PostgreSQL启动恢复读取checkpoint记录失败的条件

1、首先读取ControlFile->checkPoint指向的checkpoint

2、如果读取失败，slave直接abort退出，master再次读取ControlFile->prevCheckPoint指向的checkpoint

StartupXLOG->

|--checkPointLoc = ControlFile->checkPoint;

|--record = ReadCheckpointRecord(xlogreader, checkPointLoc, 1, true):

|-- if (record != NULL){

...

}else if (StandbyMode){

ereport(PANIC,(errmsg("could not locate a valid checkpoint record")));

}else{

checkPointLoc = ControlFile->prevCheckPoint;

record = ReadCheckpointRecord(xlogreader, checkPointLoc, 2, true);

if (record != NULL){

InRecovery = true;//标记下面进入recovery

}else{

ereport(PANIC,(errmsg("could not locate a valid checkpoint record")));

}

}

一、那么什么条件下读取的checkpoint记录record==NULL?

1、ControlFile->checkPoint % XLOG_BLCKSZ < SizeOfXLogShortPHD

2、ReadRecord(xlogreader, ControlFile->checkPoint, LOG, true)返回NULL

3、ReadRecord读到的record!=NULL && record->xl_rmid != RM_XLOG_ID

4、ReadRecord读到的record!=NULL && info != XLOG_CHECKPOINT_SHUTDOWN && info != XLOG_CHECKPOINT_ONLINE

5、ReadRecord读到的record!=NULL && record->xl_tot_len != SizeOfXLogRecord + SizeOfXLogRecordDataHeaderShort + sizeof(CheckPoint)

二、ReadRecord函数返回NULL的条件

ReadRecord(xlogreader, ControlFile->checkPoint, LOG, true)

|--record = XLogReadRecord(xlogreader, ControlFile->checkPoint, &errormsg);

|-- 2.1 record==NULL && ！StandbyMode

|-- 2.2 record!=NULL && !tliInHistory(xlogreader->latestPageTLI, expectedTLEs)

/*-----

note:只要读取了一页xlog,就会赋值为该页第一个记录的时间线

XLogReaderValidatePageHeader

-->xlogreader->latestPageTLI=hdr->xlp_tli;

------*/

三、XlogReadRecord读取checkpoint返回NULL的条件？

XLogReadRecord(xlogreader, ControlFile->checkPoint, &errormsg)

targetPagePtr = ControlFile->checkPoint - (ControlFile->checkPoint % XLOG_BLCKSZ);

targetRecOff = ControlFile->checkPoint % XLOG_BLCKSZ;

readOff = ReadPageInternal(state,targetPagePtr, Min(targetRecOff + SizeOfXLogRecord, XLOG_BLCKSZ));

pageHeaderSize = XLogPageHeaderSize((XLogPageHeader) state->readBuf);

record = (XLogRecord *) (state->readBuf + RecPtr % XLOG_BLCKSZ);

total_len = record->xl_tot_len;

-------------

1、readOff < 0

2、0< targetRecOff < pageHeaderSize

3、(((XLogPageHeader) state->readBuf)->xlp_info & XLP_FIRST_IS_CONTRECORD) && targetRecOff == pageHeaderSize

page头有跨页的record并且checkpoint定位的偏移正好在页头尾部

4、targetRecOff <= XLOG_BLCKSZ - SizeOfXLogRecord &&

!ValidXLogRecordHeader(state, ControlFile->checkPoint, state->ReadRecPtr, record,randAccess)

---（record->xl_tot_len < SizeOfXLogRecord || record->xl_rmid > RM_MAX_ID || record->xl_prev != state->ReadRecPtr）

5、targetRecOff > XLOG_BLCKSZ - SizeOfXLogRecord && total_len < SizeOfXLogRecord

6、total_len > state->readRecordBufSize && !allocate_recordbuf(state, total_len)

一旦该记录损坏，total_len的长度非常大的话，就需要allocate_recordbuf扩展state->readbuf，可能因此分配失败abort

记录的checksum需要等待全部读取完整记录后才校验

-------------

三、ReadPageInternal返回的readOff返回小于0的条件

ReadPageInternal(state,targetPagePtr, Min(targetRecOff + SizeOfXLogRecord, XLOG_BLCKSZ))

1、第一次read wal文件，readLen = state->read_page：读取第一页。readLen < 0

2、readLen>0 && !XLogReaderValidatePageHeader(state, targetSegmentPtr, state->readBuf)

--

3、读取checkpoint所在页readLen = state->read_page: readLen < 0

4、readLen > 0 && readLen <= SizeOfXLogShortPHD

5、!XLogReaderValidatePageHeader(state, pageptr, (char *) hdr)

四、XLogPageRead何时返回值<0 ?

/*

1、WaitForWALToBecomeAvailable open失败

2、lseek 失败 && ！StandbyMode

3、read失败 && ！StandbyMode

4、校验page头失败 && ！StandbyMode

如果是StandbyMode，则会重新retry->WaitForWALToBecomeAvailable,切换日志源进行open

*/

!WaitForWALToBecomeAvailable(targetPagePtr + reqLen,private->randAccess,1,targetRecPtr)//open

|-- return -1

readOff = targetPageOff;

if (lseek(readFile, (off_t) readOff, SEEK_SET) < 0){

！StandbyMode:: return -1

}

if (read(readFile, readBuf, XLOG_BLCKSZ) != XLOG_BLCKSZ){

！StandbyMode:: return -1

}

XLogReaderValidatePageHeader(xlogreader, targetPagePtr, readBuf)

！StandbyMode:: return -1

五、WaitForWALToBecomeAvailable何时返回false？

--XLOG_FROM_ARCHIVE | XLOG_FROM_PG_WAL

1、先XLogFileReadAnyTLI open日志：

1、遍历时间线列表里的每一个时间线，从最新的开始

2、当读取checkpoint的时候，source是XLOG_FROM_ANY

3、先找归档的日志进行open；如果open失败再找WAL日志进行open

4、如果都没有open成功，则向前找时间线，open前一个时间线segno和文件号相同的文件进行open

5、open成功后expectedTLEs被赋值为当前时间线列表的所有值

2、如果open失败，则切换日志源：XLOG_FROM_ARCHIVE | XLOG_FROM_PG_WAL -> XLOG_FROM_STREAM

3、切换日志源后，XLOG_FROM_ARCHIVE | XLOG_FROM_PG_WAL 则：

slave && promote ：return false

!StandbyMode：return false

--XLOG_FROM_STREAM

1、!WalRcvStreaming()即receiver进程挂了，切换日志源

2、CheckForStandbyTrigger()切换日志源

3、XLOG_FROM_STREAM->XLOG_FROM_ARCHIVE

Linux公社的RSS地址 ： https://www.linuxidc.com/rssFeed.aspx

本文永久更新链接地址： https://www.linuxidc.com/Linux/2019-02/157097.htm