*/
/*
* This file is part of Lustre, http://www.lustre.org/
- * Lustre is a trademark of Sun Microsystems, Inc.
*/
#define DEBUG_SUBSYSTEM S_LMV
return rc;
}
-struct obd_uuid *lmv_get_uuid(struct obd_export *exp)
+static struct obd_uuid *lmv_get_uuid(struct obd_export *exp)
{
struct lmv_obd *lmv = &exp->exp_obd->u.lmv;
struct lmv_tgt_desc *tgt = lmv_tgt(lmv, 0);
- return (tgt == NULL) ? NULL : obd_get_uuid(tgt->ltd_exp);
+ return tgt ? obd_get_uuid(tgt->ltd_exp) : NULL;
}
static int lmv_notify(struct obd_device *obd, struct obd_device *watched,
#define MAX_STRING_SIZE 128
-int lmv_connect_mdc(struct obd_device *obd, struct lmv_tgt_desc *tgt)
+static int lmv_connect_mdc(struct obd_device *obd, struct lmv_tgt_desc *tgt)
{
- struct lmv_obd *lmv = &obd->u.lmv;
- struct obd_device *mdc_obd;
- struct obd_export *mdc_exp;
- struct lu_fld_target target;
- int rc;
- ENTRY;
+ struct lmv_obd *lmv = &obd->u.lmv;
+ struct obd_device *mdc_obd;
+ struct obd_export *mdc_exp;
+ struct lu_fld_target target;
+ int rc;
+ ENTRY;
mdc_obd = class_find_client_obd(&tgt->ltd_uuid, LUSTRE_MDC_NAME,
&obd->obd_uuid);
rc = obd_connect(NULL, &mdc_exp, mdc_obd, &obd->obd_uuid,
&lmv->conn_data, lmv->lmv_cache);
- if (rc) {
- CERROR("target %s connect error %d\n", tgt->ltd_uuid.uuid, rc);
- RETURN(rc);
- }
+ if (rc) {
+ CERROR("target %s connect error %d\n", tgt->ltd_uuid.uuid, rc);
+ RETURN(rc);
+ }
/*
* Init fid sequence client for this mdc and add new fld target.
}
CDEBUG(D_CONFIG, "Connected to %s(%s) successfully (%d)\n",
- mdc_obd->obd_name, mdc_obd->obd_uuid.uuid,
- atomic_read(&obd->obd_refcount));
+ mdc_obd->obd_name, mdc_obd->obd_uuid.uuid,
+ atomic_read(&obd->obd_refcount));
lmv_statfs_check_update(obd, tgt);
static int lmv_disconnect_mdc(struct obd_device *obd, struct lmv_tgt_desc *tgt)
{
- struct lmv_obd *lmv = &obd->u.lmv;
- struct obd_device *mdc_obd;
- int rc;
- ENTRY;
+ struct lmv_obd *lmv = &obd->u.lmv;
+ struct obd_device *mdc_obd;
+ int rc;
+ ENTRY;
- LASSERT(tgt != NULL);
- LASSERT(obd != NULL);
+ LASSERT(tgt != NULL);
+ LASSERT(obd != NULL);
- mdc_obd = class_exp2obd(tgt->ltd_exp);
+ mdc_obd = class_exp2obd(tgt->ltd_exp);
- if (mdc_obd) {
- mdc_obd->obd_force = obd->obd_force;
- mdc_obd->obd_fail = obd->obd_fail;
- mdc_obd->obd_no_recov = obd->obd_no_recov;
+ if (mdc_obd) {
+ mdc_obd->obd_force = obd->obd_force;
+ mdc_obd->obd_fail = obd->obd_fail;
+ mdc_obd->obd_no_recov = obd->obd_no_recov;
if (lmv->lmv_tgts_kobj)
sysfs_remove_link(lmv->lmv_tgts_kobj,
rc = obd_fid_fini(tgt->ltd_exp->exp_obd);
if (rc)
- CERROR("Can't finanize fids factory\n");
-
- CDEBUG(D_INFO, "Disconnected from %s(%s) successfully\n",
- tgt->ltd_exp->exp_obd->obd_name,
- tgt->ltd_exp->exp_obd->obd_uuid.uuid);
-
- obd_register_observer(tgt->ltd_exp->exp_obd, NULL);
- rc = obd_disconnect(tgt->ltd_exp);
- if (rc) {
- if (tgt->ltd_active) {
- CERROR("Target %s disconnect error %d\n",
- tgt->ltd_uuid.uuid, rc);
- }
- }
+ CERROR("Can't finalize fids factory\n");
- lmv_activate_target(lmv, tgt, 0);
- tgt->ltd_exp = NULL;
- RETURN(0);
+ CDEBUG(D_INFO, "Disconnected from %s(%s) successfully\n",
+ tgt->ltd_exp->exp_obd->obd_name,
+ tgt->ltd_exp->exp_obd->obd_uuid.uuid);
+
+ obd_register_observer(tgt->ltd_exp->exp_obd, NULL);
+ rc = obd_disconnect(tgt->ltd_exp);
+ if (rc) {
+ if (tgt->ltd_active) {
+ CERROR("Target %s disconnect error %d\n",
+ tgt->ltd_uuid.uuid, rc);
+ }
+ }
+
+ lmv_activate_target(lmv, tgt, 0);
+ tgt->ltd_exp = NULL;
+ RETURN(0);
}
static int lmv_disconnect(struct obd_export *exp)
if (!tgt || !tgt->ltd_exp)
continue;
- rc = obd_statfs(env, tgt->ltd_exp, temp, max_age, flags);
+ rc = obd_statfs(env, tgt->ltd_exp, temp, max_age,
+ flags | OBD_STATFS_NESTED);
if (rc) {
CERROR("%s: can't stat MDS #%d: rc = %d\n",
tgt->ltd_exp->exp_obd->obd_name, i, rc);
GOTO(out_free_temp, rc);
}
- if (temp->os_state & OS_STATE_SUM ||
+ if (temp->os_state & OS_STATFS_SUM ||
flags == OBD_STATFS_FOR_MDT0) {
/* reset to the last aggregated values
* and don't sum with non-aggrated data */
tgt->ltd_statfs = *osfs;
tgt->ltd_statfs_age = ktime_get_seconds();
spin_unlock(&lmv->lmv_lock);
- lmv->lmv_qos.lq_dirty = 1;
+ set_bit(LQ_DIRTY, &lmv->lmv_qos.lq_flags);
}
return rc;
static struct lu_tgt_desc *lmv_locate_tgt_qos(struct lmv_obd *lmv, __u32 *mdt)
{
- struct lu_tgt_desc *tgt;
+ struct lu_tgt_desc *tgt, *cur = NULL;
+ __u64 total_avail = 0;
__u64 total_weight = 0;
__u64 cur_weight = 0;
+ int total_usable = 0;
__u64 rand;
int rc;
GOTO(unlock, tgt = ERR_PTR(rc));
lmv_foreach_tgt(lmv, tgt) {
- tgt->ltd_qos.ltq_usable = 0;
- if (!tgt->ltd_exp || !tgt->ltd_active)
+ if (!tgt->ltd_exp || !tgt->ltd_active) {
+ tgt->ltd_qos.ltq_usable = 0;
continue;
+ }
tgt->ltd_qos.ltq_usable = 1;
lu_tgt_qos_weight_calc(tgt);
+ if (tgt->ltd_index == *mdt)
+ cur = tgt;
+ total_avail += tgt->ltd_qos.ltq_avail;
total_weight += tgt->ltd_qos.ltq_weight;
+ total_usable++;
+ }
+
+ /* if current MDT has above-average space, within range of the QOS
+ * threshold, stay on the same MDT to avoid creating needless remote
+ * MDT directories.
+ */
+ rand = total_avail * (256 - lmv->lmv_qos.lq_threshold_rr) /
+ (total_usable * 256);
+ if (cur && cur->ltd_qos.ltq_avail >= rand) {
+ tgt = cur;
+ GOTO(unlock, rc = 0);
}
rand = lu_prandom_u64_max(total_weight);
ENTRY;
- spin_lock(&lmv->lmv_qos.lq_rr.lqr_alloc);
+ spin_lock(&lmv->lmv_lock);
for (i = 0; i < lmv->lmv_mdt_descs.ltd_tgts_size; i++) {
index = (i + lmv->lmv_qos_rr_index) %
lmv->lmv_mdt_descs.ltd_tgts_size;
*mdt = tgt->ltd_index;
lmv->lmv_qos_rr_index = (*mdt + 1) %
lmv->lmv_mdt_descs.ltd_tgts_size;
- spin_unlock(&lmv->lmv_qos.lq_rr.lqr_alloc);
+ spin_unlock(&lmv->lmv_lock);
RETURN(tgt);
}
- spin_unlock(&lmv->lmv_qos.lq_rr.lqr_alloc);
+ spin_unlock(&lmv->lmv_lock);
RETURN(ERR_PTR(-ENODEV));
}
+/* locate MDT by file name, for striped directory, the file name hash decides
+ * which stripe its dirent is stored.
+ */
static struct lmv_tgt_desc *
lmv_locate_tgt_by_name(struct lmv_obd *lmv, struct lmv_stripe_md *lsm,
const char *name, int namelen, struct lu_fid *fid,
*
* For plain direcotry, it just locate the MDT of op_data->op_fid1.
*
- * \param[in] lmv LMV device
- * \param[in] op_data client MD stack parameters, name, namelen
- * mds_num etc.
+ * \param[in] lmv LMV device
+ * \param[in/out] op_data client MD stack parameters, name, namelen etc,
+ * op_mds and op_fid1 will be updated if op_mea1
+ * indicates fid1 represents a striped directory.
*
* retval pointer to the lmv_tgt_desc if succeed.
* ERR_PTR(errno) if failed.
return rc;
}
+static inline bool lmv_op_user_qos_mkdir(const struct md_op_data *op_data)
+{
+ const struct lmv_user_md *lum = op_data->op_data;
+
+ return (op_data->op_cli_flags & CLI_SET_MEA) && lum &&
+ le32_to_cpu(lum->lum_magic) == LMV_USER_MAGIC &&
+ le32_to_cpu(lum->lum_stripe_offset) == LMV_OFFSET_DEFAULT;
+}
+
+static inline bool lmv_op_default_qos_mkdir(const struct md_op_data *op_data)
+{
+ const struct lmv_stripe_md *lsm = op_data->op_default_mea1;
+
+ return lsm && lsm->lsm_md_master_mdt_index == LMV_OFFSET_DEFAULT;
+}
+
/* mkdir by QoS in two cases:
* 1. 'lfs mkdir -i -1'
* 2. parent default LMV master_mdt_index is -1
*/
static inline bool lmv_op_qos_mkdir(const struct md_op_data *op_data)
{
- const struct lmv_stripe_md *lsm = op_data->op_default_mea1;
- const struct lmv_user_md *lum = op_data->op_data;
-
if (op_data->op_code != LUSTRE_OPC_MKDIR)
return false;
if (lmv_dir_striped(op_data->op_mea1))
return false;
- if (op_data->op_cli_flags & CLI_SET_MEA && lum &&
- (le32_to_cpu(lum->lum_magic) == LMV_USER_MAGIC ||
- le32_to_cpu(lum->lum_magic) == LMV_USER_MAGIC_SPECIFIC) &&
- le32_to_cpu(lum->lum_stripe_offset) == LMV_OFFSET_DEFAULT)
+ if (lmv_op_user_qos_mkdir(op_data))
return true;
- if (lsm && lsm->lsm_md_master_mdt_index == LMV_OFFSET_DEFAULT)
+ if (lmv_op_default_qos_mkdir(op_data))
return true;
return false;
}
+/* if default LMV is set, and its index is LMV_OFFSET_DEFAULT, and
+ * 1. max_inherit_rr is set and is not LMV_INHERIT_RR_NONE
+ * 2. or parent is ROOT
+ * mkdir roundrobin.
+ * NB, this also needs to check server is balanced, which is checked by caller.
+ */
+static inline bool lmv_op_default_rr_mkdir(const struct md_op_data *op_data)
+{
+ const struct lmv_stripe_md *lsm = op_data->op_default_mea1;
+
+ if (!lmv_op_default_qos_mkdir(op_data))
+ return false;
+
+ return lsm->lsm_md_max_inherit_rr != LMV_INHERIT_RR_NONE ||
+ fid_is_root(&op_data->op_fid1);
+}
+
/* 'lfs mkdir -i <specific_MDT>' */
static inline bool lmv_op_user_specific_mkdir(const struct md_op_data *op_data)
{
op_data->op_default_mea1->lsm_md_master_mdt_index !=
LMV_OFFSET_DEFAULT;
}
+
int lmv_create(struct obd_export *exp, struct md_op_data *op_data,
const void *data, size_t datalen, umode_t mode, uid_t uid,
gid_t gid, cfs_cap_t cap_effective, __u64 rdev,
if (!tgt)
RETURN(-ENODEV);
} else if (lmv_op_qos_mkdir(op_data)) {
+ struct lmv_tgt_desc *tmp = tgt;
+
tgt = lmv_locate_tgt_qos(lmv, &op_data->op_mds);
- if (tgt == ERR_PTR(-EAGAIN))
- tgt = lmv_locate_tgt_rr(lmv, &op_data->op_mds);
+ if (tgt == ERR_PTR(-EAGAIN)) {
+ if (ltd_qos_is_balanced(&lmv->lmv_mdt_descs) &&
+ !lmv_op_default_rr_mkdir(op_data) &&
+ !lmv_op_user_qos_mkdir(op_data))
+ /* if it's not necessary, don't create remote
+ * directory.
+ */
+ tgt = tmp;
+ else
+ tgt = lmv_locate_tgt_rr(lmv, &op_data->op_mds);
+ }
if (IS_ERR(tgt))
RETURN(PTR_ERR(tgt));
+
/*
* only update statfs after QoS mkdir, this means the cached
* statfs may be stale, and current mkdir may not follow QoS
return rc;
}
-int lmv_read_page(struct obd_export *exp, struct md_op_data *op_data,
- struct md_callback *cb_op, __u64 offset,
- struct page **ppage)
+static int lmv_read_page(struct obd_export *exp, struct md_op_data *op_data,
+ struct md_callback *cb_op, __u64 offset,
+ struct page **ppage)
{
struct obd_device *obd = exp->exp_obd;
struct lmv_obd *lmv = &obd->u.lmv;
* \retval 0 on success
* \retval negative negated errno on failure
*/
-int lmv_set_info_async(const struct lu_env *env, struct obd_export *exp,
- __u32 keylen, void *key, __u32 vallen, void *val,
- struct ptlrpc_request_set *set)
+static int lmv_set_info_async(const struct lu_env *env, struct obd_export *exp,
+ __u32 keylen, void *key, __u32 vallen, void *val,
+ struct ptlrpc_request_set *set)
{
- struct lmv_tgt_desc *tgt = NULL;
- struct obd_device *obd;
- struct lmv_obd *lmv;
+ struct lmv_tgt_desc *tgt;
+ struct obd_device *obd;
+ struct lmv_obd *lmv;
int rc = 0;
ENTRY;
lsm->lsm_md_stripe_count = le32_to_cpu(lmu->lum_stripe_count);
lsm->lsm_md_master_mdt_index = le32_to_cpu(lmu->lum_stripe_offset);
lsm->lsm_md_hash_type = le32_to_cpu(lmu->lum_hash_type);
+ lsm->lsm_md_max_inherit = lmu->lum_max_inherit;
+ lsm->lsm_md_max_inherit_rr = lmu->lum_max_inherit_rr;
lsm->lsm_md_pool_name[LOV_MAXPOOLNAME] = 0;
return 0;
}
if (lmv_dir_striped(lsm)) {
- for (i = 0; i < lsm->lsm_md_stripe_count; i++) {
- if (lsm->lsm_md_oinfo[i].lmo_root)
- iput(lsm->lsm_md_oinfo[i].lmo_root);
- }
+ for (i = 0; i < lsm->lsm_md_stripe_count; i++)
+ iput(lsm->lsm_md_oinfo[i].lmo_root);
lsm_size = lmv_stripe_md_size(lsm->lsm_md_stripe_count);
} else {
lsm_size = lmv_stripe_md_size(0);
RETURN(rc);
}
-enum ldlm_mode lmv_lock_match(struct obd_export *exp, __u64 flags,
- const struct lu_fid *fid, enum ldlm_type type,
- union ldlm_policy_data *policy,
- enum ldlm_mode mode, struct lustre_handle *lockh)
+static enum ldlm_mode
+lmv_lock_match(struct obd_export *exp, __u64 flags,
+ const struct lu_fid *fid, enum ldlm_type type,
+ union ldlm_policy_data *policy,
+ enum ldlm_mode mode, struct lustre_handle *lockh)
{
struct obd_device *obd = exp->exp_obd;
struct lmv_obd *lmv = &obd->u.lmv;
RETURN(0);
}
-int lmv_get_lustre_md(struct obd_export *exp, struct ptlrpc_request *req,
- struct obd_export *dt_exp, struct obd_export *md_exp,
- struct lustre_md *md)
+static int
+lmv_get_lustre_md(struct obd_export *exp, struct req_capsule *pill,
+ struct obd_export *dt_exp, struct obd_export *md_exp,
+ struct lustre_md *md)
{
struct lmv_obd *lmv = &exp->exp_obd->u.lmv;
struct lmv_tgt_desc *tgt = lmv_tgt(lmv, 0);
if (!tgt || !tgt->ltd_exp)
return -EINVAL;
- return md_get_lustre_md(tgt->ltd_exp, req, dt_exp, md_exp, md);
+ return md_get_lustre_md(tgt->ltd_exp, pill, dt_exp, md_exp, md);
}
-int lmv_free_lustre_md(struct obd_export *exp, struct lustre_md *md)
+static int lmv_free_lustre_md(struct obd_export *exp, struct lustre_md *md)
{
struct obd_device *obd = exp->exp_obd;
struct lmv_obd *lmv = &obd->u.lmv;
RETURN(md_free_lustre_md(tgt->ltd_exp, md));
}
-int lmv_set_open_replay_data(struct obd_export *exp,
- struct obd_client_handle *och,
- struct lookup_intent *it)
+static int lmv_set_open_replay_data(struct obd_export *exp,
+ struct obd_client_handle *och,
+ struct lookup_intent *it)
{
struct obd_device *obd = exp->exp_obd;
struct lmv_obd *lmv = &obd->u.lmv;
RETURN(md_set_open_replay_data(tgt->ltd_exp, och, it));
}
-int lmv_clear_open_replay_data(struct obd_export *exp,
- struct obd_client_handle *och)
+static int lmv_clear_open_replay_data(struct obd_export *exp,
+ struct obd_client_handle *och)
{
struct obd_device *obd = exp->exp_obd;
struct lmv_obd *lmv = &obd->u.lmv;
RETURN(md_clear_open_replay_data(tgt->ltd_exp, och));
}
-int lmv_intent_getattr_async(struct obd_export *exp,
- struct md_enqueue_info *minfo)
+static int lmv_intent_getattr_async(struct obd_export *exp,
+ struct md_op_item *item)
{
- struct md_op_data *op_data = &minfo->mi_data;
+ struct md_op_data *op_data = &item->mop_data;
struct obd_device *obd = exp->exp_obd;
struct lmv_obd *lmv = &obd->u.lmv;
struct lmv_tgt_desc *ptgt;
if (ctgt != ptgt)
RETURN(-EREMOTE);
- rc = md_intent_getattr_async(ptgt->ltd_exp, minfo);
+ rc = md_intent_getattr_async(ptgt->ltd_exp, item);
RETURN(rc);
}
-int lmv_revalidate_lock(struct obd_export *exp, struct lookup_intent *it,
- struct lu_fid *fid, __u64 *bits)
+static int lmv_revalidate_lock(struct obd_export *exp, struct lookup_intent *it,
+ struct lu_fid *fid, __u64 *bits)
{
struct obd_device *obd = exp->exp_obd;
struct lmv_obd *lmv = &obd->u.lmv;
RETURN(rc);
}
-int lmv_get_fid_from_lsm(struct obd_export *exp,
- const struct lmv_stripe_md *lsm,
- const char *name, int namelen, struct lu_fid *fid)
+static int lmv_get_fid_from_lsm(struct obd_export *exp,
+ const struct lmv_stripe_md *lsm,
+ const char *name, int namelen,
+ struct lu_fid *fid)
{
const struct lmv_oinfo *oinfo;
* process with other slave MDTs. The only exception is Q_GETOQUOTA for which
* we directly fetch data from the slave MDTs.
*/
-int lmv_quotactl(struct obd_device *unused, struct obd_export *exp,
- struct obd_quotactl *oqctl)
+static int lmv_quotactl(struct obd_device *unused, struct obd_export *exp,
+ struct obd_quotactl *oqctl)
{
struct obd_device *obd = class_exp2obd(exp);
struct lmv_obd *lmv = &obd->u.lmv;
static int __init lmv_init(void)
{
- return class_register_type(&lmv_obd_ops, &lmv_md_ops, true, NULL,
+ return class_register_type(&lmv_obd_ops, &lmv_md_ops, true,
LUSTRE_LMV_NAME, NULL);
}