- 18 Apr, 2008 40 commits
-
-
Tao Ma authored
Currently, o2net connects to a node on hb_up and disconnects on hb_down and net timeout. It disconnects on net timeout is ok, but it should attempt to reconnect back. This is because sometimes nodes get overloaded enough that the network connection breaks but the disk hb does not. And if we get into that situation, we either fence (unnecessarily) or wait for its disk hb to die (and sometimes hang in the process). So in this updated scheme, when the network disconnects, we keep attempting to reconnect till we succeed or we get a disk hb down event. If the other node is really dead, then we will eventually get a node down event. If not, we should be able to connect again and continue. Signed-off-by:
Tao Ma <tao.ma@oracle.com> Signed-off-by:
Mark Fasheh <mfasheh@suse.com>
-
Sunil Mushran authored
A previous patch added KERN_NOTICE to printks printing the lockres that cluttered the output. This patch removes the log level. For people concerned with syslog clutter, please note we now use this facility to print lockres only during an error. Signed-off-by:
Sunil Mushran <sunil.mushran@oracle.com> Signed-off-by:
-
Sunil Mushran authored
__dlm_print_one_lock_resource was printing lockname incorrectly. Also, we now use printk directly instead of mlog as the latter prints the line context which is not useful for this print. Signed-off-by:
Joel Becker <joel.becker@oracle.com> Signed-off-by:
-
Sunil Mushran authored
This patch helps in consolidating debugging related functions in dlmdebug.c. Signed-off-by:
-
Sunil Mushran authored
This patch dumps all the lockres' on the purgelist it can fit in one page into a debugfs file. Useful for debugging. Signed-off-by:
-
Sunil Mushran authored
This patch dumps all mles it can fit in one page into a debugfs file. Useful for debugging. Signed-off-by:
-
Sunil Mushran authored
This patch moves some mle related definitions from dlmmaster.c to dlmcommon.h. Future patches need these definitions to dump mle debugging information. Signed-off-by:
-
Sunil Mushran authored
This patch dumps all the lockres' alongwith all the locks into a debugfs file. Useful for debugging. Signed-off-by:
-
Sunil Mushran authored
This patch dumps the dlm state (dlm_ctxt) into a debugfs file. Useful for debugging. Signed-off-by:
-
Sunil Mushran authored
This patch creates the debugfs directories that will hold the files to be used to dump the dlm state. Signed-off-by:
-
Sunil Mushran authored
This patch links all the lockres' to a tracking list in dlm_ctxt. We will use this in an upcoming patch that will walk the entire list and to dump the lockres states to a debugfs file. Signed-off-by:
-
Sunil Mushran authored
This patch makes the o2dlm allocate memory for lockres, lockname and lock structures from slabcaches rather than kmalloc. This allows us to not only make these allocs more efficient but also allows us to track the memory being consumed by these structures. Signed-off-by:
-
Sunil Mushran authored
This patch renames dlm_mle_slabcache to prevent namespace clashes with fs/dlm. Signed-off-by:
-
Joel Becker authored
Add ABI documentation for these files: /sys/fs/ocfs2/max_locking_protocol /sys/fs/ocfs2/loaded_cluster_plugins /sys/fs/ocfs2/active_cluster_plugin /sys/fs/ocfs2/cluster_stack Signed-off-by:
-
Joel Becker authored
ocfs2 now supports plug-ins for the classic O2CB stack as well as userspace cluster stacks in conjunction with fs/dlm. This allows zero, one, or both of the plug-ins to be selected in Kconfig. For local mounts (non-clustered), neither plug-in is needed. Both plugins can be loaded at one time, the runtime will select the one needed for the cluster systme in use. Signed-off-by:
-
Joel Becker authored
Add ocfs2_stack_user.ko to the Makefile so that it builds. Signed-off-by:
-
Joel Becker authored
The masklog code is in the o2cb stack, but ocfs2_lockid.h now needs to be included by the user stack. The BUG() in ocfs2_lock_type_string() does not need masklog support, so change it to a regular BUG_ON(). Signed-off-by:
-
David Teigland authored
Add code to use fs/dlm. [ Modified to be part of the stack_user module -- Joel ] Signed-off-by:
David Teigland <teigland@redhat.com> Signed-off-by:
-
Joel Becker authored
The "SETV" message sets the filesystem locking protocol version as negotiated by the client. The client negotiates based on the maximum version advertised in /sys/fs/ocfs2/max_locking_protocol. Signed-off-by:
-
Joel Becker authored
This is the second part of the ocfs2_control handshake. After negotiating the ocfs2_control protocol, the daemon tells the filesystem what the local node id is via the SETN message. Signed-off-by:
-
Joel Becker authored
When the control daemon sees a node go down, it sends a DOWN message through the ocfs2_control device. Signed-off-by:
-
Joel Becker authored
When a control daemon opens the ocfs2_control device, it must perform a handshake to tell the filesystem it is something capable of monitoring cluster status. Only after the handshake is complete will the filesystem allow mounts. This is the first part of the handshake. The daemon reads all supported ocfs2_control protocols, then writes in the protocol it will use. Signed-off-by:
-
Joel Becker authored
The ocfs2_control misc device is how a userspace control daemon (controld) talks to the filesystem. Introduce the bare-bones filesystem ops. Signed-off-by:
-
Joel Becker authored
Add a skeleton for the stack_user module. It's just the barebones module code. Signed-off-by:
-
Joel Becker authored
Userspace can now query and specify the cluster stack in use via the /sys/fs/ocfs2/cluster_stack file. By default, it is 'o2cb', which is the classic stack. Thus, old tools that do not know how to modify this file will work just fine. The stack cannot be modified if there is a live filesystem. ocfs2_cluster_connect() now takes the expected cluster stack as an argument. This way, the filesystem and the stack glue ensure they are speaking to the same backend. If the stack is 'o2cb', the o2cb stack plugin is used. For any other value, the fsdlm stack plugin is selected. Signed-off-by:
-
Joel Becker authored
The filesystem gains the USERSPACE_STACK incomat bit and the s_cluster_info field on the superblock. When a userspace stack is in use, the name of the stack is stored on-disk for mount-time verification. The "cluster_stack" option is added to mount(2) processing. The mount process needs to pass the matching stack name. If the passed name and the on-disk name do not match, the mount is failed. When using the classic o2cb stack, the incompat bit is *not* set and no mount option is used other than the usual heartbeat=local. Thus, the filesystem is compatible with older tools. Signed-off-by:
-
Joel Becker authored
Introduce a set of sysfs files that describe the current stack glue state. The files live under /sys/fs/ocfs2. The locking_protocol file displays the version of ocfs2's locking code. The loaded_cluster_plugins file displays all of the currently loaded stack plugins. When filesystems are mounted, the active_cluster_plugin file will display the plugin in use. Signed-off-by:
-
Joel Becker authored
We define the ocfs2_stack_plugin structure to represent a stack driver. The o2cb stack code is split into stack_o2cb.c. This becomes the ocfs2_stack_o2cb.ko module. The stackglue generic functions are similarly split into the ocfs2_stackglue.ko module. This module now provides an interface to register drivers. The ocfs2_stack_o2cb driver registers itself. As part of this interface, ocfs2_stackglue can load drivers on demand. This is accomplished in ocfs2_cluster_connect(). ocfs2_cluster_disconnect() is now notified when a _hangup() is pending. If a hangup is pending, it will not release the driver module and will let _hangup() do that. Signed-off-by: -
Joel Becker authored
Define the ocfs2_stack_operations structure. Build o2cb_stack_ops from all of the o2cb-specific stack functions. Change the generic stack glue functions to call the stack_ops instead of the o2cb functions directly. The o2cb functions are moved to stack_o2cb.c. The headers are cleaned up to where only needed headers are included. In this code, stackglue.c and stack_o2cb.c refer to some shared extern variables. When they become modules, that will change. Signed-off-by:
-
Joel Becker authored
Split off the o2cb-specific funtionality from the generic stack glue calls. This is a precurser to wrapping the o2cb functionality in an operations vector. Signed-off-by:
-
Joel Becker authored
The stack glue initialization function needs a better name so that it can be used cleanly when stackglue becomes a module. Signed-off-by:
-
Joel Becker authored
dlmglue.c was still referencing a raw o2dlm lksb in one instance. Let's create a generic ocfs2_dlm_dump_lksb() function. This allows underlying DLMs to print whatever they want about their lock. We then move the o2dlm dump into stackglue.c where it belongs. Signed-off-by:
-
David Teigland authored
When using fsdlm, -EAGAIN is returned in the async callback for NOQUEUE requests. Fix up dlmglue to expect this. Signed-off-by:
-
Joel Becker authored
o2dlm has the non-standard behavior of providing a cancel callback (unlock_ast) even when the cancel has failed (the locking operation succeeded without canceling). This is called CANCELGRANT after the status code sent to the callback. fs/dlm does not provide this callback, so dlmglue must be changed to live without it. o2dlm_unlock_ast_wrapper() in stackglue now ignores CANCELGRANT calls. Because dlmglue no longer sees CANCELGRANT, ocfs2_unlock_ast() no longer needs to check for it. ocfs2_locking_ast() must catch that a cancel was tried and clear the cancel state. Making these changes opens up a locking race. dlmglue uses the the OCFS2_LOCK_BUSY flag to ensure only one thread is calling the dlm at any one time. But dlmglue must unlock the lockres before calling into the dlm. In the small window of time between unlocking the lockres and calling the dlm, the downconvert thread can try to cancel the lock. The downconvert thread is checking the OCFS2_LOCK_BUSY flag - it doesn't know that ocfs2_dlm_lock() has not yet been called. Because ocfs2_dlm_lock() has not yet been called, the cancel operation will just be a no-op. There's nothing to cancel. With CANCELGRANT, dlmglue uses the CANCELGRANT callback to clear up the cancel state. When it comes around again, it will retry the cancel. Eventually, the first thread will have called into ocfs2_dlm_lock(), and either the lock or the cancel will succeed. The downconvert thread can then do its downconvert. Without CANCELGRANT, there is nothing to clean up the cancellation state. The downconvert thread does not know to retry its operations. More importantly, the original lock may be blocking on the other node that is trying to cancel us. With neither able to make progress, the ast is never called and the cancellation state is never cleaned up that way. dlmglue is deadlocked. The OCFS2_LOCK_PENDING flag is introduced to remedy this window. It is set at the same time OCFS2_LOCK_BUSY is. Thus, the downconvert thread can check whether the lock is cancelable. If not, it just loops around to try again. Once ocfs2_dlm_lock() is called, the thread then clears OCFS2_LOCK_PENDING and wakes the downconvert thread. Now, if the downconvert thread finds the lock BUSY, it can safely try to cancel it. Whether the cancel works or not, the state will be properly set and the lock processing can continue. Signed-off-by:
-
Mark Fasheh authored
It doesn't make sense to query for a node number before connecting to the cluster stack. This should be safe to do because node_num is only just printed, and we're actually only moving the setting of node num a small amount further in the mount process. [ Disconnect when node query fails -- Joel ] Reviewed-by:
-
Joel Becker authored
The last bit of classic stack used directly in ocfs2 code is o2hb. Specifically, the check for heartbeat during mount and the call to ocfs2_hb_ctl during unmount. We create an extra API, ocfs2_cluster_hangup(), to encapsulate the call to ocfs2_hb_ctl. Other stacks will just leave hangup() empty. The check for heartbeat is moved into ocfs2_cluster_connect(). It will be matched by a similar check for other stacks. With this change, only stackglue.c includes cluster/ headers. Signed-off-by:
-
Joel Becker authored
ocfs2 asks the cluster stack for the local node's node number for two reasons; to fill the slot map and to print it. While the slot map isn't necessary for userspace cluster stacks, the printing is very nice for debugging. Thus we add ocfs2_cluster_this_node() as a generic API to get this value. It is anticipated that the slot map will not be used under a userspace cluster stack, so validity checks of the node num only need to exist in the slot map code. Otherwise, it just gets used and printed as an opaque value. [ Fixed up some "int" versus "unsigned int" issues and made osb->node_num truly opaque. --Mark ] Signed-off-by:
-
Joel Becker authored
This step introduces a cluster stack agnostic API for initializing and exiting. fs/ocfs2/dlmglue.c no longer uses o2cb/o2dlm knowledge to connect to the stack. It is all handled in stackglue.c. heartbeat.c no longer needs to know how it gets called. ocfs2_do_node_down() is now a clean recovery trigger. The big gotcha is the ordering of initializations and de-initializations done underneath ocfs2_cluster_connect(). ocfs2_dlm_init() used to do all o2dlm initialization in one block. Thus, the o2dlm functionality of ocfs2_cluster_connect() is very straightforward. ocfs2_dlm_shutdown(), however, did a few things between de-registration of the eviction callback and actually shutting down the domain. Now de-registration and shutdown of the domain are wrapped within the single ocfs2_cluster_disconnect() call. I've checked the code paths to make sure we can safely tear down things in ocfs2_dlm_shutdown() before calling ocfs2_cluster_disconnect(). The filesystem has already set itself to ignore the callback. Signed-off-by:
-
Joel Becker authored
Wrap the lock status block (lksb) in a union. Later we will add a union element for the fs/dlm lksb. Create accessors for the status and lvb fields. Other than a debugging function, dlmglue.c does not directly reference the o2dlm locking path anymore. Signed-off-by:
-
Joel Becker authored
Change the ocfs2_dlm_lock/unlock() functions to return -errno values. This is the first step towards elminiating dlm_status in fs/ocfs2/dlmglue.c. The change also passes -errno values to ->unlock_ast(). [ Fix a return code in dlmglue.c and change the error translation table into an array of ints. --Mark ] Signed-off-by:
-
