Skip to content

G
Gpdb

Greenplum / Gpdb

通知 7
Star 1
Fork 0
G
Gpdb

前往新版Gitcode，体验更适合开发者的 AI 搜索 >>

提交 5a7f91f7 编写于 作者: A Ashwin Agrawal 提交者： Xin Zhang
浏览文件
操作

Remove filerep reset peer process code.

上级 5d10cf35
隐藏空白更改
内联 并排
Showing with 0 addition and 383 deletion
src/backend/bootstrap/bootstrap.c
浏览文件 @ 5a7f91f7
......@@ -50,8 +50,6 @@ extern char *optarg;
uint32 bootstrap_data_checksum_version = 0; /* No checksum */
extern void FileRepResetPeer_Main(void);
#define ALLOC(t, c) ((t *) calloc((unsigned)(c), sizeof(t)))
static void CheckerModeMain(void);
......@@ -501,10 +499,6 @@ AuxiliaryProcessMain(int argc, char *argv[])
FileRep_Main();
proc_exit(1); /* should never return */
case FilerepResetPeerProcess:
FileRepResetPeer_Main();
proc_exit(1); /* should never return */
default:
elog(PANIC, "unrecognized process type: %d", MyAuxProcType);
proc_exit(1);
......
src/backend/cdb/Makefile
浏览文件 @ 5a7f91f7
......@@ -28,7 +28,6 @@ OBJS = cdbappendonlystorage.o cdbappendonlystorageformat.o \
cdbfilerepprimary.o cdbfilerepmirror.o \
cdbfilerepprimaryack.o cdbfilerepmirrorack.o \
cdbfilerepconnclient.o cdbfilerepconnserver.o \
cdbfilerepresetpeerprocess.o \
cdbfts.o \
cdbglobalsequence.o \
cdbgroup.o \
......
src/backend/cdb/cdbfilerepresetpeerprocess.c 已删除 100644 → 0
浏览文件 @ 5d10cf35
/*-------------------------------------------------------------------------
*
* cdbfilerepresetpeerprocess.c
* Reset process to be launched during postmaster reset when we want to
* coordinate the reset with the filerep peer
*
* FileRepResetPeer_Main does two things:
* 1) set up general process code
* 2) run the coordination with the peer
*
* The process will exit cleanly with one of two codes:
* 1: (EXIT_CODE_SHOULD_ENTER_FAULT) if the postmaster should transition to
* a filerep fault rather than continue with database/mirror restart
* 0: (EXIT_CODE_SHOULD_RESTART_SHMEM_CLEANLY) if the postmaster should
* continue with shared memory reset followed by startup of the database
* and mirror
*
* Since it's hard to set up two-way communication over a single channel, we
* instead have the peer reset process work by periodically polling the other
* postmaster, which spawns a process to answer the "question" quickly.
*
* Coordinated reset works by:
*
* 1) Instruct peer to reset, at the same time receiving a "reset counter"
* (an integer, 0-padded to several digits).
* 2) Poll the peer for status. If the peer is also in a reset point (and
* we are a mirror*) then we can proceed with the restart. Or, if the
* peer is running and the returned counter is >= our counter from
* step 1), we can proceed with the restart. TODO: this check can
* actually be removed now that primaries and mirrors can be started
* in any order
*
* Portions Copyright (c) 2009-2010 Greenplum Inc
* Portions Copyright (c) 2012-Present Pivotal Software, Inc.
*
*
* IDENTIFICATION
* src/backend/cdb/cdbfilerepresetpeerprocess.c
*
*-------------------------------------------------------------------------
*/
#include "postgres.h"
#include <signal.h>
#include <unistd.h>
#include <sys/wait.h>
#include <sys/time.h>
#include "libpq/ip.h"
#include "libpq/pqsignal.h"
#include "miscadmin.h"
#include "storage/ipc.h"
#include "postmaster/postmaster.h"
#include "postmaster/primary_mirror_mode.h"
#include "postmaster/primary_mirror_transition_client.h"
static bool resetProcessShouldExit = false;
extern void FileRepResetPeer_Main(void);
#define EXIT_CODE_SHOULD_ENTER_FAULT 1
#define EXIT_CODE_SHOULD_RESTART_SHMEM_CLEANLY 0
#define MESSAGE_FROM_PEER_BUF_SIZE 3000
/**
* The reset "pivot point" is the time when children have been shut down but the new system has not been started
*/
#define RESET_STATUS_IS_IN_RESET_PIVOT_POINT "isInResetPivotPoint"
/**
* The server is running right now
*/
#define RESET_STATUS_IS_RUNNING "resetStatusServerIsRunning"
/*
* The server is starting or stopping right now (perhaps because of a reset)
*/
#define RESET_STATUS_IS_STARTING_OR_STOPPING "resetStatusServerIsStartingOrStopping"
static bool
strequals(char *left, char *right)
{
return strcmp(left, right) == 0;
}
static bool
determineTargetHost(struct addrinfo **addrList, char *host, char *port)
{
/*
* I should make this function exported by
* primary_mirror_transition_client.h ?
*/
struct addrinfo hint;
int ret;
*addrList = NULL;
/* Initialize hint structure */
MemSet(&hint, 0, sizeof(hint));
hint.ai_socktype = SOCK_STREAM;
hint.ai_family = AF_UNSPEC;
/* Using pghost, so we have to look-up the hostname */
hint.ai_family = AF_UNSPEC;
/* Use pg_getaddrinfo_all() to resolve the address */
ret = pg_getaddrinfo_all(host, port, &hint, addrList);
if (ret || !*addrList)
{
elog(WARNING, "could not translate host name \"%s\" to address: %s\n", host, gai_strerror(ret));
return false;
}
return true;
}
/* global variables filled in by callback functions */
static char gResultDataBuf[MESSAGE_FROM_PEER_BUF_SIZE];
static char gErrorLogBuf[MESSAGE_FROM_PEER_BUF_SIZE];
static bool gResponseWasTooLarge = false;
/* callback function needed by code that runs transitions */
static bool
resetPeer_checkForNeedToExitFunction(void)
{
return resetProcessShouldExit;
}
/* callback function needed by code that runs transitions */
static void
resetPeer_errorLogFunction(char *buf)
{
if (strlen(buf) + 1 > sizeof(gErrorLogBuf))
{
gResponseWasTooLarge = true;
return;
}
snprintf(gErrorLogBuf, sizeof(gErrorLogBuf), "%s", buf);
}
/* callback function needed by code that runs transitions */
static void
resetPeer_receivedDataCallbackFunction(char *buf)
{
if (strlen(buf) + 1 > sizeof(gResultDataBuf))
{
gResponseWasTooLarge = true;
return;
}
snprintf(gResultDataBuf, sizeof(gResultDataBuf), "%s", buf);
}
static void
sendMessageToPeerAndExitIfProblem(struct addrinfo *addrList, char *msgBody,
char messageFromPeerOut[MESSAGE_FROM_PEER_BUF_SIZE],
char resetNumberFromPeerOut[MESSAGE_FROM_PEER_BUF_SIZE])
{
elog(DEBUG1, "peer reset: sending message to primary/mirror peer: %s", msgBody);
/* set up receipt buffers (populated by the callback functions) */
gResponseWasTooLarge = false;
gErrorLogBuf[0] = '\0';
gResultDataBuf[0] = '\0';
/* make the call and check results */
PrimaryMirrorTransitionClientInfo client;
client.receivedDataCallbackFn = resetPeer_receivedDataCallbackFunction;
client.errorLogFn = resetPeer_errorLogFunction;
client.checkForNeedToExitFn = resetPeer_checkForNeedToExitFunction;
int resultCode = sendTransitionMessage(&client, addrList, msgBody, strlen(msgBody),
10 /* numRetries */ , 3600 /* transition_timeout */ );
if (resultCode != TRANS_ERRCODE_SUCCESS)
{
elog(WARNING, "during reset, unable to contact primary/mirror peer to coordinate reset; "
"will transition to fault state. Error code %d and message '%s'",
resultCode, gErrorLogBuf);
proc_exit(EXIT_CODE_SHOULD_ENTER_FAULT);
}
/*
* extract the two fields into messageFromPeerOut and
* resetNumberFromPeerOut, skipping the first Success: line
*
* is there a way to make this simple string parser easier?
*
* The result will look like Success:\nLineToKeep1\nLineToKeep2 This pulls
* LineToKeep1 and LineToKeep2 out into messageFromPeerOut and
* resetNumberFromPeerIndex
*
* Note that because gResultDataBuf is limited to
* MESSAGE_FROM_PEER_BUF_SIZE, we don't technically need to check overflow
* here.
*/
int resetNumberFromPeerIndex = 0,
messageFromPeerIndex = 0,
whichLine = 0;
char *buf = gResultDataBuf;
while (*buf)
{
if (*buf == '\n')
{
whichLine++;
if (whichLine == 3)
{
elog(WARNING, "during reset, invalid message contacting primary/mirror peer to coordinate reset; "
"will transition to fault state. Message received: %s",
gResultDataBuf);
proc_exit(EXIT_CODE_SHOULD_ENTER_FAULT);
}
}
else
{
if (whichLine == 1)
{
messageFromPeerOut[messageFromPeerIndex] = *buf;
messageFromPeerIndex++;
/* see comments above about why this is not strictly needed */
Insist(messageFromPeerIndex < MESSAGE_FROM_PEER_BUF_SIZE);
}
else if (whichLine == 2)
{
resetNumberFromPeerOut[resetNumberFromPeerIndex] = *buf;
resetNumberFromPeerIndex++;
/* see comments above about why this is not strictly needed */
Insist(resetNumberFromPeerIndex < MESSAGE_FROM_PEER_BUF_SIZE);
}
}
buf++;
}
messageFromPeerOut[messageFromPeerIndex] = '\0';
resetNumberFromPeerOut[resetNumberFromPeerIndex] = '\0';
if (whichLine != 2)
{
elog(WARNING, "during reset, invalid message contacting primary/mirror peer to coordinate reset; "
"will transition to fault state. Message received: %s",
gResultDataBuf);
proc_exit(EXIT_CODE_SHOULD_ENTER_FAULT);
}
}
static void
FileRepReset_ShutdownHandler(SIGNAL_ARGS)
{
resetProcessShouldExit = true;
}
static void
FileRepReset_HandleCrash(SIGNAL_ARGS)
{
StandardHandlerForSigillSigsegvSigbus_OnMainThread("filerep reset process", PASS_SIGNAL_ARGS);
}
static void
FileRepReset_ConfigureSignals(void)
{
/* Accept Signals */
/* shutdowns */
pqsignal(SIGQUIT, FileRepReset_ShutdownHandler);
#ifdef SIGBUS
pqsignal(SIGBUS, FileRepReset_HandleCrash);
#endif
#ifdef SIGILL
pqsignal(SIGILL, FileRepReset_HandleCrash);
#endif
#ifdef SIGSEGV
pqsignal(SIGSEGV, FileRepReset_HandleCrash);
#endif
/* Ignore Signals */
pqsignal(SIGHUP, SIG_IGN);
pqsignal(SIGCHLD, SIG_IGN);
pqsignal(SIGUSR1, SIG_IGN);
pqsignal(SIGTERM, SIG_IGN);
pqsignal(SIGALRM, SIG_IGN);
pqsignal(SIGPIPE, SIG_IGN);
pqsignal(SIGUSR2, SIG_IGN);
/* Use default action */
pqsignal(SIGINT, SIG_DFL);
pqsignal(SIGTTIN, SIG_DFL);
pqsignal(SIGTTOU, SIG_DFL);
pqsignal(SIGCONT, SIG_DFL);
pqsignal(SIGWINCH, SIG_DFL);
}
void
FileRepResetPeer_Main(void)
{
/* BASIC PROCESS SETUP */
FileRepReset_ConfigureSignals();
/*
* If an exception is encountered, processing resumes here.
*
* See notes in postgres.c about the design of this coding and comments
* about how the error handling works.
*/
sigjmp_buf local_sigjmp_buf;
if (sigsetjmp(local_sigjmp_buf, 1) != 0)
{
HOLD_INTERRUPTS();
EmitErrorReport();
proc_exit(EXIT_CODE_SHOULD_ENTER_FAULT);
}
/* We can now handle ereport(ERROR) */
PG_exception_stack = &local_sigjmp_buf;
PG_SETMASK(&UnBlockSig);
/** NOW DO THE ACTUAL WORK */
char messageFromPeer[MESSAGE_FROM_PEER_BUF_SIZE];
char resetNumberFromPeer[MESSAGE_FROM_PEER_BUF_SIZE];
char resetNumberThatIndicatesResetComplete[MESSAGE_FROM_PEER_BUF_SIZE];
struct addrinfo *addrList = NULL;
char portStr[100];
PrimaryMirrorModeTransitionArguments args = primaryMirrorGetArgumentsFromLocalMemory();
Assert(args.mode == PMModePrimarySegment || args.mode == PMModeMirrorSegment);
snprintf(portStr, sizeof(portStr), "%d", args.peerPostmasterPort);
if (!determineTargetHost(&addrList, args.peerAddress, portStr))
{
elog(WARNING, "during reset, unable to look up address for peer host to coordinate reset; "
"will transition to fault state.");
proc_exit(EXIT_CODE_SHOULD_ENTER_FAULT);
}
sendMessageToPeerAndExitIfProblem(addrList, "beginPostmasterReset", messageFromPeer,
resetNumberThatIndicatesResetComplete);
for (;;)
{
pg_usleep(10 * 1000L); /* 10 ms */
sendMessageToPeerAndExitIfProblem(addrList, "getPostmasterResetStatus", messageFromPeer, resetNumberFromPeer);
if (strequals(messageFromPeer, RESET_STATUS_IS_IN_RESET_PIVOT_POINT))
{
if (args.mode == PMModeMirrorSegment)
{
/**
* peer is in the reset pivot point, we can break out of our checking loop and
* thus exit with a code telling the postmaster to begin the startup sequence again
*
* this is only done on the mirror as currently the mirror must execute the startup sequence
* before the primary
*/
elog(DEBUG1, "peer reset: primary peer has reached reset point");
break;
}
}
else if (strequals(messageFromPeer, RESET_STATUS_IS_RUNNING))
{
/** it's running -- is it >= than the reset number that indicates reset complete one */
if (strcmp(resetNumberFromPeer, resetNumberThatIndicatesResetComplete) >= 0)
{
/** yes, the reset is complete and so we can quit and do a restart */
elog(DEBUG1, "peer reset: mirror peer reset is complete");
break;
}
}
}
proc_exit(EXIT_CODE_SHOULD_RESTART_SHMEM_CLEANLY);
}
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或