数据库异常关闭时，数据库关闭时来不及或者没机会做checkpoint，则需要从上一个一致性检查的开始恢复。
PostgreSQL备机checkpoint是不能产生checkpoint WAL的，因为如果写这样类型的checkpoint的话，就会将接收的WAL打乱，那么日志将混乱，回放会出问题。
那么问题来了，备机支持checkpoint吗？他的checkpoint怎么做的？
PostgreSQL为了缩短恢复时间，备机上也支持checkpoint，即CreateRestartPoint。但是其pg_control文件的checkpoint记录的位点是从主机传过来WAL里面的checkpoint记录位置。
调用流程：
StartupXLOG do{ ... RmgrTable[record->xl_rmid].rm_redo(xlogreader); |-- RecoveryRestartPoint(&checkPoint); |-- lastCheckPointRecPtr = XLogCtl->lastCheckPointRecPtr; |-- ControlFile->checkPoint = lastCheckPointRecPtr; ... record = ReadRecord(xlogreader, InvalidXLogRecPtr, LOG, false); |-- record = XLogReadRecord(xlogreader, RecPtr, &errormsg); |-- ReadRecPtr = xlogreader->ReadRecPtr; } while (record != NULL); ShutdownXLOG->CreateRestartPoint: lastCheckPointRecPtr = XLogCtl->lastCheckPointRecPtr; ControlFile->checkPoint = lastCheckPointRecPtr;
1、备机回放
StartupXLOG do{ ... RmgrTable[record->xl_rmid].rm_redo(xlogreader);//回放 ... record = ReadRecord(xlogreader, InvalidXLogRecPtr, LOG, false);//读取一个xlog } while (record != NULL);
2、回放函数
void xlog_redo(XLogReaderState *record) { ... else if (info == XLOG_CHECKPOINT_SHUTDOWN){ ... memcpy(&checkPoint, XLogRecGetData(record), sizeof(CheckPoint)); ... RecoveryRestartPoint(&checkPoint); }else if (info == XLOG_CHECKPOINT_ONLINE){ ... memcpy(&checkPoint, XLogRecGetData(record), sizeof(CheckPoint)); ... RecoveryRestartPoint(&checkPoint); } ... }
3、RecoveryRestartPoint
static void RecoveryRestartPoint(const CheckPoint *checkPoint) { ... SpinLockAcquire(&XLogCtl->info_lck); XLogCtl->lastCheckPointRecPtr = ReadRecPtr;//ReadRecPtr为读取checkpoint记录后的位置 XLogCtl->lastCheckPointEndPtr = EndRecPtr; XLogCtl->lastCheckPoint = *checkPoint; SpinLockRelease(&XLogCtl->info_lck); }
4、ReadRecPtr赋值
ReadRecord for (;;) { char *errormsg; record = XLogReadRecord(xlogreader, RecPtr, &errormsg); ReadRecPtr = xlogreader->ReadRecPtr; EndRecPtr = xlogreader->EndRecPtr; ... }
5、备机createcheckpoint
bool CreateRestartPoint(int flags) { LWLockAcquire(CheckpointLock, LW_EXCLUSIVE); /* Get a local copy of the last safe checkpoint record. */ SpinLockAcquire(&XLogCtl->info_lck); lastCheckPointRecPtr = XLogCtl->lastCheckPointRecPtr;//checkpoint的位置来自XLogCtl->lastCheckPointRecPtr lastCheckPointEndPtr = XLogCtl->lastCheckPointEndPtr; lastCheckPoint = XLogCtl->lastCheckPoint; SpinLockRelease(&XLogCtl->info_lck); ... if (XLogRecPtrIsInvalid(lastCheckPointRecPtr) || lastCheckPoint.redo <= ControlFile->checkPointCopy.redo){ //回放了最后一个checkpoint记录后，备机再次手动执行checkpoint命令 UpdateMinRecoveryPoint(InvalidXLogRecPtr, true); if (flags & CHECKPOINT_IS_SHUTDOWN){ LWLockAcquire(ControlFileLock, LW_EXCLUSIVE); ControlFile->state = DB_SHUTDOWNED_IN_RECOVERY; ControlFile->time = (pg_time_t) time(NULL); UpdateControlFile(); LWLockRelease(ControlFileLock); } LWLockRelease(CheckpointLock); return false; } ... LWLockAcquire(ControlFileLock, LW_EXCLUSIVE); if (ControlFile->state == DB_IN_ARCHIVE_RECOVERY && ControlFile->checkPointCopy.redo < lastCheckPoint.redo){ ControlFile->prevCheckPoint = ControlFile->checkPoint; ControlFile->checkPoint = lastCheckPointRecPtr;//checkpoint的位置 ControlFile->checkPointCopy = lastCheckPoint; ControlFile->time = (pg_time_t) time(NULL); ... if (flags & CHECKPOINT_IS_SHUTDOWN) ControlFile->state = DB_SHUTDOWNED_IN_RECOVERY; UpdateControlFile(); } ... return true; }
6、备机shutdown
void ShutdownXLOG(int code, Datum arg) { /* * Signal walsenders to move to stopping state. */ WalSndInitStopping(); /* * Wait for WAL senders to be in stopping state. This prevents commands * from writing new WAL. */ WalSndWaitStopping(); if (RecoveryInProgress())//备机写checkpoint CreateRestartPoint(CHECKPOINT_IS_SHUTDOWN | CHECKPOINT_IMMEDIATE); else { /* * If archiving is enabled, rotate the last XLOG file so that all the * remaining records are archived (postmaster wakes up the archiver * process one more time at the end of shutdown). The checkpoint * record will go to the next XLOG file and won't be archived (yet). */ if (XLogArchivingActive() && XLogArchiveCommandSet()) RequestXLogSwitch(false); CreateCheckPoint(CHECKPOINT_IS_SHUTDOWN | CHECKPOINT_IMMEDIATE); } ShutdownCLOG(); ShutdownCommitTs(); ShutdownSUBTRANS(); Shutdown MultiXact(); }
7、总结
PostgreSQL备库也可以写检查点，目的是避免每次重启备库都需要从上一个检查点（由主库产生，在WAL中回放出来的）APPLY后面所有的WAL。但是他记录的checkpoint位点是从主库传过来的。这样的话就有问题了，如果主机很长时间都没有做checkpoint了，备机即使正常关闭，重启时，也会从上一个checkpoint开始恢复，这样也会恢复很长时间；并且多次重启也需要从上一次checkpoint开始重复恢复。
