AuxXxilium/linux_dsm_epyc7002
1
0
Fork 0
mirror of https://github.com/AuxXxilium/linux_dsm_epyc7002.git synced 2024-12-28 11:18:45 +07:00
Code Issues Actions Packages Projects Releases Wiki Activity
80b79dd0e2
linux_dsm_epyc7002/fs/ceph/mdsmap.c

396 lines
9.2 KiB
C
Raw Normal View History

ceph: factor out libceph from Ceph file system This factors out protocol and low-level storage parts of ceph into a separate libceph module living in net/ceph and include/linux/ceph. This is mostly a matter of moving files around. However, a few key pieces of the interface change as well: - ceph_client becomes ceph_fs_client and ceph_client, where the latter captures the mon and osd clients, and the fs_client gets the mds client and file system specific pieces. - Mount option parsing and debugfs setup is correspondingly broken into two pieces. - The mon client gets a generic handler callback for otherwise unknown messages (mds map, in this case). - The basic supported/required feature bits can be expanded (and are by ceph_fs_client). No functional change, aside from some subtle error handling cases that got cleaned up in the refactoring process. Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-07 05:14:15 +07:00
#include <linux/ceph/ceph_debug.h>
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
#include <linux/bug.h>
#include <linux/err.h>
#include <linux/random.h>
#include <linux/slab.h>
#include <linux/types.h>
ceph: factor out libceph from Ceph file system This factors out protocol and low-level storage parts of ceph into a separate libceph module living in net/ceph and include/linux/ceph. This is mostly a matter of moving files around. However, a few key pieces of the interface change as well: - ceph_client becomes ceph_fs_client and ceph_client, where the latter captures the mon and osd clients, and the fs_client gets the mds client and file system specific pieces. - Mount option parsing and debugfs setup is correspondingly broken into two pieces. - The mon client gets a generic handler callback for otherwise unknown messages (mds map, in this case). - The basic supported/required feature bits can be expanded (and are by ceph_fs_client). No functional change, aside from some subtle error handling cases that got cleaned up in the refactoring process. Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-07 05:14:15 +07:00
#include <linux/ceph/mdsmap.h>
#include <linux/ceph/messenger.h>
#include <linux/ceph/decode.h>
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
#include "super.h"
/*
* choose a random mds that is "up" (i.e. has a state > 0), or -1.
*/
int ceph_mdsmap_get_random_mds(struct ceph_mdsmap *m)
{
int n = 0;
int i;
ceph: Use pseudo-random numbers to choose mds We don't need to use up entropy to choose an mds, so use prandom_u32() to get a pseudo-random number. Also, we don't need to choose a random mds if only one mds is available, so add special casing for the common case. Fixes http://tracker.ceph.com/issues/3579 Signed-off-by: Sam Lang <sam.lang@inktank.com> Reviewed-by: Greg Farnum <greg@inktank.com> Reviewed-by: Alex Elder <elder@inktank.com>
2013-04-10 04:49:11 +07:00
/* special case for one mds */
ceph: allow connecting to mds whose rank >= mdsmap::m_max_mds mdsmap::m_max_mds is the expected count of active mds. It's not the max rank of active mds. User can decrease mdsmap::m_max_mds, but does not stop mds whose rank >= mdsmap::m_max_mds. Signed-off-by: "Yan, Zheng" <zyan@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-03-28 16:04:13 +07:00
if (1 == m->m_num_mds && m->m_info[0].state > 0)
ceph: Use pseudo-random numbers to choose mds We don't need to use up entropy to choose an mds, so use prandom_u32() to get a pseudo-random number. Also, we don't need to choose a random mds if only one mds is available, so add special casing for the common case. Fixes http://tracker.ceph.com/issues/3579 Signed-off-by: Sam Lang <sam.lang@inktank.com> Reviewed-by: Greg Farnum <greg@inktank.com> Reviewed-by: Alex Elder <elder@inktank.com>
2013-04-10 04:49:11 +07:00
return 0;
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
/* count */
ceph: allow connecting to mds whose rank >= mdsmap::m_max_mds mdsmap::m_max_mds is the expected count of active mds. It's not the max rank of active mds. User can decrease mdsmap::m_max_mds, but does not stop mds whose rank >= mdsmap::m_max_mds. Signed-off-by: "Yan, Zheng" <zyan@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-03-28 16:04:13 +07:00
for (i = 0; i < m->m_num_mds; i++)
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
if (m->m_info[i].state > 0)
n++;
if (n == 0)
return -1;
/* pick */
ceph: Use pseudo-random numbers to choose mds We don't need to use up entropy to choose an mds, so use prandom_u32() to get a pseudo-random number. Also, we don't need to choose a random mds if only one mds is available, so add special casing for the common case. Fixes http://tracker.ceph.com/issues/3579 Signed-off-by: Sam Lang <sam.lang@inktank.com> Reviewed-by: Greg Farnum <greg@inktank.com> Reviewed-by: Alex Elder <elder@inktank.com>
2013-04-10 04:49:11 +07:00
n = prandom_u32() % n;
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
i = 0;
for (i = 0; n > 0; i++, n--)
while (m->m_info[i].state <= 0)
i++;
return i;
}
ceph: check availability of mds cluster on mount Signed-off-by: Yan, Zheng <zyan@redhat.com>
2016-11-10 15:02:06 +07:00
#define __decode_and_drop_type(p, end, type, bad) \
do { \
if (*p + sizeof(type) > end) \
goto bad; \
*p += sizeof(type); \
} while (0)
#define __decode_and_drop_set(p, end, type, bad) \
do { \
u32 n; \
size_t need; \
ceph_decode_32_safe(p, end, n, bad); \
need = sizeof(type) * n; \
ceph_decode_need(p, end, need, bad); \
*p += need; \
} while (0)
#define __decode_and_drop_map(p, end, ktype, vtype, bad) \
do { \
u32 n; \
size_t need; \
ceph_decode_32_safe(p, end, n, bad); \
need = (sizeof(ktype) + sizeof(vtype)) * n; \
ceph_decode_need(p, end, need, bad); \
*p += need; \
} while (0)
static int __decode_and_drop_compat_set(void **p, void* end)
{
int i;
/* compat, ro_compat, incompat*/
for (i = 0; i < 3; i++) {
u32 n;
ceph_decode_need(p, end, sizeof(u64) + sizeof(u32), bad);
/* mask */
*p += sizeof(u64);
/* names (map<u64, string>) */
n = ceph_decode_32(p);
while (n-- > 0) {
u32 len;
ceph_decode_need(p, end, sizeof(u64) + sizeof(u32),
bad);
*p += sizeof(u64);
len = ceph_decode_32(p);
ceph_decode_need(p, end, len, bad);
*p += len;
}
}
return 0;
bad:
return -1;
}
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
/*
* Decode an MDS map
*
* Ignore any fields we don't care about (there are quite a few of
* them).
*/
struct ceph_mdsmap *ceph_mdsmap_decode(void **p, void *end)
{
struct ceph_mdsmap *m;
ceph: hex dump corrupt server data to KERN_DEBUG Also, print fsid using standard format, NOT hex dump. Signed-off-by: Sage Weil <sage@newdream.net>
2009-12-15 06:13:47 +07:00
const void *start = *p;
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
int i, j, n;
int err = -EINVAL;
ceph: CEPH_FEATURE_MDSENC support Signed-off-by: Yan, Zheng <zyan@redhat.com>
2016-03-31 14:53:01 +07:00
u8 mdsmap_v, mdsmap_cv;
ceph: check availability of mds cluster on mount Signed-off-by: Yan, Zheng <zyan@redhat.com>
2016-11-10 15:02:06 +07:00
u16 mdsmap_ev;
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
m = kzalloc(sizeof(*m), GFP_NOFS);
if (m == NULL)
return ERR_PTR(-ENOMEM);
ceph: CEPH_FEATURE_MDSENC support Signed-off-by: Yan, Zheng <zyan@redhat.com>
2016-03-31 14:53:01 +07:00
ceph_decode_need(p, end, 1 + 1, bad);
mdsmap_v = ceph_decode_8(p);
mdsmap_cv = ceph_decode_8(p);
if (mdsmap_v >= 4) {
u32 mdsmap_len;
ceph_decode_32_safe(p, end, mdsmap_len, bad);
if (end < *p + mdsmap_len)
goto bad;
end = *p + mdsmap_len;
ceph: update support for PGID64, PGPOOL3, OSDENC protocol features Support (and require) the PGID64, PGPOOL3, and OSDENC protocol features. These have been present in ceph.git since v0.42, Feb 2012. Require these features to simplify support; nobody is running older userspace. Note that the new request and reply encoding is still not in place, so the new code is not yet functional. Signed-off-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@inktank.com>
2013-02-24 01:41:09 +07:00
}
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
ceph_decode_need(p, end, 8*sizeof(u32) + sizeof(u64), bad);
ceph: convert encode/decode macros to inlines This avoids the fugly pass by reference and makes the code a bit easier to read. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-14 23:59:09 +07:00
m->m_epoch = ceph_decode_32(p);
m->m_client_epoch = ceph_decode_32(p);
m->m_last_failure = ceph_decode_32(p);
m->m_root = ceph_decode_32(p);
m->m_session_timeout = ceph_decode_32(p);
m->m_session_autoclose = ceph_decode_32(p);
m->m_max_file_size = ceph_decode_64(p);
m->m_max_mds = ceph_decode_32(p);
ceph: allow connecting to mds whose rank >= mdsmap::m_max_mds mdsmap::m_max_mds is the expected count of active mds. It's not the max rank of active mds. User can decrease mdsmap::m_max_mds, but does not stop mds whose rank >= mdsmap::m_max_mds. Signed-off-by: "Yan, Zheng" <zyan@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-03-28 16:04:13 +07:00
m->m_num_mds = m->m_max_mds;
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
ceph: allow connecting to mds whose rank >= mdsmap::m_max_mds mdsmap::m_max_mds is the expected count of active mds. It's not the max rank of active mds. User can decrease mdsmap::m_max_mds, but does not stop mds whose rank >= mdsmap::m_max_mds. Signed-off-by: "Yan, Zheng" <zyan@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-03-28 16:04:13 +07:00
m->m_info = kcalloc(m->m_num_mds, sizeof(*m->m_info), GFP_NOFS);
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
if (m->m_info == NULL)
ceph: check availability of mds cluster on mount Signed-off-by: Yan, Zheng <zyan@redhat.com>
2016-11-10 15:02:06 +07:00
goto nomem;
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
/* pick out active nodes from mds_info (state > 0) */
ceph: convert encode/decode macros to inlines This avoids the fugly pass by reference and makes the code a bit easier to read. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-14 23:59:09 +07:00
n = ceph_decode_32(p);
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
for (i = 0; i < n; i++) {
ceph: decode updated mdsmap format The mds map now uses the global_id as the 'key' (instead of the addr, which was a poor choice). This is protocol change. Signed-off-by: Sage Weil <sage@newdream.net>
2009-11-20 06:31:50 +07:00
u64 global_id;
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
u32 namelen;
s32 mds, inc, state;
u64 state_seq;
ceph: CEPH_FEATURE_MDSENC support Signed-off-by: Yan, Zheng <zyan@redhat.com>
2016-03-31 14:53:01 +07:00
u8 info_v;
void *info_end = NULL;
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
struct ceph_entity_addr addr;
u32 num_export_targets;
void *pexport_targets = NULL;
ceph: track laggy state of mds from mdsmap Signed-off-by: Sage Weil <sage@newdream.net>
2010-06-18 04:19:01 +07:00
struct ceph_timespec laggy_since;
ceph: tidy ceph_mdsmap_decode() a little I introduced a new temporary variable "info" instead of "m->m_info[mds]". Also I reversed the if condition and pulled everything in one indent level. Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Reviewed-by: Alex Elder <elder@inktank.com>
2013-05-29 18:46:56 +07:00
struct ceph_mds_info *info;
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
ceph: CEPH_FEATURE_MDSENC support Signed-off-by: Yan, Zheng <zyan@redhat.com>
2016-03-31 14:53:01 +07:00
ceph_decode_need(p, end, sizeof(u64) + 1, bad);
ceph: decode updated mdsmap format The mds map now uses the global_id as the 'key' (instead of the addr, which was a poor choice). This is protocol change. Signed-off-by: Sage Weil <sage@newdream.net>
2009-11-20 06:31:50 +07:00
global_id = ceph_decode_64(p);
ceph: CEPH_FEATURE_MDSENC support Signed-off-by: Yan, Zheng <zyan@redhat.com>
2016-03-31 14:53:01 +07:00
info_v= ceph_decode_8(p);
if (info_v >= 4) {
u32 info_len;
u8 info_cv;
ceph_decode_need(p, end, 1 + sizeof(u32), bad);
info_cv = ceph_decode_8(p);
info_len = ceph_decode_32(p);
info_end = *p + info_len;
if (info_end > end)
goto bad;
}
ceph_decode_need(p, end, sizeof(u64) + sizeof(u32), bad);
ceph: decode updated mdsmap format The mds map now uses the global_id as the 'key' (instead of the addr, which was a poor choice). This is protocol change. Signed-off-by: Sage Weil <sage@newdream.net>
2009-11-20 06:31:50 +07:00
*p += sizeof(u64);
ceph: convert encode/decode macros to inlines This avoids the fugly pass by reference and makes the code a bit easier to read. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-14 23:59:09 +07:00
namelen = ceph_decode_32(p); /* skip mds name */
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
*p += namelen;
ceph_decode_need(p, end,
ceph: fix mdsmap decoding when multiple mds's are present A misplaced sizeof() around namelen was throwing things off. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-08 06:38:19 +07:00
4*sizeof(u32) + sizeof(u64) +
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
sizeof(addr) + sizeof(struct ceph_timespec),
bad);
ceph: convert encode/decode macros to inlines This avoids the fugly pass by reference and makes the code a bit easier to read. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-14 23:59:09 +07:00
mds = ceph_decode_32(p);
inc = ceph_decode_32(p);
state = ceph_decode_32(p);
state_seq = ceph_decode_64(p);
ceph: decode updated mdsmap format The mds map now uses the global_id as the 'key' (instead of the addr, which was a poor choice). This is protocol change. Signed-off-by: Sage Weil <sage@newdream.net>
2009-11-20 06:31:50 +07:00
ceph_decode_copy(p, &addr, sizeof(addr));
ceph_decode_addr(&addr);
ceph: track laggy state of mds from mdsmap Signed-off-by: Sage Weil <sage@newdream.net>
2010-06-18 04:19:01 +07:00
ceph_decode_copy(p, &laggy_since, sizeof(laggy_since));
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
*p += sizeof(u32);
ceph_decode_32_safe(p, end, namelen, bad);
ceph: fix mdsmap decoding when multiple mds's are present A misplaced sizeof() around namelen was throwing things off. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-08 06:38:19 +07:00
*p += namelen;
ceph: CEPH_FEATURE_MDSENC support Signed-off-by: Yan, Zheng <zyan@redhat.com>
2016-03-31 14:53:01 +07:00
if (info_v >= 2) {
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
ceph_decode_32_safe(p, end, num_export_targets, bad);
pexport_targets = *p;
ceph: fix mdsmap decoding when multiple mds's are present A misplaced sizeof() around namelen was throwing things off. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-08 06:38:19 +07:00
*p += num_export_targets * sizeof(u32);
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
} else {
num_export_targets = 0;
}
ceph: CEPH_FEATURE_MDSENC support Signed-off-by: Yan, Zheng <zyan@redhat.com>
2016-03-31 14:53:01 +07:00
if (info_end && *p != info_end) {
if (*p > info_end)
goto bad;
*p = info_end;
}
ceph: decode updated mdsmap format The mds map now uses the global_id as the 'key' (instead of the addr, which was a poor choice). This is protocol change. Signed-off-by: Sage Weil <sage@newdream.net>
2009-11-20 06:31:50 +07:00
dout("mdsmap_decode %d/%d %lld mds%d.%d %s %s\n",
ceph: factor out libceph from Ceph file system This factors out protocol and low-level storage parts of ceph into a separate libceph module living in net/ceph and include/linux/ceph. This is mostly a matter of moving files around. However, a few key pieces of the interface change as well: - ceph_client becomes ceph_fs_client and ceph_client, where the latter captures the mon and osd clients, and the fs_client gets the mds client and file system specific pieces. - Mount option parsing and debugfs setup is correspondingly broken into two pieces. - The mon client gets a generic handler callback for otherwise unknown messages (mds map, in this case). - The basic supported/required feature bits can be expanded (and are by ceph_fs_client). No functional change, aside from some subtle error handling cases that got cleaned up in the refactoring process. Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-07 05:14:15 +07:00
i+1, n, global_id, mds, inc,
ceph_pr_addr(&addr.in_addr),
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
ceph_mds_state_name(state));
ceph: tidy ceph_mdsmap_decode() a little I introduced a new temporary variable "info" instead of "m->m_info[mds]". Also I reversed the if condition and pulled everything in one indent level. Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Reviewed-by: Alex Elder <elder@inktank.com>
2013-05-29 18:46:56 +07:00
ceph: allow connecting to mds whose rank >= mdsmap::m_max_mds mdsmap::m_max_mds is the expected count of active mds. It's not the max rank of active mds. User can decrease mdsmap::m_max_mds, but does not stop mds whose rank >= mdsmap::m_max_mds. Signed-off-by: "Yan, Zheng" <zyan@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-03-28 16:04:13 +07:00
if (mds < 0 || state <= 0)
ceph: tidy ceph_mdsmap_decode() a little I introduced a new temporary variable "info" instead of "m->m_info[mds]". Also I reversed the if condition and pulled everything in one indent level. Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Reviewed-by: Alex Elder <elder@inktank.com>
2013-05-29 18:46:56 +07:00
continue;
ceph: allow connecting to mds whose rank >= mdsmap::m_max_mds mdsmap::m_max_mds is the expected count of active mds. It's not the max rank of active mds. User can decrease mdsmap::m_max_mds, but does not stop mds whose rank >= mdsmap::m_max_mds. Signed-off-by: "Yan, Zheng" <zyan@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-03-28 16:04:13 +07:00
if (mds >= m->m_num_mds) {
int new_num = max(mds + 1, m->m_num_mds * 2);
void *new_m_info = krealloc(m->m_info,
new_num * sizeof(*m->m_info),
GFP_NOFS | __GFP_ZERO);
if (!new_m_info)
goto nomem;
m->m_info = new_m_info;
m->m_num_mds = new_num;
}
ceph: tidy ceph_mdsmap_decode() a little I introduced a new temporary variable "info" instead of "m->m_info[mds]". Also I reversed the if condition and pulled everything in one indent level. Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Reviewed-by: Alex Elder <elder@inktank.com>
2013-05-29 18:46:56 +07:00
info = &m->m_info[mds];
info->global_id = global_id;
info->state = state;
info->addr = addr;
info->laggy = (laggy_since.tv_sec != 0 ||
laggy_since.tv_nsec != 0);
info->num_export_targets = num_export_targets;
if (num_export_targets) {
info->export_targets = kcalloc(num_export_targets,
sizeof(u32), GFP_NOFS);
if (info->export_targets == NULL)
ceph: check availability of mds cluster on mount Signed-off-by: Yan, Zheng <zyan@redhat.com>
2016-11-10 15:02:06 +07:00
goto nomem;
ceph: tidy ceph_mdsmap_decode() a little I introduced a new temporary variable "info" instead of "m->m_info[mds]". Also I reversed the if condition and pulled everything in one indent level. Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Reviewed-by: Alex Elder <elder@inktank.com>
2013-05-29 18:46:56 +07:00
for (j = 0; j < num_export_targets; j++)
info->export_targets[j] =
ceph_decode_32(&pexport_targets);
} else {
info->export_targets = NULL;
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
}
}
ceph: allow connecting to mds whose rank >= mdsmap::m_max_mds mdsmap::m_max_mds is the expected count of active mds. It's not the max rank of active mds. User can decrease mdsmap::m_max_mds, but does not stop mds whose rank >= mdsmap::m_max_mds. Signed-off-by: "Yan, Zheng" <zyan@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-03-28 16:04:13 +07:00
if (m->m_num_mds > m->m_max_mds) {
/* find max up mds */
for (i = m->m_num_mds; i >= m->m_max_mds; i--) {
if (i == 0 || m->m_info[i-1].state > 0)
break;
}
m->m_num_mds = i;
}
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
/* pg_pools */
ceph_decode_32_safe(p, end, n, bad);
m->m_num_data_pg_pools = n;
ceph: update support for PGID64, PGPOOL3, OSDENC protocol features Support (and require) the PGID64, PGPOOL3, and OSDENC protocol features. These have been present in ceph.git since v0.42, Feb 2012. Require these features to simplify support; nobody is running older userspace. Note that the new request and reply encoding is still not in place, so the new code is not yet functional. Signed-off-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@inktank.com>
2013-02-24 01:41:09 +07:00
m->m_data_pg_pools = kcalloc(n, sizeof(u64), GFP_NOFS);
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
if (!m->m_data_pg_pools)
ceph: check availability of mds cluster on mount Signed-off-by: Yan, Zheng <zyan@redhat.com>
2016-11-10 15:02:06 +07:00
goto nomem;
ceph: update support for PGID64, PGPOOL3, OSDENC protocol features Support (and require) the PGID64, PGPOOL3, and OSDENC protocol features. These have been present in ceph.git since v0.42, Feb 2012. Require these features to simplify support; nobody is running older userspace. Note that the new request and reply encoding is still not in place, so the new code is not yet functional. Signed-off-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@inktank.com>
2013-02-24 01:41:09 +07:00
ceph_decode_need(p, end, sizeof(u64)*(n+1), bad);
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
for (i = 0; i < n; i++)
ceph: update support for PGID64, PGPOOL3, OSDENC protocol features Support (and require) the PGID64, PGPOOL3, and OSDENC protocol features. These have been present in ceph.git since v0.42, Feb 2012. Require these features to simplify support; nobody is running older userspace. Note that the new request and reply encoding is still not in place, so the new code is not yet functional. Signed-off-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@inktank.com>
2013-02-24 01:41:09 +07:00
m->m_data_pg_pools[i] = ceph_decode_64(p);
m->m_cas_pg_pool = ceph_decode_64(p);
ceph: check availability of mds cluster on mount Signed-off-by: Yan, Zheng <zyan@redhat.com>
2016-11-10 15:02:06 +07:00
m->m_enabled = m->m_epoch > 1;
mdsmap_ev = 1;
if (mdsmap_v >= 2) {
ceph_decode_16_safe(p, end, mdsmap_ev, bad_ext);
}
if (mdsmap_ev >= 3) {
if (__decode_and_drop_compat_set(p, end) < 0)
goto bad_ext;
}
/* metadata_pool */
if (mdsmap_ev < 5) {
__decode_and_drop_type(p, end, u32, bad_ext);
} else {
__decode_and_drop_type(p, end, u64, bad_ext);
}
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
ceph: check availability of mds cluster on mount Signed-off-by: Yan, Zheng <zyan@redhat.com>
2016-11-10 15:02:06 +07:00
/* created + modified + tableserver */
__decode_and_drop_type(p, end, struct ceph_timespec, bad_ext);
__decode_and_drop_type(p, end, struct ceph_timespec, bad_ext);
__decode_and_drop_type(p, end, u32, bad_ext);
/* in */
{
int num_laggy = 0;
ceph_decode_32_safe(p, end, n, bad_ext);
ceph_decode_need(p, end, sizeof(u32) * n, bad_ext);
for (i = 0; i < n; i++) {
s32 mds = ceph_decode_32(p);
ceph: allow connecting to mds whose rank >= mdsmap::m_max_mds mdsmap::m_max_mds is the expected count of active mds. It's not the max rank of active mds. User can decrease mdsmap::m_max_mds, but does not stop mds whose rank >= mdsmap::m_max_mds. Signed-off-by: "Yan, Zheng" <zyan@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-03-28 16:04:13 +07:00
if (mds >= 0 && mds < m->m_num_mds) {
ceph: check availability of mds cluster on mount Signed-off-by: Yan, Zheng <zyan@redhat.com>
2016-11-10 15:02:06 +07:00
if (m->m_info[mds].laggy)
num_laggy++;
}
}
m->m_num_laggy = num_laggy;
ceph: allow connecting to mds whose rank >= mdsmap::m_max_mds mdsmap::m_max_mds is the expected count of active mds. It's not the max rank of active mds. User can decrease mdsmap::m_max_mds, but does not stop mds whose rank >= mdsmap::m_max_mds. Signed-off-by: "Yan, Zheng" <zyan@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-03-28 16:04:13 +07:00
if (n > m->m_num_mds) {
void *new_m_info = krealloc(m->m_info,
n * sizeof(*m->m_info),
GFP_NOFS | __GFP_ZERO);
if (!new_m_info)
goto nomem;
m->m_info = new_m_info;
}
m->m_num_mds = n;
ceph: check availability of mds cluster on mount Signed-off-by: Yan, Zheng <zyan@redhat.com>
2016-11-10 15:02:06 +07:00
}
/* inc */
__decode_and_drop_map(p, end, u32, u32, bad_ext);
/* up */
__decode_and_drop_map(p, end, u32, u64, bad_ext);
/* failed */
__decode_and_drop_set(p, end, u32, bad_ext);
/* stopped */
__decode_and_drop_set(p, end, u32, bad_ext);
if (mdsmap_ev >= 4) {
/* last_failure_osd_epoch */
__decode_and_drop_type(p, end, u32, bad_ext);
}
if (mdsmap_ev >= 6) {
/* ever_allowed_snaps */
__decode_and_drop_type(p, end, u8, bad_ext);
/* explicitly_allowed_snaps */
__decode_and_drop_type(p, end, u8, bad_ext);
}
if (mdsmap_ev >= 7) {
/* inline_data_enabled */
__decode_and_drop_type(p, end, u8, bad_ext);
}
if (mdsmap_ev >= 8) {
u32 name_len;
/* enabled */
ceph_decode_8_safe(p, end, m->m_enabled, bad_ext);
ceph_decode_32_safe(p, end, name_len, bad_ext);
ceph_decode_need(p, end, name_len, bad_ext);
*p += name_len;
}
/* damaged */
if (mdsmap_ev >= 9) {
size_t need;
ceph_decode_32_safe(p, end, n, bad_ext);
need = sizeof(u32) * n;
ceph_decode_need(p, end, need, bad_ext);
*p += need;
m->m_damaged = n > 0;
} else {
m->m_damaged = false;
}
bad_ext:
ceph: CEPH_FEATURE_MDSENC support Signed-off-by: Yan, Zheng <zyan@redhat.com>
2016-03-31 14:53:01 +07:00
*p = end;
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
dout("mdsmap_decode success epoch %u\n", m->m_epoch);
return m;
ceph: check availability of mds cluster on mount Signed-off-by: Yan, Zheng <zyan@redhat.com>
2016-11-10 15:02:06 +07:00
nomem:
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
err = -ENOMEM;
ceph: check availability of mds cluster on mount Signed-off-by: Yan, Zheng <zyan@redhat.com>
2016-11-10 15:02:06 +07:00
goto out_err;
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
bad:
pr_err("corrupt mdsmap\n");
ceph: hex dump corrupt server data to KERN_DEBUG Also, print fsid using standard format, NOT hex dump. Signed-off-by: Sage Weil <sage@newdream.net>
2009-12-15 06:13:47 +07:00
print_hex_dump(KERN_DEBUG, "mdsmap: ",
DUMP_PREFIX_OFFSET, 16, 1,
start, end - start, true);
ceph: check availability of mds cluster on mount Signed-off-by: Yan, Zheng <zyan@redhat.com>
2016-11-10 15:02:06 +07:00
out_err:
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
ceph_mdsmap_destroy(m);
ceph: improve error handling in ceph_mdsmap_decode This patch makes the following improvements to the error handling in the ceph_mdsmap_decode function: - Add a NULL check for return value from kcalloc - Make use of the variable err Signed-off-by: Emil Goode <emilgoode@gmail.com> Signed-off-by: Sage Weil <sage@inktank.com>
2013-05-28 21:59:00 +07:00
return ERR_PTR(err);
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
}
void ceph_mdsmap_destroy(struct ceph_mdsmap *m)
{
int i;
ceph: allow connecting to mds whose rank >= mdsmap::m_max_mds mdsmap::m_max_mds is the expected count of active mds. It's not the max rank of active mds. User can decrease mdsmap::m_max_mds, but does not stop mds whose rank >= mdsmap::m_max_mds. Signed-off-by: "Yan, Zheng" <zyan@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-03-28 16:04:13 +07:00
for (i = 0; i < m->m_num_mds; i++)
ceph: MDS client The MDS (metadata server) client is responsible for submitting requests to the MDS cluster and parsing the response. We decide which MDS to submit each request to based on cached information about the current partition of the directory hierarchy across the cluster. A stateful session is opened with each MDS before we submit requests to it, and a mutex is used to control the ordering of messages within each session. An MDS request may generate two responses. The first indicates the operation was a success and returns any result. A second reply is sent when the operation commits to disk. Note that locking on the MDS ensures that the results of updates are visible only to the updating client before the operation commits. Requests are linked to the containing directory so that an fsync will wait for them to commit. If an MDS fails and/or recovers, we resubmit requests as needed. We also reconnect existing capabilities to a recovering MDS to reestablish that shared session state. Old dentry leases are invalidated. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-07 01:31:09 +07:00
kfree(m->m_info[i].export_targets);
kfree(m->m_info);
kfree(m->m_data_pg_pools);
kfree(m);
}
ceph: check availability of mds cluster on mount Signed-off-by: Yan, Zheng <zyan@redhat.com>
2016-11-10 15:02:06 +07:00
bool ceph_mdsmap_is_cluster_available(struct ceph_mdsmap *m)
{
int i, nr_active = 0;
if (!m->m_enabled)
return false;
if (m->m_damaged)
return false;
if (m->m_num_laggy > 0)
return false;
ceph: allow connecting to mds whose rank >= mdsmap::m_max_mds mdsmap::m_max_mds is the expected count of active mds. It's not the max rank of active mds. User can decrease mdsmap::m_max_mds, but does not stop mds whose rank >= mdsmap::m_max_mds. Signed-off-by: "Yan, Zheng" <zyan@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-03-28 16:04:13 +07:00
for (i = 0; i < m->m_num_mds; i++) {
ceph: check availability of mds cluster on mount Signed-off-by: Yan, Zheng <zyan@redhat.com>
2016-11-10 15:02:06 +07:00
if (m->m_info[i].state == CEPH_MDS_STATE_ACTIVE)
nr_active++;
}
return nr_active > 0;
}
Reference in New Issue Copy Permalink