IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an
email to Administrator. User accounts are meant only to access repo
and report issues and/or generate pull requests.
This is a purpose-specific Git hosting for
BaseALT
projects. Thank you for your understanding!
Только зарегистрированные пользователи имеют доступ к сервису!
Для получения аккаунта, обратитесь к администратору.
This adds the feature of the 'hostname' algorithm,
but provides it for all algorithms, including 'next_module'.
This can be used to deliberately break lock coherency, but
keep the devid/inode pair untouched, as this will only
alter file_id.extid:
vfs objects = fileid
fileid:algorithm = next_module
fileid:nolock_all_inodes = yes
This should be preferred unless someone is already using the
'hostname' algorithm.
Note this is only for testing (or read only shares if at all...)
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
This adds the feature of the 'fsname_nodirs' algorithm,
but provides it for all algorithms, including 'next_module'.
This can be used to deliberately break lock coherency, but
keep the devid/inode pair untouched, e.g.
vfs objects = fileid
fileid:algorithm = next_module
fileid:nolock_all_dirs = yes
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
This brings much more flexibility compared to:
- 'fsname_norootdir', 'fsname_norootdir_ext',
which only allow the nolock behavior for the share root
- 'fileid:nolockinode', which only gets a single inode number,
and ignores the devide id completely.
You can specify path names, which are relative to the shareroot
or absolute.
These names are only evaluated at SMB_VFS_CONNECT() time,
where they are converted into devide and inode pairs.
It means they are completely ignored if the path doesn't
exist yet, or is replaced by a new inode later.
This allows:
- 'fileid:algorithm = fsname_norootdir'
to be replaced by:
'fileid:algorithm = fsname' (the default)
'fileid:nolock_paths = .'
- 'fileid:algorithm = fsname_norootdir_ext'
to be replaced by:
'fileid:algorithm = fsname' (the default)
'fileid:nolock_paths = .'
'fileid:nolock_max_slots = 18446744073709551615'
And 'fileid:nolockinode = 1234567' and be replaced by
'fileid:nolock_paths = Very/Contended/Path' or
'fileid:nolock_paths = . Very/Contended/Path1 /data/conteded.dir',
if the share root and two additional inodes should be handled
by the 'nolock' behavior.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
This controlls the maximum number of concurrent locking slots
on each host. It specifies the maximal number of locking.tdb
records for a single inode.
It can be used to deliberately break lock coherency not
only between cluster nodes, but also between processes on
each node.
This allows administrators to control the behavior that's
currently only available by 'fsname_norootdir_ext' to
other cases as well.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
That way the file_id.extid is consistenly filled for all cases
where we deliberately break lock coherency.
This will simplify further changes and give administrators more
flexibility.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
This means we'll be able to provide the 'nolock' feature for all
directories also with other algorithms than 'fsname' in future.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
file_id.extid was filled with getpid() by 'fsname_norootdir_ext'.
However instead of forcing the existing 'hostname' algorithm for the 'nolock'
case, we'll now generate file_id.extid also based the hostname, vnn
and for 'fsname_norootdir_ext' also the pid.
This simplifies further changes and gives us the ability to generate stable
results for file_id.{devid,inode} based on the main algorithm. This is important
as we have a push_file_id_16() helper function used in places to generate a
stable identifier of the file that is also client visible and might be stored on
stable storage (acl_tdb, xattr_tdb). While the file_id.extid is only used
internally in volatile databases.
Review with: git show --patience
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
This can be use to get just bypass the fileid module for the
common case. But it allows 'fileid:nolockinode' (and in future
other things) to work in order to avoid lock contention
for all 'nolock' inodes.
If we would have started from scratch all the nolock
logic would have been in its own vfs module, just
altering file_id.extid
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
This way 'fsname_norootdir[_ext]' is not overwritten by
'fileid:nolockinode' and both can work independently.
It will also allow us to add more nolock inodes under
other conditions in the following changes.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
This makes the code much less magic (at least for me) and
it will allow further changes to be made easier.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Most places take twrp from a local struct smb_filename variable that the
function is working on. Some don't for various reasons:
o synthetic_smb_fname_split() is only called in very few places where we don't
expect twrp paths
o implementations of SMB_VFS_GETWD(), SMB_VFS_FS_CAPABILITIES() and
SMB_VFS_REALPATH() return the systems view of cwd and realpath without twrp info
o VFS modules implementing previous-versions support (vfs_ceph_snapshots,
vfs_shadow_copy2, vfs_snapper) synthesize raw paths that are passed to VFS NEXT
functions and therefor do not use twrp
o vfs_fruit: macOS doesn't support VSS
o vfs_recycle: in recycle_create_dir() we need a raw OS path to create a directory
o vfs_virusfilter: a few places where we need raw OS paths
o vfs_xattr_tdb: needs a raw OS path for SMB_VFS_NEXT_STAT()
o printing and rpc server: don't support VSS
o vfs_default_durable_reconnect: no Durable Handles on VSS handles, this might
be enhances in the future. No idea if Windows supports this.
o get_real_filename_full_scan: hm.... FIXME??
o get_original_lcomp: working on a raw path
o msdfs: doesn't support VSS
o vfs_get_ntquota: synthesizes an smb_filename from ".", so doesn't support VSS
even though VFS modules implement it
o fd_open: conn_rootdir_fname is a raw path
o msg_file_was_renamed: obvious
o open_np_file: pipes don't support VSS
o Python bindings: get's a raw path from the caller
o set_conn_connectpath: raw path
o set_conn_connectpath: raw path
o torture: gets raw paths from the caller
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
This can be used to deliberately break lock coherency between all smbd processes
in the whole cluster for the root directory of a share.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Autobuild-User(master): Ralph Böhme <slow@samba.org>
Autobuild-Date(master): Wed Sep 25 00:48:45 UTC 2019 on sn-devel-184
For this always ends up calling fileid_extid_mapping_zero(), so no change in
behaviour. This will change in a subsequent commit.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Bug: https://bugzilla.samba.org/show_bug.cgi?id=13741
Signed-off-by: Ralph Wuerthner <ralph.wuerthner@de.ibm.com>
Reviewed-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Stefan Metzmacher <metze@samba.org>
Autobuild-Date(master): Tue Jan 15 04:13:15 CET 2019 on sn-devel-144
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Ralph Böhme <slow@samba.org>
Autobuild-Date(master): Mon Jan 8 03:16:30 CET 2018 on sn-devel-144
Based-on-a-patch-by: Ralph Wuerthner <ralph.wuerthner@de.ibm.com>
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Sat Jan 6 04:41:24 CET 2018 on sn-devel-144
Based-on-a-patch-by: Ralph Wuerthner <ralph.wuerthner@de.ibm.com>
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Enabling fileid:algorithm = fsname_nodirs uses the hostname algorithm
for directories and thus breaks cluster lock coherence for directories.
Based-on-a-patch-by: Christian Ambach <ambi@samba.org>
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Using fileid:algorithm = hostname makes fileid generate
fileids based on the hostname. This breaks cluster lock coherence.
Based-on-a-patch-by: Christian Ambach <ambi@samba.org>
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
This is in preperation of adding an additional mapping function that
acts differently depending of the file type. No change in behaviour.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
When using the fsname or fsid algorithm a stat() and statfs() call is
required for all mounted file systems to generate the file_id. If e.g.
an NFS file system is unresponsive such a call might block and the smbd
process will become unresponsive. Add "fileid:fstype deny",
"fileid:fstype allow", "fileid:mntdir deny", and "fileid:mntdir allow"
options to ignore potentially unresponsive file systems.
See also https://lists.samba.org/archive/samba-technical/2016-January/111553.html
for a discussion about why this is useful.
Signed-off-by: Ralph Wuerthner <ralph.wuerthner@de.ibm.com>
Reviewed-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Signed-off-by: Christof Schmitt <cs@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Mon Dec 18 13:32:00 CET 2017 on sn-devel-144
Not currently used - no logic changes inside.
This will make it possible to pass down a long-lived talloc
context from the loading function for modules to use instead
of having them internally all use talloc_autofree_context()
which is a hidden global.
Updated all known module interface numbers, and added a
WHATSNEW.
Signed-off-by: Jeremy Allison <jra@samba.org>
Signed-off-by: Ralph Böhme <slow@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Sat Apr 22 01:17:00 CEST 2017 on sn-devel-144
They use talloc_tos() internally: hoist that up to the callers, some
of whom don't want to us talloc_tos().
A simple patch, but hits a lot of files.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
down to NEXT-> before initializing. This allows us to
do cleanup (by calling DISCONNECT) if initialization
fails. Also fix vfs_acl_xattr which was failing to
call the NEXT connect function.
Jeremy.
This patch introduces
struct stat_ex {
dev_t st_ex_dev;
ino_t st_ex_ino;
mode_t st_ex_mode;
nlink_t st_ex_nlink;
uid_t st_ex_uid;
gid_t st_ex_gid;
dev_t st_ex_rdev;
off_t st_ex_size;
struct timespec st_ex_atime;
struct timespec st_ex_mtime;
struct timespec st_ex_ctime;
struct timespec st_ex_btime; /* birthtime */
blksize_t st_ex_blksize;
blkcnt_t st_ex_blocks;
};
typedef struct stat_ex SMB_STRUCT_STAT;
It is really large because due to the friendly libc headers playing macro
tricks with fields like st_ino, so I renamed them to st_ex_xxx.
Why this change? To support birthtime, we already have quite a few #ifdef's at
places where it does not really belong. With a stat struct that we control, we
can consolidate the nanosecond timestamps and the birthtime deep in the VFS
stat calls.
At this moment it is triggered by a request to support the birthtime field for
GPFS. GPFS does not extend the system level struct stat, but instead has a
separate call that gets us the additional information beyond posix. Without
being able to do that within the VFS stat calls, that support would have to be
scattered around the main smbd code.
It will very likely break all the onefs modules, but I think the changes will
be reasonably easy to do.
and alters the device id depending on the configured algorithm.
The algorithm is configured via "fileid:algorithm":
- "fsname" (default) uses a uint64 hash over the mount point
- "fsid" uses the fsid returned from statfs()
This is needed for "clustering = yes" on some clusterfilesystems
metze
(This used to be commit 30f9171cca)
