Commit 61ca49a910 ("libceph: don't set global_id until we get an
auth ticket") delayed the setting of global_id too much. It is set
only after all tickets are received, but in pre-nautilus clusters an
auth ticket and the service tickets are obtained in separate steps
(for a total of three MAuth replies). When the service tickets are
requested, global_id is used to build an authorizer; if global_id is
still 0 we never get them and fail to establish the session.
Moving the setting of global_id into protocol implementations. This
way global_id can be set exactly when an auth ticket is received, not
sooner nor later.
Fixes: 61ca49a910 ("libceph: don't set global_id until we get an auth ticket")
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
There is no result to pass in msgr2 case because authentication
failures are reported through auth_bad_method frame and in MAuth
case an error is returned immediately.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
These options were introduced in 3.19 with support for message signing
and are rather useless, as explained in commit a51983e4dd ("libceph:
add nocephx_sign_messages option"). Deprecate them.
In case there is someone out there with a cluster that lacks support
for MSG_AUTH feature (very unlikely but has to be considered since we
haven't formally raised the bar from argonaut to bobtail yet), make
nocephx_sign_messages also waive MSG_AUTH requirement. This is probably
how it should have been done in the first place -- if we aren't going
to sign, requiring the signing feature makes no sense.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Implement msgr2.1 wire protocol, available since nautilus 14.2.11
and octopus 15.2.5. msgr2.0 wire protocol is not implemented -- it
has several security, integrity and robustness issues and therefore
considered deprecated.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
msgr2 supports two connection modes: crc (plain) and secure (on-wire
encryption). Connection mode is picked by server based on input from
client.
Introduce ms_mode option:
ms_mode=legacy - msgr1 (default)
ms_mode=crc - crc mode, if denied fail
ms_mode=secure - secure mode, if denied fail
ms_mode=prefer-crc - crc mode, if denied agree to secure mode
ms_mode=prefer-secure - secure mode, if denied agree to crc mode
ms_mode affects all connections, we don't separate connections to mons
like it's done in userspace with ms_client_mode vs ms_mon_client_mode.
For now the default is legacy, to be flipped to prefer-crc after some
time.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
For libceph, this ensures that libceph instance sharing (share option)
continues to work. For rbd, this avoids blocklisting alive lock owners
(locker addr is always LEGACY, while watcher addr is ANY in nautilus).
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
In preparation for msgr2, make the cluster send us maps with addrvecs
including both LEGACY and MSGR2 addrs instead of a single LEGACY addr.
This means advertising support for SERVER_NAUTILUS and also some older
features: SERVER_MIMIC, MONENC and MONNAMES.
MONNAMES and MONENC are actually pre-argonaut, we just never updated
ceph_monmap_decode() for them. Decoding is unconditional, see commit
23c625ce30 ("libceph: assume argonaut on the server side").
SERVER_MIMIC doesn't bear any meaning for the kernel client.
Since ceph_decode_entity_addrvec() is guarded by encoding version
checks (and in msgr2 case it is guarded implicitly by the fact that
server is speaking msgr2), we assume MSG_ADDR2 for it.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
In msgr2, initial authentication happens with an exchange of msgr2
control frames -- MAuth message and struct ceph_mon_request_header
aren't used. Make that optional.
Stop reporting cephx protocol as "x". Use "cephx" instead.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
- request service tickets together with auth ticket. Currently we get
auth ticket via CEPHX_GET_AUTH_SESSION_KEY op and then request service
tickets via CEPHX_GET_PRINCIPAL_SESSION_KEY op in a separate message.
Since nautilus, desired service tickets are shared togther with auth
ticket in CEPHX_GET_AUTH_SESSION_KEY reply.
- propagate session key and connection secret, if any. In preparation
for msgr2, update handle_reply() and verify_authorizer_reply() auth
ops to propagate session key and connection secret. Since nautilus,
if secure mode is negotiated, connection secret is shared either in
CEPHX_GET_AUTH_SESSION_KEY reply (for mons) or in a final authorizer
reply (for osds and mdses).
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
A pure move, no other changes.
Note that ceph_tcp_recv{msg,page}() and ceph_tcp_send{msg,page}()
helpers are also moved. msgr2 will bring its own, more efficient,
variants based on iov_iter. Switching msgr1 to them was considered
but decided against to avoid subtle regressions.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
In preparation for msgr2, define internal messenger <-> protocol
interface (as opposed to external messenger <-> client interface, which
is struct ceph_connection_operations) consisting of try_read(),
try_write(), revoke(), revoke_incoming(), opened(), reset_session() and
reset_protocol() ops. The semantics are exactly the same as they are
now.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
In preparation for msgr2, rename msgr1 specific states and move the
defines to the header file.
Also drop state transition comments. They don't cover all possible
transitions (e.g. NEGOTIATING -> STANDBY, etc) and currently do more
harm than good.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
unsigned long is a leftover from when con->state used to be a set of
bits managed with set_bit(), clear_bit(), etc. Save a bit of memory.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
The current setting allows the backoff to climb up to 5 minutes. This
is too high -- it becomes hard to tell whether the client is stuck on
something or just in backoff.
In userspace, ms_max_backoff is defaulted to 15 seconds. Let's do the
same.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
When we added the btime feature in mainline ceph, we had to extend
struct ceph_mds_request_args so that it could be set. Implement the same
in the kernel client.
Rename ceph_mds_request_head with a _old extension, and a union
ceph_mds_request_args_ext to allow for the extended size of the new
header format.
Add the appropriate code to handle both formats in struct
create_request_message and key the behavior on whether the peer supports
CEPH_FEATURE_FS_BTIME.
The gid_list field in the payload is now populated from the saved
credential. For now, we don't add any support for setting the btime via
setattr, but this does enable us to add that in the future.
[ idryomov: break unnecessarily long lines ]
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: Xiubo Li <xiubli@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
When setting the file/dir layout, it may need data pool info. So
in mds server, it needs to check the osdmap. At present, if mds
doesn't find the data pool specified, it will try to get the latest
osdmap. Now if pass the osd epoch for setxattr, the mds server can
only check this epoch of osdmap.
URL: https://tracker.ceph.com/issues/48504
Signed-off-by: Xiubo Li <xiubli@redhat.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
1. monitor's default port is defined by CEPH_MON_PORT
2. CEPH_PORT_START and CEPH_PORT_LAST are not needed.
Signed-off-by: Changcheng Liu <changcheng.liu@aliyun.com>
Reviewed-by: Ilya Dryomov <idryomov@gmail.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
When recovering a session (a'la recover_session=clean), we want to do
all of the operations that we do on a forced umount, but changing the
mount state to SHUTDOWN is can cause queued MDS requests to fail when
the session comes back. Most of those can idle until the session is
recovered in this situation.
Reserve SHUTDOWN state for forced umount, and make a new RECOVER state
for the forced reconnect situation. Change several tests for equality with
SHUTDOWN to test for that or RECOVER.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: Xiubo Li <xiubli@redhat.com>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Replace a global map->crush_workspace (protected by a global mutex)
with a list of workspaces, up to the number of CPUs + 1.
This is based on a patch from Robin Geuze <robing@nl.team.blue>.
Robin and his team have observed a 10-20% increase in IOPS on all
queue depths and lower CPU usage as well on a high-end all-NVMe
100GbE cluster.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
When doing some testing recently, I hit some page allocation failures
on mount, when creating the wb_pagevec_pool for the mount. That
requires 128k (32 contiguous pages), and after thrashing the memory
during an xfstests run, sometimes that would fail.
128k for each mount seems like a lot to hold in reserve for a rainy
day, so let's change this to a global mempool that gets allocated
when the module is plugged in.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: Ilya Dryomov <idryomov@gmail.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Drop duplicated words "down" and "the" in fs/ceph/.
[ idryomov: merge into a single patch ]
Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
This will send the caps/read/write/metadata metrics to any available MDS
once per second, which will be the same as the userland client. It will
skip the MDS sessions which don't support the metric collection, as the
MDSs will close socket connections when they get an unknown type
message.
We can disable the metric sending via the disable_send_metrics module
parameter.
[ jlayton: fix up endianness bug in ceph_mdsc_send_metrics() ]
URL: https://tracker.ceph.com/issues/43215
Signed-off-by: Xiubo Li <xiubli@redhat.com>
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
The caller can just ignore the return. No need for this wrapper that
just casts the other function to void.
[ idryomov: argument alignment ]
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: Ilya Dryomov <idryomov@gmail.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
osd_req_flags is overly general and doesn't suit its only user
(read_from_replica option) well:
- applying osd_req_flags in account_request() affects all OSD
requests, including linger (i.e. watch and notify). However,
linger requests should always go to the primary even though
some of them are reads (e.g. notify has side effects but it
is a read because it doesn't result in mutation on the OSDs).
- calls to class methods that are reads are allowed to go to
the replica, but most such calls issued for "rbd map" and/or
exclusive lock transitions are requested to be resent to the
primary via EAGAIN, doubling the latency.
Get rid of global osd_req_flags and set read_from_replica flag
only on specific OSD requests instead.
Fixes: 8ad44d5e0d ("libceph: read_from_replica option")
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Allow indicating future I/O pattern via flags. This is supported since
Kraken (and bluestore persists flags together with expected_object_size
and expected_write_size).
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Reviewed-by: Jason Dillaman <dillaman@redhat.com>
Expose replica reads through read_from_replica=balance and
read_from_replica=localize. The default is to read from primary
(read_from_replica=no).
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
OSD-side issues with reads from replica have been resolved in
Octopus. Reading from replica should be safe wrt. unstable or
uncommitted state now, so add support for balanced and localized
reads.
There are two cases when a read from replica can't be served:
- OSD may silently drop the request, expecting the client to
notice that the acting set has changed and resend via the usual
means (handled with t->used_replica)
- OSD may return EAGAIN, expecting the client to resend to the
primary, ignoring replica read flags (see handle_reply())
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Allow expressing client's location in terms of CRUSH hierarchy as
a set of (bucket type name, bucket name) pairs. The userspace syntax
"crush_location = key1=value1 key2=value2" is incompatible with mount
options and needed adaptation. Key-value pairs are separated by '|'
and we use ':' instead of '=' to separate keys from values. So for:
crush_location = host=foo rack=bar
one would write:
crush_location=host:foo|rack:bar
As in userspace, "multipath" locations are supported, so indicating
locality for parallel hierarchies is possible:
crush_location=rack:foo1|rack:foo2|datacenter:bar
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Needed for the next commit and useful for ceph_pg_pool_info tree as
well. I'm leaving the asserting helper in for now, but we should look
at getting rid of it in the future.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
The current codebase makes use of the zero-length array language
extension to the C90 standard, but the preferred mechanism to declare
variable-length types such as these ones is a flexible array member[1][2],
introduced in C99:
struct foo {
int stuff;
struct boo array[];
};
By making use of the mechanism above, we will get a compiler warning
in case the flexible array does not occur last in the structure, which
will help us prevent some kind of undefined behavior bugs from being
inadvertently introduced[3] to the codebase from now on.
Also, notice that, dynamic memory allocations won't be affected by
this change:
"Flexible array members have incomplete type, and so the sizeof operator
may not be applied. As a quirk of the original implementation of
zero-length arrays, sizeof evaluates to zero."[1]
sizeof(flexible-array-member) triggers a warning because flexible array
members have incomplete type[1]. There are some instances of code in
which the sizeof operator is being incorrectly/erroneously applied to
zero-length arrays and the result is zero. Such instances may be hiding
some bugs. So, this work (flexible-array member conversions) will also
help to get completely rid of those sorts of issues.
This issue was found with the help of Coccinelle.
[1] https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html
[2] https://github.com/KSPP/linux/issues/21
[3] commit 7649773293 ("cxgb3/l2t: Fix undefined behaviour")
Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Reviewed-by: Ilya Dryomov <idryomov@gmail.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Calculate the latency for OSD read requests. Add a new r_end_stamp
field to struct ceph_osd_request that will hold the time of that
the reply was received. Use that to calculate the RTT for each call,
and divide the sum of those by number of calls to get averate RTT.
Keep a tally of RTT for OSD writes and number of calls to track average
latency of OSD writes.
URL: https://tracker.ceph.com/issues/43215
Signed-off-by: Xiubo Li <xiubli@redhat.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Add i_last_rd and i_last_wr to ceph_inode_info. These fields are
used to track the last time the client acquired read/write caps for
the inode.
If there is no read/write on an inode for 'caps_wanted_delay_max'
seconds, __ceph_caps_file_wanted() does not request caps for read/write
even there are open files.
Call __ceph_touch_fmode() for dir operations. __ceph_caps_file_wanted()
calculates dir's wanted caps according to last dir read/modification. If
there is recent dir read, dir inode wants CEPH_CAP_ANY_SHARED caps. If
there is recent dir modification, also wants CEPH_CAP_FILE_EXCL.
Readdir is a special case. Dir inode wants CEPH_CAP_FILE_EXCL after
readdir, as with that, modifications do not need to release
CEPH_CAP_FILE_SHARED or invalidate all dentry leases issued by readdir.
Signed-off-by: "Yan, Zheng" <zyan@redhat.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
With the Octopus release, the MDS will hand out directory create caps.
If we have Fxc caps on the directory, and complete directory information
or a known negative dentry, then we can return without waiting on the
reply, allowing the open() call to return very quickly to userland.
We use the normal ceph_fill_inode() routine to fill in the inode, so we
have to gin up some reply inode information with what we'd expect the
newly-created inode to have. The client assumes that it has a full set
of caps on the new inode, and that the MDS will revoke them when there
is conflicting access.
This functionality is gated on the wsync/nowsync mount options.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Track and correctly handle directory caps for asynchronous operations.
Add aliases for Frc caps that we now designate at Dcu caps (when dealing
with directories).
Unlike file caps, we don't reclaim these when the session goes away, and
instead preemptively release them. In-flight async dirops are instead
handled during reconnect phase. The client needs to re-do a synchronous
operation in order to re-get directory caps.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Newer versions of the MDS will flag a dentry as "primary". In later
patches, we'll need to consult this info, so track it in di->flags.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
...and ensure that such requests are never queued. The MDS has need to
know that a request is asynchronous so add flags and proper
infrastructure for that.
Also, delegated inode numbers and directory caps are associated with the
session, so ensure that async requests are always transmitted on the
first attempt and are never queued to wait for session reestablishment.
If it does end up looking like we'll need to queue the request, then
have it return -EJUKEBOX so the caller can reattempt with a synchronous
request.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
On my machine (x86_64) this struct is 952 bytes, which gets rounded up
to 1024 by kmalloc. Move this to a dedicated slabcache, so we can
allocate them without the extra 72 bytes of overhead per.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: Ilya Dryomov <idryomov@gmail.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Since these helpers are only used by ceph.ko, move them there and
rename them with _sync_ qualifiers.
Signed-off-by: Xiubo Li <xiubli@redhat.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Although CEPH_DEFINE_SHOW_FUNC is much older, it now duplicates
DEFINE_SHOW_ATTRIBUTE from linux/seq_file.h.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Ilya Dryomov
Jeff Layton
Xiubo Li
Liu, Changcheng
Randy Dunlap
Gustavo A. R. Silva
Yan, Zheng