prep for flex file layout server
Subsystems (2)
cover prep for flex file layout server
These patches are in preparation for the pynfs implementation of a flex file layout server.
I already have patches for a working flex file server, but I figure those should wait until the spec is complete ;)
I pushed these patches to my linux-nfs.org tree on branch 'ff_prep':
git://git.linux-nfs.org/projects/dros/pynfs.git
The prep includes:
- patches that enable pynfs mds using real v4.1 DSes (on the backend, not usable from clients as there is no stateid sharing between mds and ds)
- fix several bugs in pnfs path
- move .x files to subdir 'xdrdef' - the nfs4.1 directory was getting crowded and we're going to add several more .x files along with all of the autogenerated files. Note this change touches server tests too.
- clean up the nfs4server's output - no longer print out EVERY part of EVERY operation which was very efficient at hiding tracebacks and made pynfs even slower. Old functionality is now switched on with -v flag. Also added an in-between mode "-s" that prints a summary of whats going on.
- support for NFSv3 backend communication - this is not actually used yet, but can be tested with file layout by always returning NFS4ERR_LAYOUTUNAVAILABLE in layoutget and using DataServer3 class instead of DataServer41
Weston Andros Adamson (17):
4.1 client: reclaim_complete after create_session 4.1 server: service RECLAIM_COMPLETE operations
dataserver: only catch connection error 4.1 server: avoid traceback in DS disconnect() move .x files to subdir 'xdrdef'
4.1 client: remove unused imports
4.1 server: add -v flag & silence random output 4.1 server: add -s option to print summary of ops
dataserver: make generic interface to ops
dataserver: don't import * from nfs4 specific mods 4.1 server: move nfs4_ops.py to nfs_ops.py add mntv3, portmapv2 and nfsv3 .x files
dataserver: separate generic and 4.1 code 4.1 server: add support for NFSv3 data servers 4.1 server: get rid of old op_getdeviceinfo
rpc: on socket error, close and mark pipe inactive
nfs3clnt: reconnect when sending on inactive pipe
nfs4.1/block.py | 8 +- nfs4.1/client41tests/environment.py | 4 +- nfs4.1/config.py | 8 +- nfs4.1/dataserver.py | 306 ++- nfs4.1/fs.py | 38 +- nfs4.1/nfs3client.py | 185 ++ nfs4.1/nfs4.x | 3290 --------------------------- nfs4.1/nfs4.x.diff | 48 - nfs4.1/nfs4_ops.py | 61 - nfs4.1/nfs4client.py | 48 +- nfs4.1/nfs4commoncode.py | 10 +- nfs4.1/nfs4lib.py | 62 +- nfs4.1/nfs4server.py | 166 +- nfs4.1/nfs4state.py | 15 +- nfs4.1/nfs_ops.py | 89 + nfs4.1/pnfs_block.x | 149 -- nfs4.1/sctrl.x | 145 -- nfs4.1/server41tests/environment.py | 4 +- nfs4.1/server41tests/st_block.py | 4 +- nfs4.1/server41tests/st_compound.py | 4 +- nfs4.1/server41tests/st_create_session.py | 4 +- nfs4.1/server41tests/st_current_stateid.py | 8 +- nfs4.1/server41tests/st_debug.py | 4 +- nfs4.1/server41tests/st_delegation.py | 4 +- nfs4.1/server41tests/st_destroy_clientid.py | 2 +- nfs4.1/server41tests/st_destroy_session.py | 4 +- nfs4.1/server41tests/st_exchange_id.py | 4 +- nfs4.1/server41tests/st_getdevicelist.py | 4 +- nfs4.1/server41tests/st_lookup.py | 2 +- nfs4.1/server41tests/st_lookupp.py | 2 +- nfs4.1/server41tests/st_open.py | 8 +- nfs4.1/server41tests/st_putfh.py | 2 +- nfs4.1/server41tests/st_reboot.py | 4 +- nfs4.1/server41tests/st_reclaim_complete.py | 2 +- nfs4.1/server41tests/st_rename.py | 4 +- nfs4.1/server41tests/st_secinfo.py | 4 +- nfs4.1/server41tests/st_secinfo_no_name.py | 4 +- nfs4.1/server41tests/st_sequence.py | 4 +- nfs4.1/server41tests/st_trunking.py | 4 +- nfs4.1/server41tests/st_verify.py | 2 +- nfs4.1/server_exports.py | 7 +- nfs4.1/setup.py | 33 +- nfs4.1/testclient.py | 2 +- nfs4.1/testserver.py | 2 +- nfs4.1/xdrdef/__init__.py | 0 nfs4.1/xdrdef/mnt3.x | 68 + nfs4.1/xdrdef/nfs3.x | 818 +++++++ nfs4.1/xdrdef/nfs4.x | 3290 +++++++++++++++++++++++++++ nfs4.1/xdrdef/nfs4.x.diff | 48 + nfs4.1/xdrdef/pnfs_block.x | 149 ++ nfs4.1/xdrdef/portmap.x | 51 + nfs4.1/xdrdef/sctrl.x | 145 ++ rpc/rpc.py | 26 +- 53 files changed, 5337 insertions(+), 4022 deletions(-) create mode 100644 nfs4.1/nfs3client.py delete mode 100644 nfs4.1/nfs4.x delete mode 100644 nfs4.1/nfs4.x.diff delete mode 100644 nfs4.1/nfs4_ops.py create mode 100644 nfs4.1/nfs_ops.py delete mode 100644 nfs4.1/pnfs_block.x delete mode 100644 nfs4.1/sctrl.x create mode 100644 nfs4.1/xdrdef/__init__.py create mode 100644 nfs4.1/xdrdef/mnt3.x create mode 100644 nfs4.1/xdrdef/nfs3.x create mode 100644 nfs4.1/xdrdef/nfs4.x create mode 100644 nfs4.1/xdrdef/nfs4.x.diff create mode 100644 nfs4.1/xdrdef/pnfs_block.x create mode 100644 nfs4.1/xdrdef/portmap.x create mode 100644 nfs4.1/xdrdef/sctrl.x
Signature · 1 line
1.8.5.2 (Apple Git-48)
PATCH 1/17 4.1 client: reclaim_complete after create_session
Send RECLAIM_COMPLETE after CREATE_SESSION. This enables backend communication to v4.1 linux nfsd servers (i/o through MDS only).
Signed-off-by Weston Andros Adamson <[email protected]>
--- nfs4.1/nfs4client.py | 1 + 1 file changed, 1 insertion(+)
nfs4.1/nfs4client.py +1 -0
@@ -367,6 +367,7 @@ class ClientRecord(object): self.seqid = inc_u32(csr.csr_sequence) # XXX Do we need to check this? sess = SessionRecord(csr, self) self.c.sessions[sess.sessionid] = sess + sess.compound([op.reclaim_complete(FALSE)]) return sess def _cb_hook(self, prefix, opname, funct):
Signature · 1 line
1.8.5.2 (Apple Git-48)
Re: [PATCH pynfs 01/17] 4.1 client: reclaim_complete after create_session
Quoted 2 lines · Weston Andros Adamson
Send
RECLAIM_COMPLETEafterCREATE_SESSION. This enables backend communication to v4.1 linux nfsd servers (i/o through MDS only).
If that's unconditionally adding a reclaim_complete after every create_sesion, that's probably not what we want since there are tests that want to do the reclaim_complete by hand. You probably want to use new_client_session() instead? See 219cf7b7f7d0 and the preceding commit.
--b.
Quoted 21 lines
Signed-off-by Weston Andros Adamson <[email protected]>--- nfs4.1/nfs4client.py | 1 + 1 file changed, 1 insertion(+)nfs4.1/nfs4client.py +1
-0@@ -367,6 +367,7 @@ class ClientRecord(object): self.seqid = inc_u32(csr.csr_sequence) # XXX Do we need to check this? sess = SessionRecord(csr, self) self.c.sessions[sess.sessionid] = sess + sess.compound([op.reclaim_complete(FALSE)]) return sess def _cb_hook(self, prefix, opname, funct):Signature · 1 line
1.8.5.2 (Apple Git-48)
Re: [PATCH pynfs 01/17] 4.1 client: reclaim_complete after create_session
Quoted 11 lines · J. Bruce Fields
Quoted 2 lines · Weston Andros Adamson
Send
RECLAIM_COMPLETEafterCREATE_SESSION. This enables backend communication to v4.1 linux nfsd servers (i/o through MDS only).If that's unconditionally adding a
reclaim_completeafter everycreate_sesion, that's probably not what we want since there are tests that want to do thereclaim_completeby hand. You probably want to usenew_client_session()instead? See219cf7b7f7d0and the preceding commit.—b.
Ok, thats a good point, but AFAIC new_client_session is never called - at least by nfs4server.
I’ll figure out a better way to do this.
-dros
Quoted 22 lines
Quoted 22 lines
Signed-off-by Weston Andros Adamson <[email protected]>---
nfs4.1/nfs4client.py| 1 +1 file changed, 1 insertion(+)nfs4.1/nfs4client.py +1
-0@@ -367,6 +367,7 @@ class ClientRecord(object): self.seqid = inc_u32(csr.csr_sequence) # XXX Do we need to check this? sess = SessionRecord(csr, self) self.c.sessions[sess.sessionid] = sess + sess.compound([op.reclaim_complete(FALSE)]) return sess def _cb_hook(self, prefix, opname, funct):Signature · 1 line
1.8.5.2 (Apple Git-48)
PATCH 2/17 4.1 server: service RECLAIM_COMPLETE operations
In reply to Weston Andros AdamsonJust return ok!
Signed-off-by Weston Andros Adamson <[email protected]>
--- nfs4.1/nfs4server.py | 3 +++ 1 file changed, 3 insertions(+)
nfs4.1/nfs4server.py +3 -0
@@ -1809,6 +1809,9 @@ class NFS4Server(rpc.Server): with find_state(env, arg.deleg_stateid, allow_0=False) as state: state.delegreturn() return encode_status(NFS4_OK) + + def op_reclaim_complete(self, arg, env): + return encode_status(NFS4_OK) def op_getdevicelist(self, arg, env): # STUB check_session(env)
Signature · 1 line
1.8.5.2 (Apple Git-48)
Re: [PATCH pynfs 02/17] 4.1 server: service RECLAIM_COMPLETE operations
Quoted 1 line · Weston Andros Adamson
Just return ok!
Technically it should record whether or not the reclaim_complete has happened and return a GRACE error on any non-reclaim open performed before the reclaim_complete--but for your purposes you may not care...
--b.
Quoted 23 lines
Signed-off-by Weston Andros Adamson <[email protected]>--- nfs4.1/nfs4server.py | 3 +++ 1 file changed, 3 insertions(+)nfs4.1/nfs4server.py +3
-0@@ -1809,6 +1809,9 @@ class NFS4Server(rpc.Server): with find_state(env, arg.deleg_stateid, allow_0=False) as state: state.delegreturn() return encode_status(NFS4_OK) + + def op_reclaim_complete(self, arg, env): + return encode_status(NFS4_OK) def op_getdevicelist(self, arg, env): # STUB check_session(env)Signature · 1 line
1.8.5.2 (Apple Git-48)
Re: [PATCH pynfs 02/17] 4.1 server: service RECLAIM_COMPLETE operations
Quoted 7 lines · J. Bruce Fields
Quoted 1 line · Weston Andros Adamson
Just return ok!
Technically it should record whether or not the
reclaim_completehas happened and return a GRACE error on any non-reclaim open performed before the reclaim_complete--but for your purposes you may not care...
...and a NOGRACE error on any reclaim opens performed by that client after the reclaim_complete?
Quoted 29 lines
--b.
Quoted 21 lines
Signed-off-by Weston Andros Adamson <[email protected]>--- nfs4.1/nfs4server.py | 3 +++ 1 file changed, 3 insertions(+)nfs4.1/nfs4server.py +3
-1@@ -1809,6 +1809,9 @@ class NFS4Server(rpc.Server): with find_state(env, arg.deleg_stateid, allow_0=False) as state: state.delegreturn() return encode_status(NFS4_OK) + + def op_reclaim_complete(self, arg, env): + return encode_status(NFS4_OK) def op_getdevicelist(self, arg, env): # STUB check_session(env) --Quoted 2 lines
1.8.5.2 (Apple Git-48)
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH pynfs 02/17] 4.1 server: service RECLAIM_COMPLETE operations
Are you saying that the pynfs server supports state recovery? This has not been my experience. I’ll double check.
-dros
Quoted 41 lines · Trond Myklebust
Quoted 7 lines · J. Bruce Fields
Quoted 1 line · Weston Andros Adamson
Just return ok!
Technically it should record whether or not the
reclaim_completehas happened and return a GRACE error on any non-reclaim open performed before the reclaim_complete--but for your purposes you may not care......and a NOGRACE error on any reclaim opens performed by that client after the
reclaim_complete?Quoted 29 lines
--b.
Quoted 21 lines
Signed-off-by Weston Andros Adamson <[email protected]>---
nfs4.1/nfs4server.py| 3 +++1 file changed, 3 insertions(+)nfs4.1/nfs4server.py +3
-1@@ -1809,6 +1809,9 @@ class NFS4Server(rpc.Server): with find_state(env, arg.deleg_stateid, allow_0=False) as state: state.delegreturn() return encode_status(NFS4_OK) + + def op_reclaim_complete(self, arg, env): + return encode_status(NFS4_OK) def op_getdevicelist(self, arg, env): # STUB check_session(env) --Quoted 2 lines
1.8.5.2 (Apple Git-48)
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH pynfs 02/17] 4.1 server: service RECLAIM_COMPLETE operations
Quoted 2 lines · Weston Andros Adamson
Are you saying that the pynfs server supports state recovery? This has not been my experience. I’ll double check.
If you don't support state recovery, then I think the minimal correct behavior would be to have no grace period at all: return NO_GRACE on *every* reclaim operation and GRACE only on non-reclaims not preceded by a global (one_fs == FALSE) RECLAIM_COMPLETE for that client.
All this does is catch misbehaving clients, and maybe that's not a priority. But it's easy enough to implement.
--b.
Quoted 48 lines
-dros
Quoted 41 lines · Trond Myklebust
Quoted 7 lines · J. Bruce Fields
Quoted 1 line · Weston Andros Adamson
Just return ok!
Technically it should record whether or not the
reclaim_completehas happened and return a GRACE error on any non-reclaim open performed before the reclaim_complete--but for your purposes you may not care......and a NOGRACE error on any reclaim opens performed by that client after the
reclaim_complete?Quoted 29 lines
--b.
Quoted 21 lines
Signed-off-by Weston Andros Adamson <[email protected]>---
nfs4.1/nfs4server.py| 3 +++1 file changed, 3 insertions(+)nfs4.1/nfs4server.py +3
-1@@ -1809,6 +1809,9 @@ class NFS4Server(rpc.Server): with find_state(env, arg.deleg_stateid, allow_0=False) as state: state.delegreturn() return encode_status(NFS4_OK) + + def op_reclaim_complete(self, arg, env): + return encode_status(NFS4_OK) def op_getdevicelist(self, arg, env): # STUB check_session(env) --Quoted 2 lines
1.8.5.2 (Apple Git-48)
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH pynfs 02/17] 4.1 server: service RECLAIM_COMPLETE operations
Quoted 11 lines · J. Bruce Fields
Quoted 2 lines · Weston Andros Adamson
Are you saying that the pynfs server supports state recovery? This has not been my experience. I’ll double check.
If you don't support state recovery, then I think the minimal correct behavior would be to have no grace period at all: return
NO_GRACEon *every* reclaim operation and GRACE only on non-reclaims not preceded by a global (one_fs== FALSE)RECLAIM_COMPLETEfor that client.All this does is catch misbehaving clients, and maybe that's not a priority. But it's easy enough to implement.
Yeah, that sounds good.
-dros
Quoted 51 lines
—b.
Quoted 48 lines
-dros
Quoted 41 lines · Trond Myklebust
Quoted 7 lines · J. Bruce Fields
Quoted 1 line · Weston Andros Adamson
Just return ok!
Technically it should record whether or not the
reclaim_completehas happened and return a GRACE error on any non-reclaim open performed before the reclaim_complete--but for your purposes you may not care......and a NOGRACE error on any reclaim opens performed by that client after the
reclaim_complete?Quoted 29 lines
--b.
Quoted 21 lines
Signed-off-by Weston Andros Adamson <[email protected]>---
nfs4.1/nfs4server.py| 3 +++1 file changed, 3 insertions(+)nfs4.1/nfs4server.py +3
-1@@ -1809,6 +1809,9 @@ class NFS4Server(rpc.Server): with find_state(env, arg.deleg_stateid, allow_0=False) as state: state.delegreturn() return encode_status(NFS4_OK) + + def op_reclaim_complete(self, arg, env): + return encode_status(NFS4_OK) def op_getdevicelist(self, arg, env): # STUB check_session(env) --Quoted 2 lines
1.8.5.2 (Apple Git-48)
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH pynfs 02/17] 4.1 server: service RECLAIM_COMPLETE operations
On second thought, I’m just going to drop this patch.
I only added it to avoid a NFS4ERR_NOTSUPP when connecting the file layout MDS to pynfs DSes, but it is harmless and outside the scope of what I’m doing.
Not worth that can of worms.
-dros
Quoted 71 lines · Weston Andros Adamson
Quoted 11 lines · J. Bruce Fields
Quoted 2 lines · Weston Andros Adamson
Are you saying that the pynfs server supports state recovery? This has not been my experience. I’ll double check.
If you don't support state recovery, then I think the minimal correct behavior would be to have no grace period at all: return
NO_GRACEon *every* reclaim operation and GRACE only on non-reclaims not preceded by a global (one_fs== FALSE)RECLAIM_COMPLETEfor that client.All this does is catch misbehaving clients, and maybe that's not a priority. But it's easy enough to implement.
Yeah, that sounds good.
-dros
Quoted 51 lines
—b.
Quoted 48 lines
-dros
Quoted 41 lines · Trond Myklebust
Quoted 7 lines · J. Bruce Fields
Quoted 1 line · Weston Andros Adamson
Just return ok!
Technically it should record whether or not the
reclaim_completehas happened and return a GRACE error on any non-reclaim open performed before the reclaim_complete--but for your purposes you may not care......and a NOGRACE error on any reclaim opens performed by that client after the
reclaim_complete?Quoted 29 lines
--b.
Quoted 21 lines
Signed-off-by Weston Andros Adamson <[email protected]>---
nfs4.1/nfs4server.py| 3 +++1 file changed, 3 insertions(+)nfs4.1/nfs4server.py +3
-1@@ -1809,6 +1809,9 @@ class NFS4Server(rpc.Server): with find_state(env, arg.deleg_stateid, allow_0=False) as state: state.delegreturn() return encode_status(NFS4_OK) + + def op_reclaim_complete(self, arg, env): + return encode_status(NFS4_OK) def op_getdevicelist(self, arg, env): # STUB check_session(env) --Quoted 2 lines
1.8.5.2 (Apple Git-48)
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH pynfs 02/17] 4.1 server: service RECLAIM_COMPLETE operations
Quoted 6 lines · Weston Andros Adamson
On second thought, I’m just going to drop this patch.
I only added it to avoid a
NFS4ERR_NOTSUPPwhen connecting the file layout MDS to pynfs DSes, but it is harmless and outside the scope of what I’m doing.Not worth that can of worms.
The MDS shouldn't really have to handle NOTSUPP on RECLAIM_COMPLETE (even if yours currently happens to). I think your no-op patch as it is would be better than nothing.
--b.
Quoted 79 lines
-dros
Quoted 71 lines · Weston Andros Adamson
Quoted 11 lines · J. Bruce Fields
Quoted 2 lines · Weston Andros Adamson
Are you saying that the pynfs server supports state recovery? This has not been my experience. I’ll double check.
If you don't support state recovery, then I think the minimal correct behavior would be to have no grace period at all: return
NO_GRACEon *every* reclaim operation and GRACE only on non-reclaims not preceded by a global (one_fs== FALSE)RECLAIM_COMPLETEfor that client.All this does is catch misbehaving clients, and maybe that's not a priority. But it's easy enough to implement.
Yeah, that sounds good.
-dros
Quoted 51 lines
—b.
Quoted 48 lines
-dros
Quoted 41 lines · Trond Myklebust
Quoted 7 lines · J. Bruce Fields
Quoted 1 line · Weston Andros Adamson
Just return ok!
Technically it should record whether or not the
reclaim_completehas happened and return a GRACE error on any non-reclaim open performed before the reclaim_complete--but for your purposes you may not care......and a NOGRACE error on any reclaim opens performed by that client after the
reclaim_complete?Quoted 29 lines
--b.
Quoted 21 lines
Signed-off-by Weston Andros Adamson <[email protected]>---
nfs4.1/nfs4server.py| 3 +++1 file changed, 3 insertions(+)nfs4.1/nfs4server.py +3
-1@@ -1809,6 +1809,9 @@ class NFS4Server(rpc.Server): with find_state(env, arg.deleg_stateid, allow_0=False) as state: state.delegreturn() return encode_status(NFS4_OK) + + def op_reclaim_complete(self, arg, env): + return encode_status(NFS4_OK) def op_getdevicelist(self, arg, env): # STUB check_session(env) --Quoted 2 lines
1.8.5.2 (Apple Git-48)
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH pynfs 02/17] 4.1 server: service RECLAIM_COMPLETE operations
Well, it works fine without it - the pynfs MDS just ignores the NOTSUPP!
If you really want it I can submit it later, but it’s unneeded and unrelated to the goals of this patchset.
-dros
Quoted 93 lines · J. Bruce Fields
Quoted 6 lines · Weston Andros Adamson
On second thought, I’m just going to drop this patch.
I only added it to avoid a
NFS4ERR_NOTSUPPwhen connecting the file layout MDS to pynfs DSes, but it is harmless and outside the scope of what I’m doing.Not worth that can of worms.
The MDS shouldn't really have to handle NOTSUPP on
RECLAIM_COMPLETE(even if yours currently happens to). I think your no-op patch as it is would be better than nothing.--b.
Quoted 79 lines
-dros
Quoted 71 lines · Weston Andros Adamson
Quoted 11 lines · J. Bruce Fields
Quoted 2 lines · Weston Andros Adamson
Are you saying that the pynfs server supports state recovery? This has not been my experience. I’ll double check.
If you don't support state recovery, then I think the minimal correct behavior would be to have no grace period at all: return
NO_GRACEon *every* reclaim operation and GRACE only on non-reclaims not preceded by a global (one_fs== FALSE)RECLAIM_COMPLETEfor that client.All this does is catch misbehaving clients, and maybe that's not a priority. But it's easy enough to implement.
Yeah, that sounds good.
-dros
Quoted 51 lines
—b.
Quoted 48 lines
-dros
Quoted 41 lines · Trond Myklebust
Quoted 7 lines · J. Bruce Fields
Quoted 1 line · Weston Andros Adamson
Just return ok!
Technically it should record whether or not the
reclaim_completehas happened and return a GRACE error on any non-reclaim open performed before the reclaim_complete--but for your purposes you may not care......and a NOGRACE error on any reclaim opens performed by that client after the
reclaim_complete?Quoted 29 lines
--b.
Quoted 21 lines
Signed-off-by Weston Andros Adamson <[email protected]>---
nfs4.1/nfs4server.py| 3 +++1 file changed, 3 insertions(+)nfs4.1/nfs4server.py +3
-1@@ -1809,6 +1809,9 @@ class NFS4Server(rpc.Server): with find_state(env, arg.deleg_stateid, allow_0=False) as state: state.delegreturn() return encode_status(NFS4_OK) + + def op_reclaim_complete(self, arg, env): + return encode_status(NFS4_OK) def op_getdevicelist(self, arg, env): # STUB check_session(env) --Quoted 2 lines
1.8.5.2 (Apple Git-48)
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
PATCH 3/17 dataserver: only catch connection error
In reply to Weston Andros AdamsonThis specifically checks for the error being reported, otherwise a other exceptions like a programming error will be caught and treated as connection errors.
Signed-off-by Weston Andros Adamson <[email protected]>
--- nfs4.1/dataserver.py | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
nfs4.1/dataserver.py +2 -1
@@ -9,6 +9,7 @@ import nfs4client import hashlib import sys import nfs4_ops as op +import socket log = logging.getLogger("Dataserver Manager")
@@ -197,7 +198,7 @@ class DSDevice(object): ds = DataServer(server, port, path, mdsds=self.mdsds, multipath_servers=server_list) self.list.append(ds) - except: + except socket.error: log.critical("cannot access %s:%i/%s" % (server, port, '/'.join(path))) sys.exit(1)
Signature · 1 line
1.8.5.2 (Apple Git-48)
PATCH 4/17 4.1 server: avoid traceback in DS disconnect()
In reply to Weston Andros AdamsonSession objects have no destroy method.
Signed-off-by Weston Andros Adamson <[email protected]>
--- nfs4.1/dataserver.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
nfs4.1/dataserver.py +1 -1
@@ -58,7 +58,7 @@ class DataServer(object): self.make_root() def disconnect(self): - self.sess.destroy() + pass def execute(self, ops, exceptions=[], delay=5, maxretries=3): """ execute the NFS call
Signature · 1 line
1.8.5.2 (Apple Git-48)
PATCH 6/17 4.1 client: remove unused imports
In reply to Weston Andros AdamsonSigned-off-by Weston Andros Adamson <[email protected]>
--- nfs4.1/nfs4client.py | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-)
nfs4.1/nfs4client.py +1 -3
@@ -8,15 +8,13 @@ from xdrdef.sctrl_pack import SCTRLPacker, SCTRLUnpacker import nfs4_ops as op import time, struct import threading -import collections import hmac -from locking import Lock from nfs4commoncode import CBCompoundState as CompoundState, \ cb_encode_status as encode_status, \ cb_encode_status_by_name as encode_status_by_name -import sys, traceback +import traceback import logging logging.basicConfig(level=logging.INFO, format="%(levelname)-7s:%(name)s:%(message)s")
Signature · 1 line
1.8.5.2 (Apple Git-48)
PATCH 7/17 4.1 server: add -v flag & silence random output
In reply to Weston Andros AdamsonBy default the nfs4server spews so much information to stdout/stderr that it's almost useless. Add the -v flag (aka --verbose) to enable the old output.
Also remove some random prints and change some to log.info().
Signed-off-by Weston Andros Adamson <[email protected]>
--- nfs4.1/dataserver.py | 3 +-- nfs4.1/fs.py | 4 ---- nfs4.1/nfs4client.py | 11 ++-------- nfs4.1/nfs4server.py | 61 +++++++++++++++++++++++----------------------------- nfs4.1/nfs4state.py | 1 - 5 files changed, 30 insertions(+), 50 deletions(-)
nfs4.1/dataserver.py +1 -2
@@ -181,7 +181,7 @@ class DSDevice(object): line = line.strip() if not line or line.startswith('#'): continue - print "Analyzing: %r" % line + log.info("Analyzing: %r" % line) try: server_list, path = nfs4lib.parse_nfs_url(line) except:
@@ -191,7 +191,6 @@ class DSDevice(object): # for now, just use the last path for local connections server, port = server_list[-1] server_list = server_list[:-1] - print server, port, path try: log.info("Adding dataserver ip:%s port:%s path:%s" % (server, port, '/'.join(path)))
nfs4.1/fs.py +0 -4
@@ -13,7 +13,6 @@ from xdrdef.nfs4_pack import NFS4Packer log_o = logging.getLogger("fs.obj") log_fs = logging.getLogger("fs") logging.addLevelName(5, "FUNCT") -log_fs.setLevel(20) class MetaData(object): """Contains everything that needs to be stored
@@ -743,7 +742,6 @@ class ConfigObj(FSObject): line = line.strip() if line and not line.startswith("#"): lines.append(line) - print lines if len(lines) != 1: self._reset() return
@@ -1154,7 +1152,6 @@ class LayoutFSObj(FSObject): disk_offset, e.state)) block_layout = pnfs_block_layout4(elist) - print block_layout p = block.Packer() p.pack_pnfs_block_layout4(block_layout) ## if self.id <= 4:
@@ -1194,7 +1191,6 @@ class LayoutFSObj(FSObject): log_o.exception("Problem decoding opaque") raise NFS4Error(NFS4ERR_BADLAYOUT, tag="Error decoding opaque") upd_list = update.blu_commit_list - print upd_list # Error check for e in upd_list: if e.bex_state != block.PNFS_BLOCK_READWRITE_DATA:
nfs4.1/nfs4client.py +2 -9
@@ -19,9 +19,7 @@ import logging logging.basicConfig(level=logging.INFO, format="%(levelname)-7s:%(name)s:%(message)s") log_cb = logging.getLogger("nfs.client.cb") -log_cb.setLevel(logging.DEBUG) -SHOW_TRAFFIC = True # Debugging aid, prints out client traffic class NFS4Client(rpc.Client, rpc.Server): def __init__(self, host='localhost', port=2049, minorversion=1, ctrl_proc=16): rpc.Client.__init__(self, 100003, 4)
@@ -76,9 +74,7 @@ class NFS4Client(rpc.Client, rpc.Server): pipe = self.c1 p = packer(check_enum=checks, check_array=checks) c4 = COMPOUND4args(tag, version, ops) - if SHOW_TRAFFIC: - print - print c4 + log_cb.info("compound args = %r" % (c4,)) p.pack_COMPOUND4args(c4) return self.send_call(pipe, 1, p.get_buffer(), credinfo)
@@ -86,19 +82,16 @@ class NFS4Client(rpc.Client, rpc.Server): xid = self.compound_async(*args, **kwargs) pipe = kwargs.get("pipe", None) res = self.listen(xid, pipe=pipe) - if SHOW_TRAFFIC: - print res + log_cb.info("compound result = %r" % (res,)) return res def listen(self, xid, pipe=None, timeout=10.0): if pipe is None: pipe = self.c1 header, data = pipe.listen(xid, timeout) - #print "HEADER", header if data: p = nfs4lib.FancyNFS4Unpacker(data) data = p.unpack_COMPOUND4res() - #print "DATA", repr(data) return data def handle_0(self, data, cred):
nfs4.1/nfs4server.py +27 -34
150 lines not on this page · raw message
nfs4.1/nfs4state.py +0 -1
@@ -13,7 +13,6 @@ import rpc import logging log = logging.getLogger("nfs.server.state") -log.setLevel(10) POSIXLOCK = False
Signature · 1 line
1.8.5.2 (Apple Git-48)
PATCH 8/17 4.1 server: add -s option to print summary of ops
In reply to Weston Andros AdamsonAs a middle ground between verbose (-v) mode and silent mode, add "-s" (aka "--show_summary") to print one line per operation.
The summary mode will print a "role" header whenever the role changes.
Example output - mds starts, talks to v3 ds (.200), client (.11) mounts, touches an existing file:
call v3 172.16.200.200:2049
access
Mounting (2, 6) on '/files'
handle v4.1 ::ffff:172.16.200.11:758
exchange_id
create_session
sequence, reclaim_complete
sequence, putrootfh, secinfo_no_name
sequence, putrootfh, getfh, getattr
sequence, putfh, getattr
(repeated 6 times)
sequence, putfh, access, getattr
sequence, putfh, lookup, getfh, getattr
(repeated 1 times)
sequence, putfh, secinfo -> NFS4ERR_NOTSUPP
sequence, putfh, getattr
(repeated 4 times)
sequence, putfh, access, getattr
call v3 172.16.200.200:2049
create -> NFS3ERR_EXIST
lookup
getattr
(repeated 1 times)
handle v4.1 ::ffff:172.16.200.11:758
sequence, putfh, open, getfh, access, getattr
call v3 172.16.200.200:2049
getattr
(repeated 1 times)
handle v4.1 ::ffff:172.16.200.11:758
sequence, putfh, setattr, getattr
sequence, putfh, close, getattr
Signed-off-by Weston Andros Adamson <[email protected]>
--- nfs4.1/dataserver.py | 16 ++++++++++------ nfs4.1/nfs4client.py | 7 ++++++- nfs4.1/nfs4server.py | 47 ++++++++++++++++++++++++++++++++++++++++++++++- nfs4.1/server_exports.py | 7 ++++--- 4 files changed, 66 insertions(+), 11 deletions(-)
nfs4.1/dataserver.py +10 -6
@@ -13,8 +13,8 @@ import socket log = logging.getLogger("Dataserver Manager") -class DataServer(object): - def __init__(self, server, port, path, flavor=rpc.AUTH_SYS, active=True, mdsds=True, multipath_servers=None): +class DataServer41(object): + def __init__(self, server, port, path, flavor=rpc.AUTH_SYS, active=True, mdsds=True, multipath_servers=None, summary=None): self.mdsds = mdsds self.server = server self.port = int(port)
@@ -32,6 +32,8 @@ class DataServer(object): else: self.multipath_servers = [] + self.summary = summary + if active: self.up()
@@ -47,7 +49,8 @@ class DataServer(object): # only support root with AUTH_SYS for now s1 = rpc.security.instance(rpc.AUTH_SYS) self.cred1 = s1.init_cred(uid=0, gid=0) - self.c1 = nfs4client.NFS4Client(self.server, self.port) + self.c1 = nfs4client.NFS4Client(self.server, self.port, + summary=self.summary) self.c1.set_cred(self.cred1) self.c1.null() c = self.c1.new_client("DS.init_%s" % self.server)
@@ -172,7 +175,7 @@ class DSDevice(object): self.address_body = None # set by load() self.mdsds = mdsds # if you are both the DS and the MDS we are the only server - def load(self, filename): + def load(self, filename, server_obj): """ Read dataservers from configuration file: where each line has format e.g. server[:[port][/path]] """
@@ -194,8 +197,9 @@ class DSDevice(object): try: log.info("Adding dataserver ip:%s port:%s path:%s" % (server, port, '/'.join(path))) - ds = DataServer(server, port, path, mdsds=self.mdsds, - multipath_servers=server_list) + ds = DataServer41(server, port, path, mdsds=self.mdsds, + multipath_servers=server_list, + summary=server_obj.summary) self.list.append(ds) except socket.error: log.critical("cannot access %s:%i/%s" %
nfs4.1/nfs4client.py +6 -1
@@ -21,7 +21,7 @@ logging.basicConfig(level=logging.INFO, log_cb = logging.getLogger("nfs.client.cb") class NFS4Client(rpc.Client, rpc.Server): - def __init__(self, host='localhost', port=2049, minorversion=1, ctrl_proc=16): + def __init__(self, host='localhost', port=2049, minorversion=1, ctrl_proc=16, summary=None): rpc.Client.__init__(self, 100003, 4) self.prog = 0x40000000 self.versions = [1] # List of supported versions of prog
@@ -36,6 +36,7 @@ class NFS4Client(rpc.Client, rpc.Server): self.c1 = self.connect(self.server_address) self.sessions = {} # XXX Really, this should be per server self.ctrl_proc = ctrl_proc + self.summary = summary def set_cred(self, credinfo): self.default_cred = credinfo
@@ -83,6 +84,10 @@ class NFS4Client(rpc.Client, rpc.Server): pipe = kwargs.get("pipe", None) res = self.listen(xid, pipe=pipe) log_cb.info("compound result = %r" % (res,)) + if self.summary: + self.summary.show_op('call v4.1 %s:%s' % self.server_address, + [ nfs_opnum4[a.argop].lower()[3:] for a in args[0] ], + nfsstat4[res.status]) return res def listen(self, xid, pipe=None, timeout=10.0):
nfs4.1/nfs4server.py +46 -1
87 lines not on this page · raw message
nfs4.1/server_exports.py +4 -3
@@ -15,7 +15,7 @@ def mount_stuff(server, opts): E = BlockLayoutFS(5, backing_device=dev) server.mount(E, path="/block") if opts.use_files: - dservers = _load_dataservers(opts.dataservers, server.is_ds and server.is_mds) + dservers = _load_dataservers(opts.dataservers, server) if dservers is None: return F = FileLayoutFS(6, dservers)
@@ -33,7 +33,8 @@ def _create_simple_block_dev(): c1 = Concat([s3, s1]) return BlockVolume(c1) -def _load_dataservers(file, connect_to_ds=False): +def _load_dataservers(filename, server): + connect_to_ds = server.is_ds and server.is_mds dss = DSDevice(connect_to_ds) - dss.load(file) + dss.load(filename, server) return dss;
Signature · 1 line
1.8.5.2 (Apple Git-48)
PATCH 9/17 dataserver: make generic interface to ops
In reply to Weston Andros AdamsonHide the execute method to make DataServer objects more generic.
Signed-off-by Weston Andros Adamson <[email protected]>
--- nfs4.1/dataserver.py | 41 ++++++++++++++++++++++++++++++++++------- nfs4.1/fs.py | 24 ++++-------------------- 2 files changed, 38 insertions(+), 27 deletions(-)
nfs4.1/dataserver.py +34 -7
@@ -63,7 +63,7 @@ class DataServer41(object): def disconnect(self): pass - def execute(self, ops, exceptions=[], delay=5, maxretries=3): + def _execute(self, ops, exceptions=[], delay=5, maxretries=3): """ execute the NFS call If an error code is specified in the exceptions it means that the caller wants to handle the error himself
@@ -117,16 +117,16 @@ class DataServer41(object): kind = createtype4(NF4DIR) for comp in self.path: existing_path.append(comp) - res = self.execute(nfs4lib.use_obj(existing_path), + res = self._execute(nfs4lib.use_obj(existing_path), exceptions=[NFS4ERR_NOENT]) if res.status == NFS4ERR_NOENT: cr_ops = nfs4lib.use_obj(existing_path[:-1]) + \ [op.create(kind, comp, attrs)] - self.execute(cr_ops) - res = self.execute(nfs4lib.use_obj(self.path) + [op.getfh()]) + self._execute(cr_ops) + res = self._execute(nfs4lib.use_obj(self.path) + [op.getfh()]) self.path_fh = res.resarray[-1].object need = ACCESS4_READ | ACCESS4_LOOKUP | ACCESS4_MODIFY | ACCESS4_EXTEND - res = self.execute(nfs4lib.use_obj(self.path_fh) + [op.access(need)]) + res = self._execute(nfs4lib.use_obj(self.path_fh) + [op.access(need)]) if res.resarray[-1].access != need: raise RuntimeError # XXX clean DS directory
@@ -146,7 +146,7 @@ class DataServer41(object): open_op = op.open(seqid, access, deny, open_owner4(self.sess.client.clientid, owner), openflag, open_claim4(CLAIM_NULL, name)) - res = self.execute(nfs4lib.use_obj(self.path_fh) + [open_op, op.getfh()], exceptions=[NFS4ERR_EXIST]) + res = self._execute(nfs4lib.use_obj(self.path_fh) + [open_op, op.getfh()], exceptions=[NFS4ERR_EXIST]) if res.status == NFS4_OK: ds_fh = res.resarray[-1].opgetfh.resok4.object ds_openstateid = stateid4(0, res.resarray[-2].stateid.other)
@@ -162,10 +162,37 @@ class DataServer41(object): seqid=0 #FIXME: seqid must be !=0 fh, stateid = self.filehandles[mds_fh] ops = [op.putfh(fh)] + [op.close(seqid, stateid)] - res = self.execute(ops) + res = self._execute(ops) # ignoring return del self.filehandles[mds_fh] + def read(self, fh, pos, count): + ops = [op.putfh(fh), + op.read(nfs4lib.state00, pos, count)] + # There are all sorts of error handling issues here + res = self._execute(ops) + data = res.resarray[-1].data + return data + + def write(self, fh, pos, data): + ops = [op.putfh(fh), + op.write(nfs4lib.state00, pos, FILE_SYNC4, data)] + # There are all sorts of error handling issues here + res = self._execute(ops) + + def truncate(self, fh, size): + ops = [op.putfh(fh), + op.setattr(nfs4lib.state00, {FATTR4_SIZE: size})] + res = self._execute(ops) + + def get_size(self, fh): + ops = [op.putfh(fh), + op.getattr(1L << FATTR4_SIZE)] + res = self._execute(ops) + attrdict = res.resarray[-1].obj_attributes + return attrdict.get(FATTR4_SIZE, 0) + + class DSDevice(object): def __init__(self, mdsds): self.list = [] # list of DataServer instances
nfs4.1/fs.py +4 -20
@@ -1567,12 +1567,7 @@ class FilelayoutVolWrapper(object): self._pos = 0 def read(self, count): - # STUB stateid0 is illegal to a ds - ops = [op.putfh(self._fh), - op.read(nfs4lib.state00, self._pos, count)] - # There are all sorts of error handling issues here - res = self._ds.execute(ops) - data = res.resarray[-1].data + data = self._ds.read(self._fh, self._pos, count) self._pos += len(data) return data
@@ -1580,25 +1575,14 @@ class FilelayoutVolWrapper(object): self._pos = offset def write(self, data): - ops = [op.putfh(self._fh), - op.write(nfs4lib.state00, self._pos, FILE_SYNC4, data)] - # There are all sorts of error handling issues here - res = self._ds.execute(ops) + self._ds.write(self._fh, self._pos, data) self._pos += len(data) - return def truncate(self, size): - ops = [op.putfh(self._fh), - op.setattr(nfs4lib.state00, {FATTR4_SIZE: size})] - res = self._ds.execute(ops) - return + self._ds.truncate(self._fh, size) def get_size(self): - ops = [op.putfh(self._fh), - op.getattr(1L << FATTR4_SIZE)] - res = self._ds.execute(ops) - attrdict = res.resarray[-1].obj_attributes - return attrdict.get(FATTR4_SIZE, 0) + return self._ds.get_size(self._fh) ################################################
Signature · 1 line
1.8.5.2 (Apple Git-48)
PATCH 10/17 dataserver: don't import * from nfs4 specific mods
In reply to Weston Andros Adamsondon't pollute the global namespace with constants and types from nfs4_*
this is in preparation for NFSv3 DSes.
Signed-off-by Weston Andros Adamson <[email protected]>
--- nfs4.1/dataserver.py | 60 ++++++++++++++++++++++++++-------------------------- 1 file changed, 30 insertions(+), 30 deletions(-)
nfs4.1/dataserver.py +30 -30
@@ -1,8 +1,8 @@ import rpc import nfs4lib -from xdrdef.nfs4_type import * +import xdrdef.nfs4_type as type4 from xdrdef.nfs4_pack import NFS4Packer -from xdrdef.nfs4_const import * +import xdrdef.nfs4_const as const4 import time import logging import nfs4client
@@ -56,7 +56,7 @@ class DataServer41(object): c = self.c1.new_client("DS.init_%s" % self.server) # This is a hack to ensure MDS/DS communication path is at least # as wide as the client/MDS channel (at least for linux client) - fore_attrs = channel_attrs4(0, 16384, 16384, 2868, 8, 8, []) + fore_attrs = type4.channel_attrs4(0, 16384, 16384, 2868, 8, 8, []) self.sess = c.create_session(fore_attrs=fore_attrs) self.make_root()
@@ -68,12 +68,12 @@ class DataServer41(object): If an error code is specified in the exceptions it means that the caller wants to handle the error himself """ - retry_errors = [NFS4ERR_DELAY, NFS4ERR_GRACE] - state_errors = [NFS4ERR_STALE_CLIENTID, NFS4ERR_BADSESSION, - NFS4ERR_BADSLOT, NFS4ERR_DEADSESSION] + retry_errors = [const4.NFS4ERR_DELAY, const4.NFS4ERR_GRACE] + state_errors = [const4.NFS4ERR_STALE_CLIENTID, const4.NFS4ERR_BADSESSION, + const4.NFS4ERR_BADSLOT, const4.NFS4ERR_DEADSESSION] while True: res = self.sess.compound(ops) - if res.status == NFS4_OK or res.status in exceptions: + if res.status == const4.NFS4_OK or res.status in exceptions: return res elif res.status in retry_errors: if maxretries > 0:
@@ -95,7 +95,7 @@ class DataServer41(object): uaddr = '.'.join([self.server, str(self.port >> 8), str(self.port & 0xff)]) - return netaddr4(self.proto, uaddr) + return type4.netaddr4(self.proto, uaddr) def get_multipath_netaddr4s(self): netaddr4s = []
@@ -108,24 +108,24 @@ class DataServer41(object): if server.find(':') >= 0: proto = "tcp6" - netaddr4s.append(netaddr4(proto, uaddr)) + netaddr4s.append(type4.netaddr4(proto, uaddr)) return netaddr4s - def make_root(self, attrs={FATTR4_MODE:0777}): + def make_root(self, attrs={const4.FATTR4_MODE:0777}): existing_path = [] - kind = createtype4(NF4DIR) + kind = type4.createtype4(const4.NF4DIR) for comp in self.path: existing_path.append(comp) res = self._execute(nfs4lib.use_obj(existing_path), - exceptions=[NFS4ERR_NOENT]) - if res.status == NFS4ERR_NOENT: + exceptions=[const4.NFS4ERR_NOENT]) + if res.status == const4.NFS4ERR_NOENT: cr_ops = nfs4lib.use_obj(existing_path[:-1]) + \ [op.create(kind, comp, attrs)] self._execute(cr_ops) res = self._execute(nfs4lib.use_obj(self.path) + [op.getfh()]) self.path_fh = res.resarray[-1].object - need = ACCESS4_READ | ACCESS4_LOOKUP | ACCESS4_MODIFY | ACCESS4_EXTEND + need = const4.ACCESS4_READ | const4.ACCESS4_LOOKUP | const4.ACCESS4_MODIFY | const4.ACCESS4_EXTEND res = self._execute(nfs4lib.use_obj(self.path_fh) + [op.access(need)]) if res.resarray[-1].access != need: raise RuntimeError
@@ -135,25 +135,25 @@ class DataServer41(object): return hashlib.sha1("%r" % mds_fh).hexdigest() def open_file(self, mds_fh, seqid=0, - access=OPEN4_SHARE_ACCESS_BOTH, deny=OPEN4_SHARE_DENY_NONE, - attrs={FATTR4_MODE: 0777}, owner = "mds", mode=GUARDED4): + access=const4.OPEN4_SHARE_ACCESS_BOTH, deny=const4.OPEN4_SHARE_DENY_NONE, + attrs={const4.FATTR4_MODE: 0777}, owner = "mds", mode=const4.GUARDED4): verifier = self.sess.c.verifier - openflag = openflag4(OPEN4_CREATE, createhow4(mode, attrs, verifier)) + openflag = type4.openflag4(const4.OPEN4_CREATE, type4.createhow4(mode, attrs, verifier)) name = self.fh_to_name(mds_fh) while True: if mds_fh in self.filehandles: return open_op = op.open(seqid, access, deny, - open_owner4(self.sess.client.clientid, owner), - openflag, open_claim4(CLAIM_NULL, name)) - res = self._execute(nfs4lib.use_obj(self.path_fh) + [open_op, op.getfh()], exceptions=[NFS4ERR_EXIST]) - if res.status == NFS4_OK: + type4.open_owner4(self.sess.client.clientid, owner), + openflag, type4.open_claim4(const4.CLAIM_NULL, name)) + res = self._execute(nfs4lib.use_obj(self.path_fh) + [open_op, op.getfh()], exceptions=[const4.NFS4ERR_EXIST]) + if res.status == const4.NFS4_OK: ds_fh = res.resarray[-1].opgetfh.resok4.object - ds_openstateid = stateid4(0, res.resarray[-2].stateid.other) + ds_openstateid = type4.stateid4(0, res.resarray[-2].stateid.other) self.filehandles[mds_fh] = (ds_fh, ds_openstateid) return - elif res.status == NFS4ERR_EXIST: - openflag = openflag4(OPEN4_NOCREATE) + elif res.status == const4.NFS4ERR_EXIST: + openflag = type4.openflag4(const4.OPEN4_NOCREATE) else: raise RuntimeError
@@ -176,26 +176,26 @@ class DataServer41(object): def write(self, fh, pos, data): ops = [op.putfh(fh), - op.write(nfs4lib.state00, pos, FILE_SYNC4, data)] + op.write(nfs4lib.state00, pos, const4.FILE_SYNC4, data)] # There are all sorts of error handling issues here res = self._execute(ops) def truncate(self, fh, size): ops = [op.putfh(fh), - op.setattr(nfs4lib.state00, {FATTR4_SIZE: size})] + op.setattr(nfs4lib.state00, {const4.FATTR4_SIZE: size})] res = self._execute(ops) def get_size(self, fh): ops = [op.putfh(fh), - op.getattr(1L << FATTR4_SIZE)] + op.getattr(1L << const4.FATTR4_SIZE)] res = self._execute(ops) attrdict = res.resarray[-1].obj_attributes - return attrdict.get(FATTR4_SIZE, 0) + return attrdict.get(const4.FATTR4_SIZE, 0) class DSDevice(object): def __init__(self, mdsds): - self.list = [] # list of DataServer instances + self.list = [] # list of DataServer41 instances # STUB only one data group supported for now self.devid = 0 self.active = 0
@@ -247,7 +247,7 @@ class DSDevice(object): netaddrs.append(multipath) stripe_indices.append(index) index = index + 1 - addr = nfsv4_1_file_layout_ds_addr4(stripe_indices, netaddrs) + addr = type4.nfsv4_1_file_layout_ds_addr4(stripe_indices, netaddrs) p = NFS4Packer() p.pack_nfsv4_1_file_layout_ds_addr4(addr) return p.get_buffer()
Signature · 1 line
1.8.5.2 (Apple Git-48)
PATCH 11/17 4.1 server: move nfs4_ops.py to nfs_ops.py
In reply to Weston Andros AdamsonAlso replace ugly exec & inspect code to just define a class with a __getattr__ switch.
Signed-off-by Weston Andros Adamson <[email protected]>
--- nfs4.1/dataserver.py | 32 ++++++++++--------- nfs4.1/fs.py | 2 -- nfs4.1/nfs4_ops.py | 61 ----------------------------------- nfs4.1/nfs4client.py | 16 +++++----- nfs4.1/nfs4lib.py | 8 +++-- nfs4.1/nfs4state.py | 8 +++-- nfs4.1/nfs_ops.py | 89 ++++++++++++++++++++++++++++++++++++++++++++++++++++ 7 files changed, 125 insertions(+), 91 deletions(-) delete mode 100644 nfs4.1/nfs4_ops.py create mode 100644 nfs4.1/nfs_ops.py
nfs4.1/dataserver.py +17 -15
@@ -8,11 +8,13 @@ import logging import nfs4client import hashlib import sys -import nfs4_ops as op +import nfs_ops import socket log = logging.getLogger("Dataserver Manager") +op4 = nfs_ops.NFS4ops() + class DataServer41(object): def __init__(self, server, port, path, flavor=rpc.AUTH_SYS, active=True, mdsds=True, multipath_servers=None, summary=None): self.mdsds = mdsds
@@ -121,12 +123,12 @@ class DataServer41(object): exceptions=[const4.NFS4ERR_NOENT]) if res.status == const4.NFS4ERR_NOENT: cr_ops = nfs4lib.use_obj(existing_path[:-1]) + \ - [op.create(kind, comp, attrs)] + [op4.create(kind, comp, attrs)] self._execute(cr_ops) - res = self._execute(nfs4lib.use_obj(self.path) + [op.getfh()]) + res = self._execute(nfs4lib.use_obj(self.path) + [op4.getfh()]) self.path_fh = res.resarray[-1].object need = const4.ACCESS4_READ | const4.ACCESS4_LOOKUP | const4.ACCESS4_MODIFY | const4.ACCESS4_EXTEND - res = self._execute(nfs4lib.use_obj(self.path_fh) + [op.access(need)]) + res = self._execute(nfs4lib.use_obj(self.path_fh) + [op4.access(need)]) if res.resarray[-1].access != need: raise RuntimeError # XXX clean DS directory
@@ -143,10 +145,10 @@ class DataServer41(object): while True: if mds_fh in self.filehandles: return - open_op = op.open(seqid, access, deny, + open_op = op4.open(seqid, access, deny, type4.open_owner4(self.sess.client.clientid, owner), openflag, type4.open_claim4(const4.CLAIM_NULL, name)) - res = self._execute(nfs4lib.use_obj(self.path_fh) + [open_op, op.getfh()], exceptions=[const4.NFS4ERR_EXIST]) + res = self._execute(nfs4lib.use_obj(self.path_fh) + [open_op, op4.getfh()], exceptions=[const4.NFS4ERR_EXIST]) if res.status == const4.NFS4_OK: ds_fh = res.resarray[-1].opgetfh.resok4.object ds_openstateid = type4.stateid4(0, res.resarray[-2].stateid.other)
@@ -161,33 +163,33 @@ class DataServer41(object): """close the given file""" seqid=0 #FIXME: seqid must be !=0 fh, stateid = self.filehandles[mds_fh] - ops = [op.putfh(fh)] + [op.close(seqid, stateid)] + ops = [op4.putfh(fh)] + [op4.close(seqid, stateid)] res = self._execute(ops) # ignoring return del self.filehandles[mds_fh] def read(self, fh, pos, count): - ops = [op.putfh(fh), - op.read(nfs4lib.state00, pos, count)] + ops = [op4.putfh(fh), + op4.read(nfs4lib.state00, pos, count)] # There are all sorts of error handling issues here res = self._execute(ops) data = res.resarray[-1].data return data def write(self, fh, pos, data): - ops = [op.putfh(fh), - op.write(nfs4lib.state00, pos, const4.FILE_SYNC4, data)] + ops = [op4.putfh(fh), + op4.write(nfs4lib.state00, pos, const4.FILE_SYNC4, data)] # There are all sorts of error handling issues here res = self._execute(ops) def truncate(self, fh, size): - ops = [op.putfh(fh), - op.setattr(nfs4lib.state00, {const4.FATTR4_SIZE: size})] + ops = [op4.putfh(fh), + op4.setattr(nfs4lib.state00, {const4.FATTR4_SIZE: size})] res = self._execute(ops) def get_size(self, fh): - ops = [op.putfh(fh), - op.getattr(1L << const4.FATTR4_SIZE)] + ops = [op4.putfh(fh), + op4.getattr(1L << const4.FATTR4_SIZE)] res = self._execute(ops) attrdict = res.resarray[-1].obj_attributes return attrdict.get(const4.FATTR4_SIZE, 0)
nfs4.1/fs.py +0 -2
@@ -1557,8 +1557,6 @@ class FileLayoutFile(object): # XXX This should inherit from fs_base.py vol = FilelayoutVolWrapper(self._obj, device.list[index]) return vol, v_pos, remaining -import nfs4_ops as op - class FilelayoutVolWrapper(object): def __init__(self, obj, dataserver): self._obj = obj
nfs4.1/nfs4_ops.py deleted +0 -61
61 lines not on this page · raw message
nfs4.1/nfs4client.py +9 -7
@@ -5,7 +5,7 @@ from nfs4lib import NFS4Error, NFS4Replay, inc_u32 from xdrdef.nfs4_type import * from xdrdef.nfs4_const import * from xdrdef.sctrl_pack import SCTRLPacker, SCTRLUnpacker -import nfs4_ops as op +import nfs_ops import time, struct import threading import hmac
@@ -20,6 +20,8 @@ logging.basicConfig(level=logging.INFO, format="%(levelname)-7s:%(name)s:%(message)s") log_cb = logging.getLogger("nfs.client.cb") +op4 = nfs_ops.NFS4ops() + class NFS4Client(rpc.Client, rpc.Server): def __init__(self, host='localhost', port=2049, minorversion=1, ctrl_proc=16, summary=None): rpc.Client.__init__(self, 100003, 4)
@@ -275,7 +277,7 @@ class NFS4Client(rpc.Client, rpc.Server): owner = client_owner4(verf, name) if protect is None: protect = state_protect4_a(SP4_NONE) - res = self.compound([op.exchange_id(owner, flags, protect, + res = self.compound([op4.exchange_id(owner, flags, protect, [self.impl_id])], cred) nfs4lib.check(res, expect)
@@ -287,7 +289,7 @@ class NFS4Client(rpc.Client, rpc.Server): def new_client_session(self, name, flags=0, sec=None): c = self.new_client(name, flags=flags) s = c.create_session(sec=sec) - s.compound([op.reclaim_complete(FALSE)]) + s.compound([op4.reclaim_complete(FALSE)]) return s class ClientStateProtection(object):
@@ -339,7 +341,7 @@ class ClientRecord(object): if prog is None: prog = self.c.prog for item in xrange(max_retries): - res = self.c.compound([op.create_session(self.clientid, self.seqid, + res = self.c.compound([op4.create_session(self.clientid, self.seqid, flags, fore_attrs, back_attrs, prog, sec)],
@@ -362,7 +364,7 @@ class ClientRecord(object): self.seqid = inc_u32(csr.csr_sequence) # XXX Do we need to check this? sess = SessionRecord(csr, self) self.c.sessions[sess.sessionid] = sess - sess.compound([op.reclaim_complete(FALSE)]) + sess.compound([op4.reclaim_complete(FALSE)]) return sess def _cb_hook(self, prefix, opname, funct):
@@ -430,7 +432,7 @@ class SessionRecord(object): raise RuntimeError slot = self.fore_channel.slots[slot] # STUB, need to properly set highest - return op.sequence(self.sessionid, slot.get_seqid(seq_delta), + return op4.sequence(self.sessionid, slot.get_seqid(seq_delta), slot.id, slot.id, cache_this) def set_ssv(self, ssv=None, *args, **kwargs):
@@ -443,7 +445,7 @@ class SessionRecord(object): p = nfs4lib.FancyNFS4Packer() p.pack_SEQUENCE4args(seq_op.opsequence) digest = protect.context.hmac(p.get_buffer(), SSV4_SUBKEY_MIC_I2T) - ssv_op = op.set_ssv(ssv, digest) + ssv_op = op4.set_ssv(ssv, digest) res = self.c.compound([seq_op, ssv_op], *args, **kwargs) # STUB - do some checking protect.context.set_ssv(ssv)
nfs4.1/nfs4lib.py +5 -3
@@ -3,7 +3,7 @@ import rpc import xdrdef.nfs4_const from xdrdef.nfs4_pack import NFS4Packer, NFS4Unpacker import xdrdef.nfs4_type -import nfs4_ops as op +import nfs_ops import time import collections import hmac
@@ -30,6 +30,8 @@ state01 = xdrdef.nfs4_type.stateid4(1, "\0" * 12) import hashlib # Note this requires 2.5 or higher +op4 = nfs_ops.NFS4ops() + # Note that all the oid strings have tag and length bytes prepended, as # per description of sec_oid4 in draft26 sect 3.2
@@ -626,9 +628,9 @@ def use_obj(file): if file is None or file == [None]: return [] elif type(file) is str: - return [op.putfh(file)] + return [op4.putfh(file)] else: - return [op.putrootfh()] + [op.lookup(comp) for comp in file] + return [op4.putrootfh()] + [op4.lookup(comp) for comp in file] ############################################### # Attribute information
nfs4.1/nfs4state.py +5 -3
@@ -8,12 +8,14 @@ from nfs4lib import NFS4Error #from xdrdef.nfs4_type import stateid4 from xdrdef.nfs4_type import * from xdrdef.nfs4_const import * -import nfs4_ops as op +import nfs_ops import rpc import logging log = logging.getLogger("nfs.server.state") +op4 = nfs_ops.NFS4ops() + POSIXLOCK = False SHARE, BYTE, DELEG, LAYOUT, ANON = range(5) # State types
@@ -748,9 +750,9 @@ class DelegEntry(StateTableEntry): # ANSWER - we care about self.status, which can be set to # INVALID anytime by deleg_return slot = session.channel_back.choose_slot() - seq_op = op.cb_sequence(session.sessionid, slot.get_seqid(), + seq_op = op4.cb_sequence(session.sessionid, slot.get_seqid(), slot.id, slot.id, True, []) # STUB - recall_op = op.cb_recall(self.get_id(cb=True), False, self.file.fh) + recall_op = op4.cb_recall(self.get_id(cb=True), False, self.file.fh) if self.invalid: # Race here doesn't matter, but would like to avoid the # RPC if possible.
nfs4.1/nfs_ops.py new file +89 -0
89 lines not on this page · raw message
Signature · 1 line
1.8.5.2 (Apple Git-48)
PATCH 12/17 add mntv3, portmapv2 and nfsv3 .x files
In reply to Weston Andros Adamsonmntv3 and nfsv3 .x files made from IETF specs by Tom Haynes <[email protected]>.
portmap2 .x files made from IETF specs by Weston Andros Adamson <[email protected]>.
Signed-off-by Weston Andros Adamson <[email protected]>
--- nfs4.1/xdrdef/mnt3.x | 68 ++++ nfs4.1/xdrdef/nfs3.x | 818 ++++++++++++++++++++++++++++++++++++++++++++++++ nfs4.1/xdrdef/portmap.x | 51 +++ 3 files changed, 937 insertions(+) create mode 100644 nfs4.1/xdrdef/mnt3.x create mode 100644 nfs4.1/xdrdef/nfs3.x create mode 100644 nfs4.1/xdrdef/portmap.x
nfs4.1/xdrdef/mnt3.x new file +68 -0
@@ -0,0 +1,68 @@ +const MNTPATHLEN = 1024; /* Maximum bytes in a path name */ +const MNTNAMLEN = 255; /* Maximum bytes in a name */ +const FHSIZE3 = 64; /* Maximum bytes in a V3 file handle */ + +typedef opaque fhandle3 <FHSIZE3>; +typedef string dirpath <MNTPATHLEN>; +typedef string name <MNTNAMLEN>; + +struct mountres3_ok { + fhandle3 fhandle; + int auth_flavors <>; +}; + +union mountres3 switch (mountstat3 fhs_status) { +case MNT3_OK: + mountres3_ok mountinfo; +default: + void; +}; + +struct mountbody { + name ml_hostname; + dirpath ml_directory; + mountlist ml_next; +}; + +typedef mountbody *mountlist; + + +struct groupnode { + name gr_name; + groups gr_next; +}; + +typedef groupnode *groups; + + +struct exportnode { + dirpath ex_dir; + groups ex_groups; + exports ex_next; +}; + +typedef exportnode *exports; + +enum mountstat3 { + MNT3_OK = 0, /* no error */ + MNT3ERR_PERM = 1, /* Not owner */ + MNT3ERR_NOENT = 2, /* No such file or directory */ + MNT3ERR_IO = 5, /* I/O error */ + MNT3ERR_ACCES = 13, /* Permission denied */ + MNT3ERR_NOTDIR = 20, /* Not a directory */ + MNT3ERR_INVAL = 22, /* Invalid argument */ + MNT3ERR_NAMETOOLONG = 63, /* Filename too long */ + MNT3ERR_NOTSUPP = 10004, /* Operation not supported */ + MNT3ERR_SERVERFAULT = 10006 /* A failure on the server */ +}; + +program MOUNT_PROGRAM { + version MOUNT_V3 { + void MOUNTPROC3_NULL(void) = 0; + mountres3 MOUNTPROC3_MNT(dirpath) = 1; + mountlist MOUNTPROC3_DUMP(void) = 2; + void MOUNTPROC3_UMNT(dirpath) = 3; + void MOUNTPROC3_UMNTALL(void) = 4; + exports MOUNTPROC3_EXPORT(void) = 5; + } = 3; +} = 100005;
nfs4.1/xdrdef/nfs3.x new file +818 -0
818 lines not on this page · raw message
nfs4.1/xdrdef/portmap.x new file +51 -0
51 lines not on this page · raw message
Signature · 1 line
1.8.5.2 (Apple Git-48)
Re: [PATCH pynfs 12/17] add mntv3, portmapv2 and nfsv3 .x files
Quoted 2 lines · Weston Andros Adamson
mntv3 and nfsv3 .x files made from IETF specs by Tom Haynes <[email protected]>.
Some folks actually submitted v3 client code at some point and I never managed to get it merged. Maybe I should dig that up again while we're doing this.... But probably it's still out of scope.
--b.
Quoted 972 lines
portmap2 .x files made from IETF specs by Weston Andros Adamson <[email protected]>.
Signed-off-by Weston Andros Adamson <[email protected]>--- nfs4.1/xdrdef/mnt3.x | 68 ++++ nfs4.1/xdrdef/nfs3.x | 818 ++++++++++++++++++++++++++++++++++++++++++++++++ nfs4.1/xdrdef/portmap.x | 51 +++ 3 files changed, 937 insertions(+) create mode 100644 nfs4.1/xdrdef/mnt3.x create mode 100644 nfs4.1/xdrdef/nfs3.x create mode 100644 nfs4.1/xdrdef/portmap.xnfs4.1/xdrdef/mnt3.x new file +68
-0@@ -0,0 +1,68 @@ +const MNTPATHLEN = 1024; /* Maximum bytes in a path name */ +const MNTNAMLEN = 255; /* Maximum bytes in a name */ +const FHSIZE3 = 64; /* Maximum bytes in a V3 file handle */ + +typedef opaque fhandle3 <FHSIZE3>; +typedef string dirpath <MNTPATHLEN>; +typedef string name <MNTNAMLEN>; + +struct mountres3_ok { + fhandle3 fhandle; + int auth_flavors <>; +}; + +union mountres3 switch (mountstat3 fhs_status) { +case MNT3_OK: + mountres3_ok mountinfo; +default: + void; +}; + +struct mountbody { + name ml_hostname; + dirpath ml_directory; + mountlist ml_next; +}; + +typedef mountbody *mountlist; + + +struct groupnode { + name gr_name; + groups gr_next; +}; + +typedef groupnode *groups; + + +struct exportnode { + dirpath ex_dir; + groups ex_groups; + exports ex_next; +}; + +typedef exportnode *exports; + +enum mountstat3 { + MNT3_OK = 0, /* no error */ + MNT3ERR_PERM = 1, /* Not owner */ + MNT3ERR_NOENT = 2, /* No such file or directory */ + MNT3ERR_IO = 5, /* I/O error */ + MNT3ERR_ACCES = 13, /* Permission denied */ + MNT3ERR_NOTDIR = 20, /* Not a directory */ + MNT3ERR_INVAL = 22, /* Invalid argument */ + MNT3ERR_NAMETOOLONG = 63, /* Filename too long */ + MNT3ERR_NOTSUPP = 10004, /* Operation not supported */ + MNT3ERR_SERVERFAULT = 10006 /* A failure on the server */ +}; + +program MOUNT_PROGRAM { + version MOUNT_V3 { + void MOUNTPROC3_NULL(void) = 0; + mountres3 MOUNTPROC3_MNT(dirpath) = 1; + mountlist MOUNTPROC3_DUMP(void) = 2; + void MOUNTPROC3_UMNT(dirpath) = 3; + void MOUNTPROC3_UMNTALL(void) = 4; + exports MOUNTPROC3_EXPORT(void) = 5; + } = 3; +} = 100005;nfs4.1/xdrdef/nfs3.x new file +818
-0@@ -0,0 +1,818 @@ +const NFS3_FHSIZE = 64; /* Maximum bytes in a V3 file handle */ + +const FALSE = 0; +const TRUE = 1; + +const NFS3_COOKIEVERFSIZE = 8; +const NFS3_CREATEVERFSIZE = 8; +const NFS3_WRITEVERFSIZE = 8; + +typedef unsigned hyper uint64; +typedef hyper int64; +typedef unsigned int uint32; +typedef int int32; +typedef string filename3 <>; +typedef string nfspath3 <>; +typedef uint64 fileid3; +typedef uint64 cookie3; +typedef opaque cookieverf3[NFS3_COOKIEVERFSIZE]; +typedef opaque createverf3[NFS3_CREATEVERFSIZE]; +typedef opaque writeverf3[NFS3_WRITEVERFSIZE]; +typedef uint32 uid3; +typedef uint32 gid3; +typedef uint64 size3; +typedef uint64 offset3; +typedef uint32 mode3; +typedef uint32 count3; + +enum nfsstat3 { + NFS3_OK = 0, + NFS3ERR_PERM = 1, + NFS3ERR_NOENT = 2, + NFS3ERR_IO = 5, + NFS3ERR_NXIO = 6, + NFS3ERR_ACCES = 13, + NFS3ERR_EXIST = 17, + NFS3ERR_XDEV = 18, + NFS3ERR_NODEV = 19, + NFS3ERR_NOTDIR = 20, + NFS3ERR_ISDIR = 21, + NFS3ERR_INVAL = 22, + NFS3ERR_FBIG = 27, + NFS3ERR_NOSPC = 28, + NFS3ERR_ROFS = 30, + NFS3ERR_MLINK = 31, + NFS3ERR_NAMETOOLONG = 63, + NFS3ERR_NOTEMPTY = 66, + NFS3ERR_DQUOT = 69, + NFS3ERR_STALE = 70, + NFS3ERR_REMOTE = 71, + NFS3ERR_BADHANDLE = 10001, + NFS3ERR_NOT_SYNC = 10002, + NFS3ERR_BAD_COOKIE = 10003, + NFS3ERR_NOTSUPP = 10004, + NFS3ERR_TOOSMALL = 10005, + NFS3ERR_SERVERFAULT = 10006, + NFS3ERR_BADTYPE = 10007, + NFS3ERR_JUKEBOX = 10008 +}; + +enum ftype3 { + NF3REG = 1, + NF3DIR = 2, + NF3BLK = 3, + NF3CHR = 4, + NF3LNK = 5, + NF3SOCK = 6, + NF3FIFO = 7 +}; + +struct specdata3 { + uint32 specdata1; + uint32 specdata2; +}; + +struct nfs_fh3 { + opaque data <NFS3_FHSIZE>; +}; + +struct nfstime3 { + uint32 seconds; + uint32 nseconds; +}; + +struct fattr3 { + ftype3 type; + mode3 mode; + uint32 nlink; + uid3 uid; + gid3 gid; + size3 size; + size3 used; + specdata3 rdev; + uint64 fsid; + fileid3 fileid; + nfstime3 atime; + nfstime3 mtime; + nfstime3 ctime; +}; + +union post_op_attr switch (bool attributes_follow) { +case TRUE: + fattr3 attributes; +case FALSE: + void; +}; + +struct wcc_attr { + size3 size; + nfstime3 mtime; + nfstime3 ctime; +}; + +union pre_op_attr switch (bool attributes_follow) { +case TRUE: + wcc_attr attributes; +case FALSE: + void; +}; + +struct wcc_data { + pre_op_attr before; + post_op_attr after; +}; + +union post_op_fh3 switch (bool handle_follows) { +case TRUE: + nfs_fh3 handle; +case FALSE: + void; +}; + +enum time_how { + DONT_CHANGE = 0, + SET_TO_SERVER_TIME = 1, + SET_TO_CLIENT_TIME = 2 +}; + +union set_mode3 switch (bool set_it) { +case TRUE: + mode3 mode; +default: + void; +}; + +union set_uid3 switch (bool set_it) { +case TRUE: + uid3 uid; +default: + void; +}; + +union set_gid3 switch (bool set_it) { +case TRUE: + gid3 gid; +default: + void; +}; + +union set_size3 switch (bool set_it) { +case TRUE: + size3 size; +default: + void; +}; + +union set_atime switch (time_how set_it) { +case SET_TO_CLIENT_TIME: + nfstime3 atime; +default: + void; +}; + +union set_mtime switch (time_how set_it) { +case SET_TO_CLIENT_TIME: + nfstime3 mtime; +default: + void; +}; + +struct sattr3 { + set_mode3 mode; + set_uid3 uid; + set_gid3 gid; + set_size3 size; + set_atime atime; + set_mtime mtime; +}; + +struct diropargs3 { + nfs_fh3 dir; + filename3 name; +}; + +struct GETATTR3args { + nfs_fh3 object; +}; + +struct GETATTR3resok { + fattr3 obj_attributes; +}; + +union GETATTR3res switch (nfsstat3 status) { +case NFS3_OK: + GETATTR3resok resok; +default: + void; +}; + +union sattrguard3 switch (bool check) { +case TRUE: + nfstime3 obj_ctime; +case FALSE: + void; +}; + +struct SETATTR3args { + nfs_fh3 object; + sattr3 new_attributes; + sattrguard3 guard; +}; + +struct SETATTR3resok { + wcc_data obj_wcc; +}; + +struct SETATTR3resfail { + wcc_data obj_wcc; +}; + +union SETATTR3res switch (nfsstat3 status) { +case NFS3_OK: + SETATTR3resok resok; +default: + SETATTR3resfail resfail; +}; + +struct LOOKUP3args { + diropargs3 what; +}; + +struct LOOKUP3resok { + nfs_fh3 object; + post_op_attr obj_attributes; + post_op_attr dir_attributes; +}; + +struct LOOKUP3resfail { + post_op_attr dir_attributes; +}; + +union LOOKUP3res switch (nfsstat3 status) { +case NFS3_OK: + LOOKUP3resok resok; +default: + LOOKUP3resfail resfail; +}; + +const ACCESS3_READ = 0x0001; +const ACCESS3_LOOKUP = 0x0002; +const ACCESS3_MODIFY = 0x0004; +const ACCESS3_EXTEND = 0x0008; +const ACCESS3_DELETE = 0x0010; +const ACCESS3_EXECUTE = 0x0020; + +struct ACCESS3args { + nfs_fh3 object; + uint32 access; +}; + +struct ACCESS3resok { + post_op_attr obj_attributes; + uint32 access; +}; + +struct ACCESS3resfail { + post_op_attr obj_attributes; +}; + +union ACCESS3res switch (nfsstat3 status) { +case NFS3_OK: + ACCESS3resok resok; +default: + ACCESS3resfail resfail; +}; + +struct READLINK3args { + nfs_fh3 symlink; +}; + +struct READLINK3resok { + post_op_attr symlink_attributes; + nfspath3 data; +}; + +struct READLINK3resfail { + post_op_attr symlink_attributes; +}; + +union READLINK3res switch (nfsstat3 status) { +case NFS3_OK: + READLINK3resok resok; +default: + READLINK3resfail resfail; +}; + +struct READ3args { + nfs_fh3 file; + offset3 offset; + count3 count; +}; + +struct READ3resok { + post_op_attr file_attributes; + count3 count; + bool eof; + opaque data <>; +}; + +struct READ3resfail { + post_op_attr file_attributes; +}; + +union READ3res switch (nfsstat3 status) { +case NFS3_OK: + READ3resok resok; +default: + READ3resfail resfail; +}; + +enum stable_how { + UNSTABLE = 0, + DATA_SYNC = 1, + FILE_SYNC = 2 +}; + +struct WRITE3args { + nfs_fh3 file; + offset3 offset; + count3 count; + stable_how stable; + opaque data <>; +}; + +struct WRITE3resok { + wcc_data file_wcc; + count3 count; + stable_how committed; + writeverf3 verf; +}; + +struct WRITE3resfail { + wcc_data file_wcc; +}; + +union WRITE3res switch (nfsstat3 status) { +case NFS3_OK: + WRITE3resok resok; +default: + WRITE3resfail resfail; +}; + +enum createmode3 { + UNCHECKED = 0, + GUARDED = 1, + EXCLUSIVE = 2 +}; + +union createhow3 switch (createmode3 mode) { +case UNCHECKED: +case GUARDED: + sattr3 obj_attributes; +case EXCLUSIVE: + createverf3 verf; +}; + +struct CREATE3args { + diropargs3 where; + createhow3 how; +}; + +struct CREATE3resok { + post_op_fh3 obj; + post_op_attr obj_attributes; + wcc_data dir_wcc; +}; + +struct CREATE3resfail { + wcc_data dir_wcc; +}; + +union CREATE3res switch (nfsstat3 status) { +case NFS3_OK: + CREATE3resok resok; +default: + CREATE3resfail resfail; +}; + +struct MKDIR3args { + diropargs3 where; + sattr3 attributes; +}; + +struct MKDIR3resok { + post_op_fh3 obj; + post_op_attr obj_attributes; + wcc_data dir_wcc; +}; + +struct MKDIR3resfail { + wcc_data dir_wcc; +}; + +union MKDIR3res switch (nfsstat3 status) { +case NFS3_OK: + MKDIR3resok resok; +default: + MKDIR3resfail resfail; +}; + +struct symlinkdata3 { + sattr3 symlink_attributes; + nfspath3 symlink_data; +}; + +struct SYMLINK3args { + diropargs3 where; + symlinkdata3 symlink; +}; + +struct SYMLINK3resok { + post_op_fh3 obj; + post_op_attr obj_attributes; + wcc_data dir_wcc; +}; + +struct SYMLINK3resfail { + wcc_data dir_wcc; +}; + +union SYMLINK3res switch (nfsstat3 status) { +case NFS3_OK: + SYMLINK3resok resok; +default: + SYMLINK3resfail resfail; +}; + +struct devicedata3 { + sattr3 dev_attributes; + specdata3 spec; +}; + +union mknoddata3 switch (ftype3 type) { +case NF3CHR: +case NF3BLK: + devicedata3 device; +case NF3SOCK: +case NF3FIFO: + sattr3 pipe_attributes; +default: + void; +}; + +struct MKNOD3args { + diropargs3 where; + mknoddata3 what; +}; + +struct MKNOD3resok { + post_op_fh3 obj; + post_op_attr obj_attributes; + wcc_data dir_wcc; +}; + +struct MKNOD3resfail { + wcc_data dir_wcc; +}; + +union MKNOD3res switch (nfsstat3 status) { +case NFS3_OK: + MKNOD3resok resok; +default: + MKNOD3resfail resfail; +}; + +struct REMOVE3args { + diropargs3 object; +}; + +struct REMOVE3resok { + wcc_data dir_wcc; +}; + +struct REMOVE3resfail { + wcc_data dir_wcc; +}; + +union REMOVE3res switch (nfsstat3 status) { +case NFS3_OK: + REMOVE3resok resok; +default: + REMOVE3resfail resfail; +}; + +struct RMDIR3args { + diropargs3 object; +}; + +struct RMDIR3resok { + wcc_data dir_wcc; +}; + +struct RMDIR3resfail { + wcc_data dir_wcc; +}; + +union RMDIR3res switch (nfsstat3 status) { +case NFS3_OK: + RMDIR3resok resok; +default: + RMDIR3resfail resfail; +}; + +struct RENAME3args { + diropargs3 fromfile; + diropargs3 tofile; +}; + +struct RENAME3resok { + wcc_data fromdir_wcc; + wcc_data todir_wcc; +}; + +struct RENAME3resfail { + wcc_data fromdir_wcc; + wcc_data todir_wcc; +}; + +union RENAME3res switch (nfsstat3 status) { +case NFS3_OK: + RENAME3resok resok; +default: + RENAME3resfail resfail; +}; + +struct LINK3args { + nfs_fh3 file; + diropargs3 link; +}; + +struct LINK3resok { + post_op_attr file_attributes; + wcc_data linkdir_wcc; +}; + +struct LINK3resfail { + post_op_attr file_attributes; + wcc_data linkdir_wcc; +}; + +union LINK3res switch (nfsstat3 status) { +case NFS3_OK: + LINK3resok resok; +default: + LINK3resfail resfail; +}; + +struct READDIR3args { + nfs_fh3 dir; + cookie3 cookie; + cookieverf3 cookieverf; + count3 count; +}; + +struct entry3 { + fileid3 fileid; + filename3 name; + cookie3 cookie; + entry3 *nextentry; +}; + +struct dirlist3 { + entry3 *entries; + bool eof; +}; + +struct READDIR3resok { + post_op_attr dir_attributes; + cookieverf3 cookieverf; + dirlist3 reply; +}; + +struct READDIR3resfail { + post_op_attr dir_attributes; +}; + +union READDIR3res switch (nfsstat3 status) { +case NFS3_OK: + READDIR3resok resok; +default: + READDIR3resfail resfail; +}; + +struct READDIRPLUS3args { + nfs_fh3 dir; + cookie3 cookie; + cookieverf3 cookieverf; + count3 dircount; + count3 maxcount; +}; + +struct entryplus3 { + fileid3 fileid; + filename3 name; + cookie3 cookie; + post_op_attr name_attributes; + post_op_fh3 name_handle; + entryplus3 *nextentry; +}; + +struct dirlistplus3 { + entryplus3 *entries; + bool eof; +}; + +struct READDIRPLUS3resok { + post_op_attr dir_attributes; + cookieverf3 cookieverf; + dirlistplus3 reply; +}; + +struct READDIRPLUS3resfail { + post_op_attr dir_attributes; +}; + +union READDIRPLUS3res switch (nfsstat3 status) { +case NFS3_OK: + READDIRPLUS3resok resok; +default: + READDIRPLUS3resfail resfail; +}; + +struct FSSTAT3args { + nfs_fh3 fsroot; +}; + +struct FSSTAT3resok { + post_op_attr obj_attributes; + size3 tbytes; + size3 fbytes; + size3 abytes; + size3 tfiles; + size3 ffiles; + size3 afiles; + uint32 invarsec; +}; + +struct FSSTAT3resfail { + post_op_attr obj_attributes; +}; + +union FSSTAT3res switch (nfsstat3 status) { +case NFS3_OK: + FSSTAT3resok resok; +default: + FSSTAT3resfail resfail; +}; + +const FSF3_LINK = 0x0001; +const FSF3_SYMLINK = 0x0002; +const FSF3_HOMOGENEOUS = 0x0008; +const FSF3_CANSETTIME = 0x0010; + +struct FSINFOargs { + nfs_fh3 fsroot; +}; + +struct FSINFO3resok { + post_op_attr obj_attributes; + uint32 rtmax; + uint32 rtpref; + uint32 rtmult; + uint32 wtmax; + uint32 wtpref; + uint32 wtmult; + uint32 dtpref; + size3 maxfilesize; + nfstime3 time_delta; + uint32 properties; +}; + +struct FSINFO3resfail { + post_op_attr obj_attributes; +}; + +union FSINFO3res switch (nfsstat3 status) { +case NFS3_OK: + FSINFO3resok resok; +default: + FSINFO3resfail resfail; +}; + +struct PATHCONF3args { + nfs_fh3 object; +}; + +struct PATHCONF3resok { + post_op_attr obj_attributes; + uint32 linkmax; + uint32 name_max; + bool no_trunc; + bool chown_restricted; + bool case_insensitive; + bool case_preserving; +}; + +struct PATHCONF3resfail { + post_op_attr obj_attributes; +}; + +union PATHCONF3res switch (nfsstat3 status) { +case NFS3_OK: + PATHCONF3resok resok; +default: + PATHCONF3resfail resfail; +}; + +struct COMMIT3args { + nfs_fh3 file; + offset3 offset; + count3 count; +}; + +struct COMMIT3resok { + wcc_data file_wcc; + writeverf3 verf; +}; + +struct COMMIT3resfail { + wcc_data file_wcc; +}; + +union COMMIT3res switch (nfsstat3 status) { +case NFS3_OK: + COMMIT3resok resok; +default: + COMMIT3resfail resfail; +}; + +program NFS_PROGRAM { + version NFS_V3 { + + void + NFSPROC3_NULL(void) = 0; + + GETATTR3res + NFSPROC3_GETATTR(GETATTR3args) = 1; + + SETATTR3res + NFSPROC3_SETATTR(SETATTR3args) = 2; + + LOOKUP3res + NFSPROC3_LOOKUP(LOOKUP3args) = 3; + + ACCESS3res + NFSPROC3_ACCESS(ACCESS3args) = 4; + + READLINK3res + NFSPROC3_READLINK(READLINK3args) = 5; + + READ3res + NFSPROC3_READ(READ3args) = 6; + + WRITE3res + NFSPROC3_WRITE(WRITE3args) = 7; + + CREATE3res + NFSPROC3_CREATE(CREATE3args) = 8; + + MKDIR3res + NFSPROC3_MKDIR(MKDIR3args) = 9; + + SYMLINK3res + NFSPROC3_SYMLINK(SYMLINK3args) = 10; + + MKNOD3res + NFSPROC3_MKNOD(MKNOD3args) = 11; + + REMOVE3res + NFSPROC3_REMOVE(REMOVE3args) = 12; + + RMDIR3res + NFSPROC3_RMDIR(RMDIR3args) = 13; + + RENAME3res + NFSPROC3_RENAME(RENAME3args) = 14; + + LINK3res + NFSPROC3_LINK(LINK3args) = 15; + + READDIR3res + NFSPROC3_READDIR(READDIR3args) = 16; + + READDIRPLUS3res + NFSPROC3_READDIRPLUS(READDIRPLUS3args) = 17; + + FSSTAT3res + NFSPROC3_FSSTAT(FSSTAT3args) = 18; + + FSINFO3res + NFSPROC3_FSINFO(FSINFO3args) = 19; + + PATHCONF3res + NFSPROC3_PATHCONF(PATHCONF3args) = 20; + + COMMIT3res + NFSPROC3_COMMIT(COMMIT3args) = 21; + } = 3; +} = 100003;nfs4.1/xdrdef/portmap.x new file +51
-0@@ -0,0 +1,51 @@ +const PMAP_PORT = 111; /* portmapper port number */ + +struct mapping { + unsigned int prog; + unsigned int vers; + unsigned int prot; + unsigned int port; +}; + +const IPPROTO_TCP = 6; /* protocol number for TCP/IP */ +const IPPROTO_UDP = 17; /* protocol number for UDP/IP */ + +struct pmaplist { + mapping map; + pmaplist next; +}; + +struct call_args { + unsigned int prog; + unsigned int vers; + unsigned int proc; + opaque args<>; +}; + +struct call_result { + unsigned int port; + opaque res<>; +}; + +program PMAP_PROG { + version PMAP_VERS { + void + PMAPPROC_NULL(void) = 0; + + bool + PMAPPROC_SET(mapping) = 1; + + bool + PMAPPROC_UNSET(mapping) = 2; + + unsigned int + PMAPPROC_GETPORT(mapping) = 3; + + pmaplist + PMAPPROC_DUMP(void) = 4; + + call_result + PMAPPROC_CALLIT(call_args) = 5; + } = 2; +} = 100000; +Signature · 1 line
1.8.5.2 (Apple Git-48)
PATCH 13/17 dataserver: separate generic and 4.1 code
In reply to Weston Andros AdamsonSigned-off-by Weston Andros Adamson <[email protected]>
--- nfs4.1/dataserver.py | 99 ++++++++++++++++++++++++++++------------------------ 1 file changed, 54 insertions(+), 45 deletions(-)
nfs4.1/dataserver.py +54 -45
@@ -15,7 +15,7 @@ log = logging.getLogger("Dataserver Manager") op4 = nfs_ops.NFS4ops() -class DataServer41(object): +class DataServer(object): def __init__(self, server, port, path, flavor=rpc.AUTH_SYS, active=True, mdsds=True, multipath_servers=None, summary=None): self.mdsds = mdsds self.server = server
@@ -43,28 +43,44 @@ class DataServer41(object): self.active = True if not self.mdsds: self.connect() + self.make_root() def down(self): + self.disconnect() self.active = False - def connect(self): - # only support root with AUTH_SYS for now - s1 = rpc.security.instance(rpc.AUTH_SYS) - self.cred1 = s1.init_cred(uid=0, gid=0) - self.c1 = nfs4client.NFS4Client(self.server, self.port, - summary=self.summary) - self.c1.set_cred(self.cred1) - self.c1.null() - c = self.c1.new_client("DS.init_%s" % self.server) - # This is a hack to ensure MDS/DS communication path is at least - # as wide as the client/MDS channel (at least for linux client) - fore_attrs = type4.channel_attrs4(0, 16384, 16384, 2868, 8, 8, []) - self.sess = c.create_session(fore_attrs=fore_attrs) - self.make_root() + def reset(self): + self.down() + self.up() + + def get_netaddr4(self): + # STUB server multipathing not supported yet + uaddr = '.'.join([self.server, + str(self.port >> 8), + str(self.port & 0xff)]) + return type4.netaddr4(self.proto, uaddr) + + def get_multipath_netaddr4s(self): + netaddr4s = [] + for addr in self.multipath_servers: + server, port = addr + uaddr = '.'.join([server, + str(port >> 8), + str(port & 0xff)]) + proto = "tcp" + if server.find(':') >= 0: + proto = "tcp6" + + netaddr4s.append(type4.netaddr4(proto, uaddr)) + return netaddr4s + + def fh_to_name(self, mds_fh): + return hashlib.sha1("%r" % mds_fh).hexdigest() def disconnect(self): pass +class DataServer41(DataServer): def _execute(self, ops, exceptions=[], delay=5, maxretries=3): """ execute the NFS call If an error code is specified in the exceptions it means that the
@@ -85,36 +101,28 @@ class DataServer41(object): log.error("Too many retries with DS %s" % self.server) raise Exception("Dataserver communication retry error") elif res.status in state_errors: - self.disconnect() - self.connect() + self.reset() else: log.error("Unhandled status %s from DS %s" % (nfsstat4[res.status], self.server)) raise Exception("Dataserver communication error") - def get_netaddr4(self): - # STUB server multipathing not supported yet - uaddr = '.'.join([self.server, - str(self.port >> 8), - str(self.port & 0xff)]) - return type4.netaddr4(self.proto, uaddr) - - def get_multipath_netaddr4s(self): - netaddr4s = [] - for addr in self.multipath_servers: - server, port = addr - uaddr = '.'.join([server, - str(port >> 8), - str(port & 0xff)]) - proto = "tcp" - if server.find(':') >= 0: - proto = "tcp6" - - netaddr4s.append(type4.netaddr4(proto, uaddr)) - return netaddr4s - + def connect(self): + # only support root with AUTH_SYS for now + s1 = rpc.security.instance(rpc.AUTH_SYS) + self.cred1 = s1.init_cred(uid=0, gid=0) + self.c1 = nfs4client.NFS4Client(self.server, self.port, + summary=self.summary) + self.c1.set_cred(self.cred1) + self.c1.null() + c = self.c1.new_client("DS.init_%s" % self.server) + # This is a hack to ensure MDS/DS communication path is at least + # as wide as the client/MDS channel (at least for linux client) + fore_attrs = type4.channel_attrs4(0, 16384, 16384, 2868, 8, 8, []) + self.sess = c.create_session(fore_attrs=fore_attrs) - def make_root(self, attrs={const4.FATTR4_MODE:0777}): + def make_root(self): + attrs = {const4.FATTR4_MODE:0777} existing_path = [] kind = type4.createtype4(const4.NF4DIR) for comp in self.path:
@@ -133,12 +141,13 @@ class DataServer41(object): raise RuntimeError # XXX clean DS directory - def fh_to_name(self, mds_fh): - return hashlib.sha1("%r" % mds_fh).hexdigest() - - def open_file(self, mds_fh, seqid=0, - access=const4.OPEN4_SHARE_ACCESS_BOTH, deny=const4.OPEN4_SHARE_DENY_NONE, - attrs={const4.FATTR4_MODE: 0777}, owner = "mds", mode=const4.GUARDED4): + def open_file(self, mds_fh): + seqid=0 + access = const4.OPEN4_SHARE_ACCESS_BOTH + deny = const4.OPEN4_SHARE_DENY_NONE + attrs = {const4.FATTR4_MODE: 0777} + owner = "mds" + mode = const4.GUARDED4 verifier = self.sess.c.verifier openflag = type4.openflag4(const4.OPEN4_CREATE, type4.createhow4(mode, attrs, verifier)) name = self.fh_to_name(mds_fh)
Signature · 1 line
1.8.5.2 (Apple Git-48)
PATCH 14/17 4.1 server: add support for NFSv3 data servers
In reply to Weston Andros AdamsonAdd the NFSv3 client and a new DataServer class that handles DS ops using the v3 client.
DataServer3 is not used yet, as it requires flexfile layouts in order to pass a v3 DS to clients.
Tested with linux client mounting pnfs MDS via v4.1 (disabling pnfs) and a linux server acting as the v3 DS.
Signed-off-by Weston Andros Adamson <[email protected]>
--- nfs4.1/dataserver.py | 100 ++++++++++++++++++++++++++++ nfs4.1/nfs3client.py | 180 +++++++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 280 insertions(+) create mode 100644 nfs4.1/nfs3client.py
nfs4.1/dataserver.py +100 -0
@@ -3,9 +3,12 @@ import nfs4lib import xdrdef.nfs4_type as type4 from xdrdef.nfs4_pack import NFS4Packer import xdrdef.nfs4_const as const4 +import xdrdef.nfs3_type as type3 +import xdrdef.nfs3_const as const3 import time import logging import nfs4client +import nfs3client import hashlib import sys import nfs_ops
@@ -14,6 +17,7 @@ import socket log = logging.getLogger("Dataserver Manager") op4 = nfs_ops.NFS4ops() +op3 = nfs_ops.NFS3ops() class DataServer(object): def __init__(self, server, port, path, flavor=rpc.AUTH_SYS, active=True, mdsds=True, multipath_servers=None, summary=None):
@@ -203,6 +207,102 @@ class DataServer41(DataServer): attrdict = res.resarray[-1].obj_attributes return attrdict.get(const4.FATTR4_SIZE, 0) +class DataServer3(DataServer): + def _execute(self, procnum, procarg, exceptions=(), delay=5, maxretries=3): + """ execute the NFS call + If an error code is specified in the exceptions it means that the + caller wants to handle the error himself + """ + retry_errors = [] + while True: + res = self.c1.proc(procnum, procarg) + if res.status == const3.NFS3_OK or res.status in exceptions: + return res + elif res.status in retry_errors: + if maxretries > 0: + maxretries -= 1 + time.sleep(delay) + else: + log.error("Too many retries with DS %s" % self.server) + raise Exception("Dataserver communication retry error") + else: + log.error("Unhandled status %s from DS %s" % + (const3.nfsstat3[res.status], self.server)) + raise Exception("Dataserver communication error") + + def connect(self): + # only support root with AUTH_SYS for now + s1 = rpc.security.instance(rpc.AUTH_SYS) + self.cred1 = s1.init_cred(uid=0, gid=0) + self.c1 = nfs3client.NFS3Client(self.server, self.port, + summary=self.summary) + self.c1.set_cred(self.cred1) + self.rootfh = type3.nfs_fh3(self.c1.mntclnt.get_rootfh(self.path)) + self.c1.null() + + def make_root(self): + """ don't actually make a root path - we must use it as the export """ + need = const3.ACCESS3_READ | const3.ACCESS3_LOOKUP | \ + const3.ACCESS3_MODIFY | const3.ACCESS3_EXTEND + arg = op3.access(self.rootfh, need) + res = self._execute(const3.NFSPROC3_ACCESS, arg) + if res.resok.access != need: + raise RuntimeError + # XXX clean DS directory + + def open_file(self, mds_fh): + name = self.fh_to_name(mds_fh) + where = type3.diropargs3(self.rootfh, name) + attr = type3.sattr3(mode=type3.set_mode3(True, 0777), + uid=type3.set_uid3(True, 0), + gid=type3.set_gid3(True, 0), + size=type3.set_size3(False), + atime=type3.set_atime(False), + mtime=type3.set_mtime(False)) + how = type3.createhow3(const3.GUARDED, attr) + arg = op3.create(where, how) + res = self._execute(const3.NFSPROC3_CREATE, arg, + exceptions=(const3.NFS3ERR_EXIST,)) + + if res.status == const3.NFS3_OK: + self.filehandles[mds_fh] = (res.resok.obj.handle, None) + + else: + arg = op3.lookup(type3.diropargs3(self.rootfh, name)) + res = self._execute(const3.NFSPROC3_LOOKUP, arg) + + self.filehandles[mds_fh] = (res.resok.object, None) + + def close_file(self, mds_fh): + del self.filehandles[mds_fh] + + def read(self, fh, pos, count): + arg = op3.read(fh, pos, count) + res = self._execute(const3.NFSPROC3_READ, arg) + # XXX check res.status? + return res.resok.data + + def write(self, fh, pos, data): + arg = op3.write(fh, pos, len(data), const3.FILE_SYNC, data) + # There are all sorts of error handling issues here + res = self._execute(const3.NFSPROC3_WRITE, arg) + + def truncate(self, fh, size): + attr = type3.sattr3(mode=type3.set_mode3(False), + uid=type3.set_uid3(False), + gid=type3.set_gid3(False), + size=type3.set_size3(True, size), + atime=type3.set_atime(False), + mtime=type3.set_mtime(False)) + arg = op3.setattr(fh, attr, type3.sattrguard3(check=False)) + res = self._execute(const3.NFSPROC3_SETATTR, arg) + + def get_size(self, fh): + arg = op3.getattr(fh) + res = self._execute(const3.NFSPROC3_GETATTR, arg) + # XXX check res.status? + return res.resok.obj_attributes.size + class DSDevice(object): def __init__(self, mdsds):
nfs4.1/nfs3client.py new file +180 -0
180 lines not on this page · raw message
Signature · 1 line
1.8.5.2 (Apple Git-48)
PATCH 15/17 4.1 server: get rid of old op_getdeviceinfo
In reply to Weston Andros AdamsonThis has been commented out for a while.
Signed-off-by Weston Andros Adamson <[email protected]>
--- nfs4.1/nfs4server.py | 21 --------------------- 1 file changed, 21 deletions(-)
nfs4.1/nfs4server.py +0 -21
@@ -1870,27 +1870,6 @@ class NFS4Server(rpc.Server): new_cookie >= len(list)) return encode_status(NFS4_OK, res) -# def op_getdeviceinfo(self, arg, env): # STUB -# check_session(env) -# # check_cfh(env) -# # fs = env.cfh.fs -# # STUB - only deals with block volumes -# kind = arg.gdia_layout_type -# if kind != LAYOUT4_BLOCK_VOLUME: -# return encode_status(NFS4ERR_INVAL) -# # STUB - want to pull this from fs, not block module -# d = block.devices.get(arg.gdia_device_id, None) -# if d is None: -# return encode_status(NFS4ERR_INVAL) -# address = device_addr4(LAYOUT4_BLOCK_VOLUME, d.get_addr()) -# # Check that we don't exceed count -# p = nfs4lib.FancyNFS4Packer() -# p.pack_device_addr4(address) -# if len(p.get_buffer()) > arg.gdia_maxcount: -# return encode_status(NFS4ERR_TOOSMALL, gdir_mincount = len(p.get_buffer())) -# res = GETDEVICEINFO4resok(address, 0) -# return encode_status(NFS4_OK, res) - def op_getdeviceinfo(self, arg, env): # STUB # STUB - ignoring notifications check_session(env)
Signature · 1 line
1.8.5.2 (Apple Git-48)
PATCH 15/17 nfs41 svr: get rid of old op_getdeviceinfo
In reply to Weston Andros AdamsonThis has been commented out for a while.
Signed-off-by Weston Andros Adamson <[email protected]>
--- nfs4.1/nfs4server.py | 21 --------------------- 1 file changed, 21 deletions(-)
nfs4.1/nfs4server.py +0 -21
@@ -1870,27 +1870,6 @@ class NFS4Server(rpc.Server): new_cookie >= len(list)) return encode_status(NFS4_OK, res) -# def op_getdeviceinfo(self, arg, env): # STUB -# check_session(env) -# # check_cfh(env) -# # fs = env.cfh.fs -# # STUB - only deals with block volumes -# kind = arg.gdia_layout_type -# if kind != LAYOUT4_BLOCK_VOLUME: -# return encode_status(NFS4ERR_INVAL) -# # STUB - want to pull this from fs, not block module -# d = block.devices.get(arg.gdia_device_id, None) -# if d is None: -# return encode_status(NFS4ERR_INVAL) -# address = device_addr4(LAYOUT4_BLOCK_VOLUME, d.get_addr()) -# # Check that we don't exceed count -# p = nfs4lib.FancyNFS4Packer() -# p.pack_device_addr4(address) -# if len(p.get_buffer()) > arg.gdia_maxcount: -# return encode_status(NFS4ERR_TOOSMALL, gdir_mincount = len(p.get_buffer())) -# res = GETDEVICEINFO4resok(address, 0) -# return encode_status(NFS4_OK, res) - def op_getdeviceinfo(self, arg, env): # STUB # STUB - ignoring notifications check_session(env)
Signature · 1 line
1.8.5.2 (Apple Git-48)
PATCH 16/17 rpc: on socket error, close and mark pipe inactive
In reply to Weston Andros Adamsoncall _event_close() on socket errors instead of tracing back and mark the pipe as inactive so callers can reconnect as needed.
Signed-off-by Weston Andros Adamson <[email protected]>
--- rpc/rpc.py | 26 +++++++++++++++++++++++--- 1 file changed, 23 insertions(+), 3 deletions(-)
rpc/rpc.py +23 -3
@@ -351,6 +351,7 @@ class RpcPipe(Pipe): self._pending = {} # {xid:defer} self._lock = threading.Lock() # Protects fields below self._xid = 0 + self.set_active() def _get_xid(self): with self._lock:
@@ -358,6 +359,15 @@ class RpcPipe(Pipe): self._xid = inc_u32(out) return out + def set_active(self): + self._active = True + + def clear_active(self): + self._active = False + + def is_active(self): + return self._active + def listen(self, xid, timeout=None): """Wait for a reply to a CALL.""" self._pending[xid].wait(timeout)
@@ -500,15 +510,24 @@ class ConnectionHandler(object): log_p.warn(1, "polling error from %i" % fd) # STUB - now what? for fd in w: - self._event_write(fd) + try: + self._event_write(fd) + except socket.error, e: + self._event_close(fd) for fd in r: if fd in self.listeners: - self._event_connect_incoming(fd) + try: + self._event_connect_incoming(fd) + except socket.error, e: + self._event_close(fd) elif fd == self._alarm_poll.fileno(): commands = self._alarm_poll.recv(self.rsize) for c in commands: data = self._alarm.pop() - switch[c](data) + try: + switch[c](data) + except socket.error, e: + self._event_close(fd) else: try: data = self.sockets[fd].recv_records(self.rsize)
@@ -557,6 +576,7 @@ class ConnectionHandler(object): self.writelist -= temp self.readlist -= temp self.errlist -= temp + self.sockets[fd].clear_active() self.sockets[fd].close() del self.sockets[fd]
Signature · 1 line
1.8.5.2 (Apple Git-48)
PATCH 17/17 nfs3clnt: reconnect when sending on inactive pipe
In reply to Weston Andros AdamsonSigned-off-by Weston Andros Adamson <[email protected]>
--- nfs4.1/nfs3client.py | 45 +++++++++++++++++++++++++-------------------- 1 file changed, 25 insertions(+), 20 deletions(-)
nfs4.1/nfs3client.py +25 -20
@@ -30,14 +30,19 @@ class PORTMAPClient(rpc.Client): def __init__(self, host='localhost', port=PMAP_PORT): rpc.Client.__init__(self, PMAP_PROG, PMAP_VERS) self.server_address = (host, port) - self.c1 = self.connect(self.server_address) + self._pipe = None + + def get_pipe(self): + if not self._pipe or not self._pipe.is_active(): + self._pipe = self.connect(self.server_address) + return self._pipe def proc_async(self, procnum, procarg, credinfo=None, pipe=None, checks=True, packer=PORTMAPPacker): if credinfo is None: credinfo = self.default_cred if pipe is None: - pipe = self.c1 + pipe = self.get_pipe() p = packer(check_enum=checks, check_array=checks) arg_packer = getattr(p, 'pack_%s' % procarg.__class__.__name__) arg_packer(procarg)
@@ -51,7 +56,7 @@ class PORTMAPClient(rpc.Client): def listen(self, xid, restypename, pipe=None, timeout=10.0): if pipe is None: - pipe = self.c1 + pipe = self.get_pipe() header, data = pipe.listen(xid, timeout) if data: p = PORTMAPUnpacker(data)
@@ -69,14 +74,19 @@ class Mnt3Client(rpc.Client): def __init__(self, host='localhost', port=None): rpc.Client.__init__(self, MOUNT_PROGRAM, MOUNT_V3) self.server_address = (host, port) - self.c1 = self.connect(self.server_address) + self._pipe = None + + def get_pipe(self): + if not self._pipe or not self._pipe.is_active(): + self._pipe = self.connect(self.server_address) + return self._pipe def proc_async(self, procnum, procarg, credinfo=None, pipe=None, checks=True, packer=MNT3Packer): if credinfo is None: credinfo = self.default_cred if pipe is None: - pipe = self.c1 + pipe = self.get_pipe() p = packer(check_enum=checks, check_array=checks) arg_packer = getattr(p, 'pack_%s' % procarg.__class__.__name__) arg_packer(procarg)
@@ -90,7 +100,7 @@ class Mnt3Client(rpc.Client): def listen(self, xid, restypename, pipe=None, timeout=10.0): if pipe is None: - pipe = self.c1 + pipe = self.get_pipe() header, data = pipe.listen(xid, timeout) if data: p = MNT3Unpacker(data)
@@ -110,15 +120,6 @@ class Mnt3Client(rpc.Client): class NFS3Client(rpc.Client): def __init__(self, host='localhost', port=None, ctrl_proc=16, summary=None): rpc.Client.__init__(self, 100003, 3) - #self.prog = 0x40000000 - #self.versions = [1] # List of supported versions of prog - - #self.minorversion = minorversion - #self.minor_versions = [minorversion] - #self.tag = "default tag" - #self.impl_id = nfs_impl_id4("citi.umich.edu", "pynfs X.X", - # nfs4lib.get_nfstime()) - self.portmap = PORTMAPClient(host=host) self.mntport = self.portmap.get_port(MOUNT_PROGRAM, MOUNT_V3) if not port:
@@ -128,17 +129,21 @@ class NFS3Client(rpc.Client): self.verifier = struct.pack('>d', time.time()) self.server_address = (host, self.port) - self.c1 = self.connect(self.server_address) - #self.sessions = {} # XXX Really, this should be per server self.ctrl_proc = ctrl_proc self.summary = summary + self._pipe = None self.mntclnt = Mnt3Client(host=host, port=self.mntport) + def get_pipe(self): + if not self._pipe or not self._pipe.is_active(): + self._pipe = self.connect(self.server_address) + return self._pipe + def set_cred(self, credinfo): self.default_cred = credinfo def null_async(self, data=""): - return self.send_call(self.c1, 0, data) + return self.send_call(self.get_pipe(), 0, data) def null(self, *args, **kwargs): xid = self.null_async(*args, **kwargs)
@@ -149,7 +154,7 @@ class NFS3Client(rpc.Client): if credinfo is None: credinfo = self.default_cred if pipe is None: - pipe = self.c1 + pipe = self.get_pipe() p = packer(check_enum=checks, check_array=checks) arg_packer = getattr(p, 'pack_%s' % procarg.__class__.__name__) arg_packer(procarg)
@@ -167,7 +172,7 @@ class NFS3Client(rpc.Client): def listen(self, xid, procarg=None, pipe=None, timeout=10.0): if pipe is None: - pipe = self.c1 + pipe = self.get_pipe() header, data = pipe.listen(xid, timeout) if data: p = NFS3Unpacker(data)
Signature · 1 line
1.8.5.2 (Apple Git-48)
A message that is not in the archive (sent off-list or not yet received): [email protected]
Re: [PATCH pynfs 05/17] move .x files to subdir 'xdrdef'
Quoted 1 line · Weston Andros Adamson
nfs4.1/nfs4.x | 3290 ---------------------------
...
Quoted 1 line
nfs4.1/xdrdef/nfs4.x | 3290 +++++++++++++++++++++++++++
git should be able to show that as a rename instead of a huge add/delete pair. Maybe git-format-patch needs the -M option or something?
--b.
Re: [PATCH pynfs 05/17] move .x files to subdir 'xdrdef'
Huh, I didn’t know there was such an option. I’ll repost.
-dros
Quoted 9 lines · J. Bruce Fields
Quoted 1 line · Weston Andros Adamson
nfs4.1/nfs4.x | 3290 ---------------------------
...
Quoted 1 line
nfs4.1/xdrdef/nfs4.x | 3290 +++++++++++++++++++++++++++
git should be able to show that as a rename instead of a huge add/delete pair. Maybe git-format-patch needs the -M option or something?
--b.