1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
|
// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*-
// vim: ts=8 sw=2 smarttab
/*
* Ceph - scalable distributed file system
*
* Copyright (C) 2004-2006 Sage Weil <sage@newdream.net>
*
* This is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License version 2.1, as published by the Free Software
* Foundation. See file COPYING.
*
*/
#ifndef CEPH_MBACKFILL_H
#define CEPH_MBACKFILL_H
#include "msg/Message.h"
#include "messages/MOSDPeeringOp.h"
#include "osd/PGPeeringEvent.h"
class MBackfillReserve : public MOSDPeeringOp {
private:
static constexpr int HEAD_VERSION = 5;
static constexpr int COMPAT_VERSION = 4;
public:
spg_t pgid;
epoch_t query_epoch;
enum {
REQUEST = 0, // primary->replica: please reserve a slot
GRANT = 1, // replica->primary: ok, i reserved it
REJECT_TOOFULL = 2, // replica->primary: too full, sorry, try again later (*)
RELEASE = 3, // primary->replcia: release the slot i reserved before
REVOKE_TOOFULL = 4, // replica->primary: too full, stop backfilling
REVOKE = 5, // replica->primary: i'm taking back the slot i gave you
// (*) NOTE: prior to luminous, REJECT was overloaded to also mean release
};
uint32_t type;
uint32_t priority;
int64_t primary_num_bytes;
int64_t shard_num_bytes;
spg_t get_spg() const {
return pgid;
}
epoch_t get_map_epoch() const {
return query_epoch;
}
epoch_t get_min_epoch() const {
return query_epoch;
}
PGPeeringEvent *get_event() override {
switch (type) {
case REQUEST:
return new PGPeeringEvent(
query_epoch,
query_epoch,
RequestBackfillPrio(priority, primary_num_bytes, shard_num_bytes));
case GRANT:
return new PGPeeringEvent(
query_epoch,
query_epoch,
RemoteBackfillReserved());
case REJECT_TOOFULL:
// NOTE: this is replica -> primary "i reject your request"
// and also primary -> replica "cancel my previously-granted request"
// (for older peers)
// and also replica -> primary "i revoke your reservation"
// (for older peers)
return new PGPeeringEvent(
query_epoch,
query_epoch,
RemoteReservationRejectedTooFull());
case RELEASE:
return new PGPeeringEvent(
query_epoch,
query_epoch,
RemoteReservationCanceled());
case REVOKE_TOOFULL:
return new PGPeeringEvent(
query_epoch,
query_epoch,
RemoteReservationRevokedTooFull());
case REVOKE:
return new PGPeeringEvent(
query_epoch,
query_epoch,
RemoteReservationRevoked());
default:
ceph_abort();
}
}
MBackfillReserve()
: MOSDPeeringOp{MSG_OSD_BACKFILL_RESERVE, HEAD_VERSION, COMPAT_VERSION},
query_epoch(0), type(-1), priority(-1), primary_num_bytes(0),
shard_num_bytes(0) {}
MBackfillReserve(int type,
spg_t pgid,
epoch_t query_epoch, unsigned prio = -1,
int64_t primary_num_bytes = 0,
int64_t shard_num_bytes = 0)
: MOSDPeeringOp{MSG_OSD_BACKFILL_RESERVE, HEAD_VERSION, COMPAT_VERSION},
pgid(pgid), query_epoch(query_epoch),
type(type), priority(prio), primary_num_bytes(primary_num_bytes),
shard_num_bytes(shard_num_bytes) {}
std::string_view get_type_name() const override {
return "MBackfillReserve";
}
void inner_print(std::ostream& out) const override {
switch (type) {
case REQUEST:
out << "REQUEST";
break;
case GRANT:
out << "GRANT";
break;
case REJECT_TOOFULL:
out << "REJECT_TOOFULL";
break;
case RELEASE:
out << "RELEASE";
break;
case REVOKE_TOOFULL:
out << "REVOKE_TOOFULL";
break;
case REVOKE:
out << "REVOKE";
break;
}
if (type == REQUEST) out << " prio: " << priority;
return;
}
void decode_payload() override {
auto p = payload.cbegin();
using ceph::decode;
decode(pgid.pgid, p);
decode(query_epoch, p);
decode(type, p);
decode(priority, p);
decode(pgid.shard, p);
if (header.version >= 5) {
decode(primary_num_bytes, p);
decode(shard_num_bytes, p);
} else {
primary_num_bytes = 0;
shard_num_bytes = 0;
}
}
void encode_payload(uint64_t features) override {
using ceph::encode;
if (!HAVE_FEATURE(features, RECOVERY_RESERVATION_2)) {
header.version = 3;
header.compat_version = 3;
encode(pgid.pgid, payload);
encode(query_epoch, payload);
encode((type == RELEASE || type == REVOKE_TOOFULL || type == REVOKE) ?
REJECT_TOOFULL : type, payload);
encode(priority, payload);
encode(pgid.shard, payload);
return;
}
header.version = HEAD_VERSION;
header.compat_version = COMPAT_VERSION;
encode(pgid.pgid, payload);
encode(query_epoch, payload);
encode(type, payload);
encode(priority, payload);
encode(pgid.shard, payload);
encode(primary_num_bytes, payload);
encode(shard_num_bytes, payload);
}
};
#endif
|
