1 /*-------------------------------------------------------------------------
4 * The system control file "pg_control" is not a heap relation.
5 * However, we define it here so that the format is documented.
8 * Portions Copyright (c) 1996-2017, PostgreSQL Global Development Group
9 * Portions Copyright (c) 1994, Regents of the University of California
11 * src/include/catalog/pg_control.h
13 *-------------------------------------------------------------------------
18 #include "access/xlogdefs.h"
19 #include "pgtime.h" /* for pg_time_t */
20 #include "port/pg_crc32c.h"
23 #define MOCK_AUTH_NONCE_LEN 32
25 /* Version identifier for this pg_control format */
26 #define PG_CONTROL_VERSION 1002
29 * Body of CheckPoint XLOG records. This is declared here because we keep
30 * a copy of the latest one in pg_control for possible disaster recovery.
31 * Changing this struct requires a PG_CONTROL_VERSION bump.
33 typedef struct CheckPoint
35 XLogRecPtr redo; /* next RecPtr available when we began to
36 * create CheckPoint (i.e. REDO start point) */
37 TimeLineID ThisTimeLineID; /* current TLI */
38 TimeLineID PrevTimeLineID; /* previous TLI, if this record begins a new
39 * timeline (equals ThisTimeLineID otherwise) */
40 bool fullPageWrites; /* current full_page_writes */
41 uint32 nextXidEpoch; /* higher-order bits of nextXid */
42 TransactionId nextXid; /* next free XID */
43 Oid nextOid; /* next free OID */
44 MultiXactId nextMulti; /* next free MultiXactId */
45 MultiXactOffset nextMultiOffset; /* next free MultiXact offset */
46 TransactionId oldestXid; /* cluster-wide minimum datfrozenxid */
47 Oid oldestXidDB; /* database with minimum datfrozenxid */
48 MultiXactId oldestMulti; /* cluster-wide minimum datminmxid */
49 Oid oldestMultiDB; /* database with minimum datminmxid */
50 pg_time_t time; /* time stamp of checkpoint */
51 TransactionId oldestCommitTsXid; /* oldest Xid with valid commit
53 TransactionId newestCommitTsXid; /* newest Xid with valid commit
57 * Oldest XID still running. This is only needed to initialize hot standby
58 * mode from an online checkpoint, so we only bother calculating this for
59 * online checkpoints and only when wal_level is replica. Otherwise it's
60 * set to InvalidTransactionId.
62 TransactionId oldestActiveXid;
65 /* XLOG info values for XLOG rmgr */
66 #define XLOG_CHECKPOINT_SHUTDOWN 0x00
67 #define XLOG_CHECKPOINT_ONLINE 0x10
68 #define XLOG_NOOP 0x20
69 #define XLOG_NEXTOID 0x30
70 #define XLOG_SWITCH 0x40
71 #define XLOG_BACKUP_END 0x50
72 #define XLOG_PARAMETER_CHANGE 0x60
73 #define XLOG_RESTORE_POINT 0x70
74 #define XLOG_FPW_CHANGE 0x80
75 #define XLOG_END_OF_RECOVERY 0x90
76 #define XLOG_FPI_FOR_HINT 0xA0
81 * System status indicator. Note this is stored in pg_control; if you change
82 * it, you must bump PG_CONTROL_VERSION
88 DB_SHUTDOWNED_IN_RECOVERY,
91 DB_IN_ARCHIVE_RECOVERY,
96 * Contents of pg_control.
98 * NOTE: try to keep this under 512 bytes so that it will fit on one physical
99 * sector of typical disk drives. This reduces the odds of corruption due to
100 * power failure midway through a write.
103 typedef struct ControlFileData
106 * Unique system identifier --- to ensure we match up xlog files with the
107 * installation that produced them.
109 uint64 system_identifier;
112 * Version identifier information. Keep these fields at the same offset,
113 * especially pg_control_version; they won't be real useful if they move
114 * around. (For historical reasons they must be 8 bytes into the file
115 * rather than immediately at the front.)
117 * pg_control_version identifies the format of pg_control itself.
118 * catalog_version_no identifies the format of the system catalogs.
120 * There are additional version identifiers in individual files; for
121 * example, WAL logs contain per-page magic numbers that can serve as
122 * version cues for the WAL log.
124 uint32 pg_control_version; /* PG_CONTROL_VERSION */
125 uint32 catalog_version_no; /* see catversion.h */
130 DBState state; /* see enum above */
131 pg_time_t time; /* time stamp of last pg_control update */
132 XLogRecPtr checkPoint; /* last check point record ptr */
133 XLogRecPtr prevCheckPoint; /* previous check point record ptr */
135 CheckPoint checkPointCopy; /* copy of last check point record */
137 XLogRecPtr unloggedLSN; /* current fake LSN value, for unlogged rels */
140 * These two values determine the minimum point we must recover up to
141 * before starting up:
143 * minRecoveryPoint is updated to the latest replayed LSN whenever we
144 * flush a data change during archive recovery. That guards against
145 * starting archive recovery, aborting it, and restarting with an earlier
146 * stop location. If we've already flushed data changes from WAL record X
147 * to disk, we mustn't start up until we reach X again. Zero when not
148 * doing archive recovery.
150 * backupStartPoint is the redo pointer of the backup start checkpoint, if
151 * we are recovering from an online backup and haven't reached the end of
152 * backup yet. It is reset to zero when the end of backup is reached, and
153 * we mustn't start up before that. A boolean would suffice otherwise, but
154 * we use the redo pointer as a cross-check when we see an end-of-backup
155 * record, to make sure the end-of-backup record corresponds the base
156 * backup we're recovering from.
158 * backupEndPoint is the backup end location, if we are recovering from an
159 * online backup which was taken from the standby and haven't reached the
160 * end of backup yet. It is initialized to the minimum recovery point in
161 * pg_control which was backed up last. It is reset to zero when the end
162 * of backup is reached, and we mustn't start up before that.
164 * If backupEndRequired is true, we know for sure that we're restoring
165 * from a backup, and must see a backup-end record before we can safely
166 * start up. If it's false, but backupStartPoint is set, a backup_label
167 * file was found at startup but it may have been a leftover from a stray
168 * pg_start_backup() call, not accompanied by pg_stop_backup().
170 XLogRecPtr minRecoveryPoint;
171 TimeLineID minRecoveryPointTLI;
172 XLogRecPtr backupStartPoint;
173 XLogRecPtr backupEndPoint;
174 bool backupEndRequired;
177 * Parameter settings that determine if the WAL can be used for archival
183 int max_worker_processes;
184 int max_prepared_xacts;
185 int max_locks_per_xact;
186 bool track_commit_timestamp;
189 * This data is used to check for hardware-architecture compatibility of
190 * the database and the backend executable. We need not check endianness
191 * explicitly, since the pg_control version will surely look wrong to a
192 * machine of different endianness, but we do need to worry about MAXALIGN
193 * and floating-point format. (Note: storage layout nominally also
194 * depends on SHORTALIGN and INTALIGN, but in practice these are the same
195 * on all architectures of interest.)
197 * Testing just one double value is not a very bulletproof test for
198 * floating-point compatibility, but it will catch most cases.
200 uint32 maxAlign; /* alignment requirement for tuples */
201 double floatFormat; /* constant 1234567.0 */
202 #define FLOATFORMAT_VALUE 1234567.0
205 * This data is used to make sure that configuration of this database is
206 * compatible with the backend executable.
208 uint32 blcksz; /* data block size for this DB */
209 uint32 relseg_size; /* blocks per segment of large relation */
211 uint32 xlog_blcksz; /* block size within WAL files */
212 uint32 xlog_seg_size; /* size of each WAL segment */
214 uint32 nameDataLen; /* catalog name field width */
215 uint32 indexMaxKeys; /* max number of columns in an index */
217 uint32 toast_max_chunk_size; /* chunk size in TOAST tables */
218 uint32 loblksize; /* chunk size in pg_largeobject */
220 /* flags indicating pass-by-value status of various types */
221 bool float4ByVal; /* float4 pass-by-value? */
222 bool float8ByVal; /* float8, int8, etc pass-by-value? */
224 /* Are data pages protected by checksums? Zero if no checksum version */
225 uint32 data_checksum_version;
228 * Random nonce, used in authentication requests that need to proceed
229 * based on values that are cluster-unique, like a SASL exchange that
230 * failed at an early stage.
232 char mock_authentication_nonce[MOCK_AUTH_NONCE_LEN];
234 /* CRC of all above ... MUST BE LAST! */
239 * Physical size of the pg_control file. Note that this is considerably
240 * bigger than the actually used size (ie, sizeof(ControlFileData)).
241 * The idea is to keep the physical size constant independent of format
242 * changes, so that ReadControlFile will deliver a suitable wrong-version
243 * message instead of a read error if it's looking at an incompatible file.
245 #define PG_CONTROL_SIZE 8192
247 #endif /* PG_CONTROL_H */