variables belong into header files
(and in this case, they are already in dlmdomain.h)
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
dlmfs: A minimal dlm userspace interface implemented via a virtual
file system.
Most of the OCFS2 tools make use of this to take cluster locks when
doing operations on the file system.
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
Signed-off-by: Kurt Hackel <kurt.hackel@oracle.com>
A distributed lock manager built with the cluster file system use case
in mind. The OCFS2 dlm exposes a VMS style API, though things have
been simplified internally. The only lock levels implemented currently
are NLMODE, PRMODE and EXMODE.
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
Signed-off-by: Kurt Hackel <kurt.hackel@oracle.com>
