Implement event-driven PG peering

blocking-uring-test
Vitaliy Filippov 2020-02-11 02:30:46 +03:00
parent 327f310868
commit a66b34e04d
8 changed files with 281 additions and 119 deletions

View File

@ -24,7 +24,7 @@ libblockstore.so: $(BLOCKSTORE_OBJS)
libfio_blockstore.so: ./libblockstore.so fio_engine.cpp json11.o
g++ $(CXXFLAGS) -shared -o libfio_blockstore.so fio_engine.cpp json11.o ./libblockstore.so -ltcmalloc_minimal -luring
OSD_OBJS := osd.o osd_exec_secondary.o osd_receive.o osd_send.o osd_peering.o osd_peering_pg.o osd_primary.o json11.o
OSD_OBJS := osd.o osd_exec_secondary.o osd_receive.o osd_send.o osd_peering.o osd_peering_pg.o osd_primary.o json11.o timerfd_interval.o
osd_exec_secondary.o: osd_exec_secondary.cpp osd.h osd_ops.h
g++ $(CXXFLAGS) -c -o $@ $<
osd_receive.o: osd_receive.cpp osd.h osd_ops.h

View File

@ -12,6 +12,7 @@ osd_t::osd_t(blockstore_config_t & config, blockstore_t *bs, ring_loop_t *ringlo
this->config = config;
this->bs = bs;
this->ringloop = ringloop;
this->tick_tfd = new timerfd_interval(ringloop, 1, []() {});
this->bs_block_size = bs->get_block_size();
// FIXME: use bitmap granularity instead
this->bs_disk_alignment = bs->get_disk_alignment();
@ -85,6 +86,7 @@ osd_t::osd_t(blockstore_config_t & config, blockstore_t *bs, ring_loop_t *ringlo
osd_t::~osd_t()
{
delete tick_tfd;
ringloop->unregister_consumer(consumer);
close(epoll_fd);
close(listen_fd);
@ -229,7 +231,10 @@ void osd_t::stop_client(int peer_fd)
auto it = clients.find(peer_fd);
if (it->second.osd_num)
{
// FIXME cancel outbound operations
osd_peer_fds.erase(it->second.osd_num);
repeer_pgs(it->second.osd_num, false);
peering_state |= OSD_PEERING_PEERS;
}
for (auto rit = read_ready_clients.begin(); rit != read_ready_clients.end(); rit++)
{

9
osd.h
View File

@ -14,6 +14,7 @@
#include "blockstore.h"
#include "ringloop.h"
#include "timerfd_interval.h"
#include "osd_ops.h"
#include "osd_peering_pg.h"
@ -34,6 +35,8 @@
#define PEER_CONNECTING 1
#define PEER_CONNECTED 2
#define OSD_PEERING_PEERS 1
#define OSD_PEERING_PGS 2
//#define OSD_STUB
@ -186,6 +189,7 @@ class osd_t
blockstore_t *bs;
uint32_t bs_block_size, bs_disk_alignment;
ring_loop_t *ringloop;
timerfd_interval *tick_tfd;
int wait_state = 0;
int epoll_fd = 0;
@ -217,6 +221,7 @@ class osd_t
osd_peer_def_t parse_peer(std::string peer);
void init_primary();
void handle_peers();
void repeer_pgs(osd_num_t osd_num, bool is_connected);
void start_pg_peering(int i);
// op execution
@ -235,8 +240,8 @@ class osd_t
void make_primary_reply(osd_op_t *op);
void finish_primary_op(osd_op_t *cur_op, int retval);
void handle_primary_read_subop(osd_op_t *cur_op, int ok);
int extend_missing_stripes(osd_read_stripe_t *stripes, osd_num_t *target_set, int minsize, int size);
void submit_read_subops(int read_pg_size, const uint64_t* target_set, osd_op_t *cur_op);
int extend_missing_stripes(osd_read_stripe_t *stripes, osd_num_t *osd_set, int minsize, int size);
void submit_read_subops(int read_pg_size, const uint64_t* osd_set, osd_op_t *cur_op);
public:
osd_t(blockstore_config_t & config, blockstore_t *bs, ring_loop_t *ringloop);
~osd_t();

View File

@ -17,12 +17,13 @@ void osd_t::init_primary()
throw std::runtime_error("peer1 and peer2 osd numbers are the same");
pgs.push_back((pg_t){
.state = PG_OFFLINE,
.pg_cursize = 2, // or 3
.pg_cursize = 0,
.pg_num = 1,
.target_set = { 1, 0, 3 }, // or { 1, 2, 3 }
.target_set = { 1, 2, 3 },
.cur_set = { 1, 0, 0 },
});
pg_count = 1;
peering_state = 1;
peering_state = OSD_PEERING_PEERS;
}
osd_peer_def_t osd_t::parse_peer(std::string peer)
@ -122,19 +123,19 @@ void osd_t::handle_connect_result(int peer_fd)
}
// Peering loop
// Ideally: Connect -> Ask & check config -> Start PG peering
void osd_t::handle_peers()
{
if (peering_state & 1)
if (peering_state & OSD_PEERING_PEERS)
{
for (int i = 0; i < peers.size(); i++)
{
if (osd_peer_fds.find(peers[i].osd_num) == osd_peer_fds.end() &&
time(NULL) - peers[i].last_connect_attempt > 5)
time(NULL) - peers[i].last_connect_attempt > 5) // FIXME hardcode 5
{
peers[i].last_connect_attempt = time(NULL);
connect_peer(peers[i].osd_num, peers[i].addr.c_str(), peers[i].port, [this](osd_num_t osd_num, int peer_fd)
{
// FIXME: Check peer config after connecting
if (peer_fd < 0)
{
printf("Failed to connect to peer OSD %lu: %s\n", osd_num, strerror(-peer_fd));
@ -144,83 +145,202 @@ void osd_t::handle_peers()
int i;
for (i = 0; i < peers.size(); i++)
{
auto it = osd_peer_fds.find(peers[i].osd_num);
if (it == osd_peer_fds.end() || clients[it->second].peer_state != PEER_CONNECTED)
{
if (osd_peer_fds.find(peers[i].osd_num) == osd_peer_fds.end())
break;
}
}
if (i >= peers.size())
{
// Start PG peering
pgs[0].state = PG_PEERING;
pgs[0].state_dict.clear();
pgs[0].obj_states.clear();
pgs[0].ver_override.clear();
if (pgs[0].peering_state)
delete pgs[0].peering_state;
peering_state = 2;
ringloop->wakeup();
// Connected to all peers
peering_state = peering_state & ~OSD_PEERING_PEERS;
}
repeer_pgs(osd_num, true);
});
}
}
}
if (peering_state & 2)
if (peering_state & OSD_PEERING_PGS)
{
bool still_doing_pgs = false;
for (int i = 0; i < pgs.size(); i++)
{
if (pgs[i].state == PG_PEERING)
{
if (!pgs[i].peering_state)
{
start_pg_peering(i);
}
else if (pgs[i].peering_state->list_done >= 3)
if (!pgs[i].peering_state->list_ops.size())
{
pgs[i].calc_object_states();
peering_state = 0;
}
else
{
still_doing_pgs = true;
}
}
}
if (!still_doing_pgs)
{
// Done all PGs
peering_state = peering_state & ~OSD_PEERING_PGS;
}
}
}
void osd_t::repeer_pgs(osd_num_t osd_num, bool is_connected)
{
// Re-peer affected PGs
// FIXME: We shouldn't rely just on target_set. Other OSDs may also contain PG data.
osd_num_t real_osd = (is_connected ? osd_num : 0);
for (int i = 0; i < pgs.size(); i++)
{
bool repeer = false;
for (int r = 0; r < pgs[i].target_set.size(); r++)
{
if (pgs[i].target_set[r] == osd_num &&
pgs[i].cur_set[r] != real_osd)
{
pgs[i].cur_set[r] = real_osd;
repeer = true;
break;
}
}
if (repeer)
{
// Repeer this pg
printf("Repeer PG %d because of OSD %lu\n", i, osd_num);
start_pg_peering(i);
peering_state |= OSD_PEERING_PGS;
}
}
}
// Repeer on each connect/disconnect peer event
void osd_t::start_pg_peering(int pg_idx)
{
// FIXME: Set PG_INCOMPLETE if incomplete
auto & pg = pgs[pg_idx];
auto ps = pg.peering_state = new pg_peering_state_t();
pg.state = PG_PEERING;
pg.state_dict.clear();
pg.obj_states.clear();
pg.ver_override.clear();
pg.pg_cursize = 0;
for (int role = 0; role < pg.cur_set.size(); role++)
{
osd_num_t osd_num = this->osd_num;
if (pg.cur_set[role] != 0)
{
pg.pg_cursize++;
}
}
if (pg.pg_cursize < pg.pg_minsize)
{
pg.state = PG_INCOMPLETE;
}
if (pg.peering_state)
{
// Adjust the peering operation that's still in progress
for (auto & p: pg.peering_state->list_ops)
{
int role;
for (role = 0; role < pg.cur_set.size(); role++)
{
if (pg.cur_set[role] == p.first)
break;
}
if (pg.state == PG_INCOMPLETE || role >= pg.cur_set.size())
{
// Discard the result after completion, which, chances are, will be unsuccessful
auto list_op = p.second;
if (list_op->peer_fd == 0)
{
// Self
list_op->bs_op.callback = [list_op](blockstore_op_t *bs_op)
{
if (list_op->bs_op.buf)
free(list_op->bs_op.buf);
delete list_op;
};
}
else
{
// Peer
list_op->callback = [](osd_op_t *list_op)
{
delete list_op;
};
}
pg.peering_state->list_ops.erase(p.first);
}
}
for (auto & p: pg.peering_state->list_results)
{
int role;
for (role = 0; role < pg.cur_set.size(); role++)
{
if (pg.cur_set[role] == p.first)
break;
}
if (pg.state == PG_INCOMPLETE || role >= pg.cur_set.size())
{
pg.peering_state->list_results.erase(p.first);
}
}
}
if (pg.state == PG_INCOMPLETE)
{
if (pg.peering_state)
{
delete pg.peering_state;
pg.peering_state = NULL;
}
printf("PG %d is incomplete\n", pg.pg_num);
return;
}
if (!pg.peering_state)
{
pg.peering_state = new pg_peering_state_t();
}
auto ps = pg.peering_state;
for (int role = 0; role < pg.cur_set.size(); role++)
{
osd_num_t role_osd = pg.cur_set[role];
if (!role_osd)
{
continue;
}
if (ps->list_ops.find(role_osd) != ps->list_ops.end() ||
ps->list_results.find(role_osd) != ps->list_results.end())
{
continue;
}
if (role_osd == this->osd_num)
{
// Self
osd_op_t *op = new osd_op_t();
op->op_type = 0;
op->peer_fd = 0;
op->bs_op.opcode = BS_OP_LIST;
op->bs_op.callback = [ps, op, osd_num](blockstore_op_t *bs_op)
op->bs_op.callback = [ps, op, role_osd](blockstore_op_t *bs_op)
{
if (op->bs_op.retval < 0)
{
throw std::runtime_error("OP_LIST failed");
throw std::runtime_error("local OP_LIST failed");
}
printf(
"Got object list from OSD %lu (local): %d objects (%lu of them stable)\n",
osd_num, bs_op->retval, bs_op->version
role_osd, bs_op->retval, bs_op->version
);
ps->list_results[osd_num] = {
ps->list_results[role_osd] = {
.buf = (obj_ver_id*)op->bs_op.buf,
.total_count = (uint64_t)op->bs_op.retval,
.stable_count = op->bs_op.version,
};
ps->list_done++;
ps->list_ops.erase(role_osd);
delete op;
};
bs->enqueue_op(&op->bs_op);
ps->list_ops[role_osd] = op;
}
for (int i = 0; i < peers.size(); i++)
else
{
osd_num_t osd_num = peers[i].osd_num;
auto & cl = clients[osd_peer_fds[peers[i].osd_num]];
// Peer
auto & cl = clients[osd_peer_fds[role_osd]];
osd_op_t *op = new osd_op_t();
op->op_type = OSD_OP_OUT;
op->peer_fd = cl.peer_fd;
@ -228,32 +348,42 @@ void osd_t::start_pg_peering(int pg_idx)
.sec_list = {
.header = {
.magic = SECONDARY_OSD_OP_MAGIC,
.id = 1,
.id = this->next_subop_id++,
.opcode = OSD_OP_SECONDARY_LIST,
},
.pgnum = pg.pg_num,
.pgtotal = pg_count,
},
};
op->callback = [ps, osd_num](osd_op_t *op)
op->callback = [this, ps, role_osd](osd_op_t *op)
{
if (op->reply.hdr.retval < 0)
{
throw std::runtime_error("OP_LIST failed");
int peer_fd = op->peer_fd;
printf("Failed to get object list from OSD %lu, disconnecting peer\n", role_osd);
delete op;
ps->list_ops.erase(role_osd);
stop_client(peer_fd);
return;
}
printf(
"Got object list from OSD %lu: %ld objects (%lu of them stable)\n",
osd_num, op->reply.hdr.retval, op->reply.sec_list.stable_count
role_osd, op->reply.hdr.retval, op->reply.sec_list.stable_count
);
ps->list_results[osd_num] = {
ps->list_results[role_osd] = {
.buf = (obj_ver_id*)op->buf,
.total_count = (uint64_t)op->reply.hdr.retval,
.stable_count = op->reply.sec_list.stable_count,
};
// so it doesn't get freed. FIXME: do it better
op->buf = NULL;
ps->list_done++;
ps->list_ops.erase(role_osd);
delete op;
};
outbox_push(cl, op);
ps->list_ops[role_osd] = op;
}
}
ringloop->wakeup();
}

View File

@ -18,7 +18,7 @@ void pg_t::remember_object(pg_obj_state_check_t &st, std::vector<obj_ver_role> &
else if (st.n_roles < pg.pg_minsize)
{
state = OBJ_INCOMPLETE;
pg.state = pg.state | PG_HAS_INCOMPLETE;
pg.state = pg.state | PG_HAS_UNFOUND;
}
else
{
@ -164,6 +164,7 @@ void pg_t::calc_object_states()
std::sort(all.begin(), all.end());
// Walk over it and check object states
pg.clean_count = 0;
pg.state = 0;
int replica = 0;
pg_obj_state_check_t st;
for (int i = 0; i < all.size(); i++)
@ -225,7 +226,7 @@ void pg_t::calc_object_states()
{
st.n_stable++;
}
if (pg.target_set[replica] == all[i].osd_num)
if (pg.cur_set[replica] == all[i].osd_num)
{
st.n_matched++;
}
@ -242,5 +243,17 @@ void pg_t::calc_object_states()
st.obj_end = st.ver_end = all.size();
remember_object(st, all);
}
if (pg.pg_cursize < pg.pg_size)
{
pg.state = pg.state | PG_DEGRADED;
}
printf(
"PG %u is active%s%s%s%s\n", pg.pg_num,
(pg.state & PG_DEGRADED) ? " + degraded" : "",
(pg.state & PG_HAS_UNFOUND) ? " + has_unfound" : "",
(pg.state & PG_HAS_DEGRADED) ? " + has_degraded" : "",
(pg.state & PG_HAS_MISPLACED) ? " + has_misplaced" : "",
(pg.state & PG_HAS_UNCLEAN) ? " + has_unclean" : ""
);
pg.state = pg.state | PG_ACTIVE;
}

View File

@ -14,10 +14,11 @@
#define PG_INCOMPLETE (1<<2)
#define PG_ACTIVE (1<<3)
// Plus any of these:
#define PG_HAS_INCOMPLETE (1<<4)
#define PG_HAS_DEGRADED (1<<5)
#define PG_HAS_MISPLACED (1<<6)
#define PG_HAS_UNCLEAN (1<<7)
#define PG_DEGRADED (1<<4)
#define PG_HAS_UNFOUND (1<<5)
#define PG_HAS_DEGRADED (1<<6)
#define PG_HAS_MISPLACED (1<<7)
#define PG_HAS_UNCLEAN (1<<8)
// OSD object states
#define OBJ_CLEAN 0x01
@ -40,7 +41,9 @@ typedef std::vector<pg_obj_loc_t> pg_osd_set_t;
struct pg_osd_set_state_t
{
// (role -> osd_num_t) map, as in pg.target_set and pg.cur_set
std::vector<osd_num_t> read_target;
// full OSD set including additional OSDs where the object is misplaced
pg_osd_set_t osd_set;
uint64_t state = 0;
uint64_t object_count = 0;
@ -53,9 +56,12 @@ struct pg_list_result_t
uint64_t stable_count;
};
struct osd_op_t;
struct pg_peering_state_t
{
// osd_num -> list result
spp::sparse_hash_map<osd_num_t, osd_op_t*> list_ops;
spp::sparse_hash_map<osd_num_t, pg_list_result_t> list_results;
int list_done = 0;
};
@ -103,9 +109,12 @@ struct pg_t
uint64_t pg_cursize = 3, pg_size = 3, pg_minsize = 2;
pg_num_t pg_num;
uint64_t clean_count = 0;
// target_set = (role => osd_num or UINT64_MAX if missing). role numbers start with zero
// target_set is the "correct" peer OSD set for this PG
std::vector<osd_num_t> target_set;
// moved object map. by default, each object is considered to reside on the target_set.
// cur_set is the current set of connected peer OSDs for this PG
// cur_set = (role => osd_num or UINT64_MAX if missing). role numbers begin with zero
std::vector<osd_num_t> cur_set;
// moved object map. by default, each object is considered to reside on the cur_set.
// this map stores all objects that differ.
// it may consume up to ~ (raw storage / object size) * 24 bytes in the worst case scenario
// which is up to ~192 MB per 1 TB in the worst case scenario

View File

@ -54,7 +54,9 @@ void osd_t::exec_primary_read(osd_op_t *cur_op)
pg_num_t pg_num = (oid % pg_count); // FIXME +1
if (((end - 1) / (bs_block_size*2)) != oid.stripe ||
(start % bs_disk_alignment) || (end % bs_disk_alignment) ||
pg_num > pgs.size())
pg_num > pgs.size() ||
// FIXME: Postpone operations in inactive PGs
!(pgs[pg_num].state & PG_ACTIVE))
{
finish_primary_op(cur_op, -EINVAL);
return;
@ -79,30 +81,28 @@ void osd_t::exec_primary_read(osd_op_t *cur_op)
auto vo_it = pgs[pg_num].ver_override.find(oid);
op_data->target_ver = vo_it != pgs[pg_num].ver_override.end() ? vo_it->second : UINT64_MAX;
}
if (pgs[pg_num].pg_cursize == pgs[pg_num].pg_size)
if (pgs[pg_num].state == PG_ACTIVE)
{
// Fast happy-path
submit_read_subops(pgs[pg_num].pg_minsize, pgs[pg_num].target_set.data(), cur_op);
submit_read_subops(pgs[pg_num].pg_minsize, pgs[pg_num].cur_set.data(), cur_op);
cur_op->send_list.push_back(cur_op->buf, cur_op->op.rw.len);
}
else
{
// PG is degraded
uint64_t* target_set;
{
auto it = pgs[pg_num].obj_states.find(oid);
target_set = (it != pgs[pg_num].obj_states.end()
? it->second->read_target.data()
: pgs[pg_num].target_set.data());
}
if (extend_missing_stripes(stripes, target_set, pgs[pg_num].pg_minsize, pgs[pg_num].pg_size) < 0)
// PG may be degraded or have misplaced objects
spp::sparse_hash_map<object_id, pg_osd_set_state_t*> obj_states;
auto st_it = pgs[pg_num].obj_states.find(oid);
uint64_t* cur_set = (st_it != pgs[pg_num].obj_states.end()
? st_it->second->read_target.data()
: pgs[pg_num].cur_set.data());
if (extend_missing_stripes(stripes, cur_set, pgs[pg_num].pg_minsize, pgs[pg_num].pg_size) < 0)
{
free(op_data);
finish_primary_op(cur_op, -EIO);
return;
}
// Submit reads
submit_read_subops(pgs[pg_num].pg_size, target_set, cur_op);
submit_read_subops(pgs[pg_num].pg_size, cur_set, cur_op);
op_data->pg_minsize = pgs[pg_num].pg_minsize;
op_data->pg_size = pgs[pg_num].pg_size;
op_data->degraded = 1;
@ -158,11 +158,11 @@ void osd_t::handle_primary_read_subop(osd_op_t *cur_op, int ok)
}
}
int osd_t::extend_missing_stripes(osd_read_stripe_t *stripes, osd_num_t *target_set, int minsize, int size)
int osd_t::extend_missing_stripes(osd_read_stripe_t *stripes, osd_num_t *osd_set, int minsize, int size)
{
for (int role = 0; role < minsize; role++)
{
if (stripes[role].end != 0 && target_set[role] == 0)
if (stripes[role].end != 0 && osd_set[role] == 0)
{
stripes[role].real_start = stripes[role].real_end = 0;
// Stripe is missing. Extend read to other stripes.
@ -170,7 +170,7 @@ int osd_t::extend_missing_stripes(osd_read_stripe_t *stripes, osd_num_t *target_
int exist = 0;
for (int j = 0; j < size; j++)
{
if (target_set[j] != 0)
if (osd_set[j] != 0)
{
if (stripes[j].real_end == 0 || j >= minsize)
{
@ -199,7 +199,7 @@ int osd_t::extend_missing_stripes(osd_read_stripe_t *stripes, osd_num_t *target_
return 0;
}
void osd_t::submit_read_subops(int read_pg_size, const uint64_t* target_set, osd_op_t *cur_op)
void osd_t::submit_read_subops(int read_pg_size, const uint64_t* osd_set, osd_op_t *cur_op)
{
osd_primary_read_t *op_data = (osd_primary_read_t*)cur_op->op_data;
osd_read_stripe_t *stripes = op_data->stripes;
@ -230,7 +230,7 @@ void osd_t::submit_read_subops(int read_pg_size, const uint64_t* target_set, osd
{
continue;
}
auto role_osd_num = target_set[role];
auto role_osd_num = osd_set[role];
if (role_osd_num != 0)
{
if (role_osd_num == this->osd_num)

View File

@ -344,9 +344,9 @@ int main05(int argc, char *argv[])
.state = PG_PEERING,
.pg_num = 1,
.target_set = { 1, 2, 3 },
.cur_set = { 1, 2, 3 },
.peering_state = new pg_peering_state_t(),
};
pg.peering_state->list_done = 3;
for (uint64_t osd_num = 1; osd_num <= 3; osd_num++)
{
pg_list_result_t r = {