int main(int argc, char* argv[])
{
cmockery_parse_arguments(argc, argv);
const UnitTest tests[] =
{
unit_test(test__resetSessionForPrimaryGangLoss),
unit_test(test__createWriterGang),
unit_test(test__createReaderGang), };
MemoryContextInit();
CurrentResourceOwner = ResourceOwnerCreate(NULL, "gang test");
Gp_role = GP_ROLE_DISPATCH;
GpIdentity.numsegments = TOTOAL_SEGMENTS;
GpIdentity.dbid = 1;
GpIdentity.segindex = -1;
gp_connections_per_thread = 64;
Port procport;
MyProcPort = &procport;
MyProcPort->database_name = "test";
MyProcPort->user_name = "gpadmin";
s_cdb = makeTestCdb(1, TOTOAL_SEGMENTS);
return run_tests(tests);
}
/*
* Main worker routine. Accepts dsm_handle as an argument
*/
static void
bg_worker_main(Datum main_arg)
{
PartitionArgs *args;
dsm_handle handle = DatumGetInt32(main_arg);
/* Create resource owner */
CurrentResourceOwner = ResourceOwnerCreate(NULL, "CreatePartitionsWorker");
/* Attach to dynamic shared memory */
if (!handle)
{
ereport(WARNING,
(errmsg("pg_pathman worker: invalid dsm_handle")));
}
segment = dsm_attach(handle);
args = dsm_segment_address(segment);
/* Establish connection and start transaction */
BackgroundWorkerInitializeConnectionByOid(args->dbid, InvalidOid);
StartTransactionCommand();
SPI_connect();
PushActiveSnapshot(GetTransactionSnapshot());
/* Create partitions */
args->result = create_partitions(args->relid, PATHMAN_GET_DATUM(args->value, args->by_val), args->value_type, &args->crashed);
/* Cleanup */
SPI_finish();
PopActiveSnapshot();
CommitTransactionCommand();
dsm_detach(segment);
}
/* Main entry point for walsender process */
int
WalSenderMain(void)
{
MemoryContext walsnd_context;
am_cascading_walsender = RecoveryInProgress();
/* Create a per-walsender data structure in shared memory */
InitWalSnd();
/*
* Create a memory context that we will do all our work in. We do this so
* that we can reset the context during error recovery and thereby avoid
* possible memory leaks. Formerly this code just ran in
* TopMemoryContext, but resetting that would be a really bad idea.
*
* XXX: we don't actually attempt error recovery in walsender, we just
* close the connection and exit.
*/
walsnd_context = AllocSetContextCreate(TopMemoryContext,
"Wal Sender",
ALLOCSET_DEFAULT_MINSIZE,
ALLOCSET_DEFAULT_INITSIZE,
ALLOCSET_DEFAULT_MAXSIZE);
MemoryContextSwitchTo(walsnd_context);
/* Set up resource owner */
CurrentResourceOwner = ResourceOwnerCreate(NULL, "walsender top-level resource owner");
/* Unblock signals (they were blocked when the postmaster forked us) */
PG_SETMASK(&UnBlockSig);
/*
* Use the recovery target timeline ID during recovery
*/
if (am_cascading_walsender)
ThisTimeLineID = GetRecoveryTargetTLI();
/* Tell the standby that walsender is ready for receiving commands */
ReadyForQuery(DestRemote);
/* Handle handshake messages before streaming */
WalSndHandshake();
/* Initialize shared memory status */
{
/* use volatile pointer to prevent code rearrangement */
volatile WalSnd *walsnd = MyWalSnd;
SpinLockAcquire(&walsnd->mutex);
walsnd->sentPtr = sentPtr;
SpinLockRelease(&walsnd->mutex);
}
SyncRepInitConfig();
/* Main loop of walsender */
return WalSndLoop();
}
/*
* CreatePortal
* Returns a new portal given a name.
*
* allowDup: if true, automatically drop any pre-existing portal of the
* same name (if false, an error is raised).
*
* dupSilent: if true, don't even emit a WARNING.
*/
Portal
CreatePortal(const char *name, bool allowDup, bool dupSilent)
{
Portal portal;
AssertArg(PointerIsValid(name));
portal = GetPortalByName(name);
if (PortalIsValid(portal))
{
if (!allowDup)
ereport(ERROR,
(errcode(ERRCODE_DUPLICATE_CURSOR),
errmsg("cursor \"%s\" already exists", name)));
if (!dupSilent)
ereport(WARNING,
(errcode(ERRCODE_DUPLICATE_CURSOR),
errmsg("closing existing cursor \"%s\"",
name)));
PortalDrop(portal, false);
}
/* make new portal structure */
portal = (Portal) MemoryContextAllocZero(PortalMemory, sizeof *portal);
/* initialize portal heap context; typically it won't store much */
portal->heap = AllocSetContextCreate(PortalMemory,
"PortalHeapMemory",
ALLOCSET_SMALL_MINSIZE,
ALLOCSET_SMALL_INITSIZE,
ALLOCSET_SMALL_MAXSIZE);
/* create a resource owner for the portal */
portal->resowner = ResourceOwnerCreate(CurTransactionResourceOwner,
"Portal");
/* initialize portal fields that don't start off zero */
portal->cleanup = PortalCleanup;
portal->createSubid = GetCurrentSubTransactionId();
portal->strategy = PORTAL_MULTI_QUERY;
portal->cursorOptions = CURSOR_OPT_NO_SCROLL;
portal->atStart = true;
portal->atEnd = true; /* disallow fetches until query is set */
/* put portal in table (sets portal->name) */
PortalHashTableInsert(portal, name);
return portal;
}
/*
* Establish an AuxProcessResourceOwner for the current process.
*/
void
CreateAuxProcessResourceOwner(void)
{
Assert(AuxProcessResourceOwner == NULL);
Assert(CurrentResourceOwner == NULL);
AuxProcessResourceOwner = ResourceOwnerCreate(NULL, "AuxiliaryProcess");
CurrentResourceOwner = AuxProcessResourceOwner;
/*
* Register a shmem-exit callback for cleanup of aux-process resource
* owner. (This needs to run after, e.g., ShutdownXLOG.)
*/
on_shmem_exit(ReleaseAuxProcessResourcesCallback, 0);
}
static void
FileRepSubProcess_InitProcess(void)
{
SetProcessingMode(InitProcessing);
/*
* Create a resource owner to keep track of our resources
*/
CurrentResourceOwner = ResourceOwnerCreate(NULL,
FileRepProcessTypeToString[fileRepProcessType]);
InitXLOGAccess();
SetProcessingMode(NormalProcessing);
InitBufferPoolAccess();
/*
* Don't add Filerep backend subprocesses to the proc array.
*
* This avoids any deadlock situations during Filerep transition. E.g. If
* a normal backend has acquired ProcArrayLock and is waiting for Filerep
* transition to finish, the Filerep backend subprocesses will deadlock
* forever as they can't acquire the ProcArray lock to remove themselves
* from the ProcArray. This directly causes the transition to stall and
* thus the whole system.
*/
/*
* Initialize my entry in the shared-invalidation manager's array of
* per-backend data.
*
* Sets up MyBackendId, a unique backend identifier.
*/
MyBackendId = InvalidBackendId;
SharedInvalBackendInit(false);
if (MyBackendId > MaxBackends || MyBackendId <= 0)
elog(FATAL, "bad backend id: %d", MyBackendId);
/*
* bufmgr needs another initialization call too
*/
InitBufferPoolBackend();
}
/*
* SetResQueueId -- set the cached value for the current resource queue.
*
* Notes
* Needs to be called at session initialization and after (or in) SET ROLE.
*/
void
SetResQueueId(void)
{
/* to cave the code of cache part, we provide a resource owner here if no
* existing */
ResourceOwner owner = NULL;
if (CurrentResourceOwner == NULL)
{
owner = ResourceOwnerCreate(NULL, "SetResQueueId");
CurrentResourceOwner = owner;
}
MyQueueId = GetResQueueForRole(GetUserId());
if (owner)
{
CurrentResourceOwner = NULL;
ResourceOwnerDelete(owner);
}
return;
}
/* attach worker to the shared memory segment, read the job structure */
static void
initialize_worker(uint32 segment)
{
dsm_segment *seg;
ResourceOwner old,
tmp;
/* Connect to dynamic shared memory segment.
*
* In order to attach a dynamic shared memory segment, we need a
* resource owner. We cannot to StartTransactionCommand here, since
* we haven't yet attached to the database: to do this, we need to
* fetch information about connection properties from the shared
* memory segment.
*/
old = CurrentResourceOwner;
CurrentResourceOwner = ResourceOwnerCreate(NULL, "Background Worker");
seg = dsm_attach(segment);
if (seg == NULL)
ereport(ERROR,
(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
errmsg("unable to map dynamic shared memory segment")));
dsm_pin_mapping(seg);
tmp = CurrentResourceOwner;
CurrentResourceOwner = old;
ResourceOwnerDelete(tmp);
job = palloc(sizeof(JobDesc));
/* copy the arguments from shared memory segment */
memcpy(job, dsm_segment_address(seg), sizeof(JobDesc));
/* and detach it right away */
dsm_detach(seg);
Assert(job->magic == JOB_MAGIC);
job_run_function.schema = quote_identifier(job->schemaname);
job_run_function.name = quote_identifier("run_job");
}
/*
* Main entry point for checkpointer process
*
* This is invoked from AuxiliaryProcessMain, which has already created the
* basic execution environment, but not enabled signals yet.
*/
void
CheckpointerMain(void)
{
sigjmp_buf local_sigjmp_buf;
MemoryContext checkpointer_context;
CheckpointerShmem->checkpointer_pid = MyProcPid;
/*
* Properly accept or ignore signals the postmaster might send us
*
* Note: we deliberately ignore SIGTERM, because during a standard Unix
* system shutdown cycle, init will SIGTERM all processes at once. We
* want to wait for the backends to exit, whereupon the postmaster will
* tell us it's okay to shut down (via SIGUSR2).
*/
pqsignal(SIGHUP, ChkptSigHupHandler); /* set flag to read config
* file */
pqsignal(SIGINT, ReqCheckpointHandler); /* request checkpoint */
pqsignal(SIGTERM, SIG_IGN); /* ignore SIGTERM */
pqsignal(SIGQUIT, chkpt_quickdie); /* hard crash time */
pqsignal(SIGALRM, SIG_IGN);
pqsignal(SIGPIPE, SIG_IGN);
pqsignal(SIGUSR1, chkpt_sigusr1_handler);
pqsignal(SIGUSR2, ReqShutdownHandler); /* request shutdown */
/*
* Reset some signals that are accepted by postmaster but not here
*/
pqsignal(SIGCHLD, SIG_DFL);
pqsignal(SIGTTIN, SIG_DFL);
pqsignal(SIGTTOU, SIG_DFL);
pqsignal(SIGCONT, SIG_DFL);
pqsignal(SIGWINCH, SIG_DFL);
/* We allow SIGQUIT (quickdie) at all times */
sigdelset(&BlockSig, SIGQUIT);
/*
* Initialize so that first time-driven event happens at the correct time.
*/
last_checkpoint_time = last_xlog_switch_time = (pg_time_t) time(NULL);
/*
* Create a resource owner to keep track of our resources (currently only
* buffer pins).
*/
CurrentResourceOwner = ResourceOwnerCreate(NULL, "Checkpointer");
/*
* Create a memory context that we will do all our work in. We do this so
* that we can reset the context during error recovery and thereby avoid
* possible memory leaks. Formerly this code just ran in
* TopMemoryContext, but resetting that would be a really bad idea.
*/
checkpointer_context = AllocSetContextCreate(TopMemoryContext,
"Checkpointer",
ALLOCSET_DEFAULT_SIZES);
MemoryContextSwitchTo(checkpointer_context);
/*
* If an exception is encountered, processing resumes here.
*
* See notes in postgres.c about the design of this coding.
*/
if (sigsetjmp(local_sigjmp_buf, 1) != 0)
{
/* Since not using PG_TRY, must reset error stack by hand */
error_context_stack = NULL;
/* Prevent interrupts while cleaning up */
HOLD_INTERRUPTS();
/* Report the error to the server log */
EmitErrorReport();
/*
* These operations are really just a minimal subset of
* AbortTransaction(). We don't have very many resources to worry
* about in checkpointer, but we do have LWLocks, buffers, and temp
* files.
*/
LWLockReleaseAll();
ConditionVariableCancelSleep();
pgstat_report_wait_end();
AbortBufferIO();
UnlockBuffers();
/* buffer pins are released here: */
ResourceOwnerRelease(CurrentResourceOwner,
RESOURCE_RELEASE_BEFORE_LOCKS,
false, true);
/* we needn't bother with the other ResourceOwnerRelease phases */
AtEOXact_Buffers(false);
AtEOXact_SMgr();
AtEOXact_Files();
AtEOXact_HashTables(false);
/* Warn any waiting backends that the checkpoint failed. */
if (ckpt_active)
{
SpinLockAcquire(&CheckpointerShmem->ckpt_lck);
CheckpointerShmem->ckpt_failed++;
CheckpointerShmem->ckpt_done = CheckpointerShmem->ckpt_started;
SpinLockRelease(&CheckpointerShmem->ckpt_lck);
ckpt_active = false;
}
/*
* Now return to normal top-level context and clear ErrorContext for
* next time.
*/
MemoryContextSwitchTo(checkpointer_context);
FlushErrorState();
/* Flush any leaked data in the top-level context */
MemoryContextResetAndDeleteChildren(checkpointer_context);
/* Now we can allow interrupts again */
RESUME_INTERRUPTS();
/*
* Sleep at least 1 second after any error. A write error is likely
* to be repeated, and we don't want to be filling the error logs as
* fast as we can.
*/
pg_usleep(1000000L);
/*
* Close all open files after any error. This is helpful on Windows,
* where holding deleted files open causes various strange errors.
* It's not clear we need it elsewhere, but shouldn't hurt.
*/
smgrcloseall();
}
/* We can now handle ereport(ERROR) */
PG_exception_stack = &local_sigjmp_buf;
/*
* Unblock signals (they were blocked when the postmaster forked us)
*/
PG_SETMASK(&UnBlockSig);
/*
* Ensure all shared memory values are set correctly for the config. Doing
* this here ensures no race conditions from other concurrent updaters.
*/
UpdateSharedMemoryConfig();
/*
* Advertise our latch that backends can use to wake us up while we're
* sleeping.
*/
ProcGlobal->checkpointerLatch = &MyProc->procLatch;
/*
* Loop forever
*/
for (;;)
{
bool do_checkpoint = false;
int flags = 0;
pg_time_t now;
int elapsed_secs;
int cur_timeout;
int rc;
/* Clear any already-pending wakeups */
ResetLatch(MyLatch);
/*
* Process any requests or signals received recently.
*/
AbsorbFsyncRequests();
if (got_SIGHUP)
{
got_SIGHUP = false;
ProcessConfigFile(PGC_SIGHUP);
/*
* Checkpointer is the last process to shut down, so we ask it to
* hold the keys for a range of other tasks required most of which
* have nothing to do with checkpointing at all.
*
* For various reasons, some config values can change dynamically
* so the primary copy of them is held in shared memory to make
* sure all backends see the same value. We make Checkpointer
* responsible for updating the shared memory copy if the
* parameter setting changes because of SIGHUP.
*/
UpdateSharedMemoryConfig();
}
if (checkpoint_requested)
{
checkpoint_requested = false;
do_checkpoint = true;
BgWriterStats.m_requested_checkpoints++;
}
if (shutdown_requested)
{
/*
* From here on, elog(ERROR) should end with exit(1), not send
* control back to the sigsetjmp block above
*/
ExitOnAnyError = true;
/* Close down the database */
ShutdownXLOG(0, 0);
/* Normal exit from the checkpointer is here */
proc_exit(0); /* done */
}
/*
* Force a checkpoint if too much time has elapsed since the last one.
* Note that we count a timed checkpoint in stats only when this
* occurs without an external request, but we set the CAUSE_TIME flag
* bit even if there is also an external request.
*/
now = (pg_time_t) time(NULL);
elapsed_secs = now - last_checkpoint_time;
if (elapsed_secs >= CheckPointTimeout)
{
if (!do_checkpoint)
BgWriterStats.m_timed_checkpoints++;
do_checkpoint = true;
flags |= CHECKPOINT_CAUSE_TIME;
}
/*
* Do a checkpoint if requested.
*/
if (do_checkpoint)
{
bool ckpt_performed = false;
bool do_restartpoint;
/*
* Check if we should perform a checkpoint or a restartpoint. As a
* side-effect, RecoveryInProgress() initializes TimeLineID if
* it's not set yet.
*/
do_restartpoint = RecoveryInProgress();
/*
* Atomically fetch the request flags to figure out what kind of a
* checkpoint we should perform, and increase the started-counter
* to acknowledge that we've started a new checkpoint.
*/
SpinLockAcquire(&CheckpointerShmem->ckpt_lck);
flags |= CheckpointerShmem->ckpt_flags;
CheckpointerShmem->ckpt_flags = 0;
CheckpointerShmem->ckpt_started++;
SpinLockRelease(&CheckpointerShmem->ckpt_lck);
/*
* The end-of-recovery checkpoint is a real checkpoint that's
* performed while we're still in recovery.
*/
if (flags & CHECKPOINT_END_OF_RECOVERY)
do_restartpoint = false;
/*
* We will warn if (a) too soon since last checkpoint (whatever
* caused it) and (b) somebody set the CHECKPOINT_CAUSE_XLOG flag
* since the last checkpoint start. Note in particular that this
* implementation will not generate warnings caused by
* CheckPointTimeout < CheckPointWarning.
*/
if (!do_restartpoint &&
(flags & CHECKPOINT_CAUSE_XLOG) &&
elapsed_secs < CheckPointWarning)
ereport(LOG,
(errmsg_plural("checkpoints are occurring too frequently (%d second apart)",
"checkpoints are occurring too frequently (%d seconds apart)",
elapsed_secs,
elapsed_secs),
errhint("Consider increasing the configuration parameter \"max_wal_size\".")));
/*
* Initialize checkpointer-private variables used during
* checkpoint.
*/
ckpt_active = true;
if (do_restartpoint)
ckpt_start_recptr = GetXLogReplayRecPtr(NULL);
else
ckpt_start_recptr = GetInsertRecPtr();
ckpt_start_time = now;
ckpt_cached_elapsed = 0;
/*
* Do the checkpoint.
*/
if (!do_restartpoint)
{
CreateCheckPoint(flags);
ckpt_performed = true;
}
else
ckpt_performed = CreateRestartPoint(flags);
/*
* After any checkpoint, close all smgr files. This is so we
* won't hang onto smgr references to deleted files indefinitely.
*/
smgrcloseall();
/*
* Indicate checkpoint completion to any waiting backends.
*/
SpinLockAcquire(&CheckpointerShmem->ckpt_lck);
CheckpointerShmem->ckpt_done = CheckpointerShmem->ckpt_started;
SpinLockRelease(&CheckpointerShmem->ckpt_lck);
if (ckpt_performed)
{
/*
* Note we record the checkpoint start time not end time as
* last_checkpoint_time. This is so that time-driven
* checkpoints happen at a predictable spacing.
*/
last_checkpoint_time = now;
}
else
{
/*
* We were not able to perform the restartpoint (checkpoints
* throw an ERROR in case of error). Most likely because we
* have not received any new checkpoint WAL records since the
* last restartpoint. Try again in 15 s.
*/
last_checkpoint_time = now - CheckPointTimeout + 15;
}
ckpt_active = false;
}
/* Check for archive_timeout and switch xlog files if necessary. */
CheckArchiveTimeout();
/*
* Send off activity statistics to the stats collector. (The reason
* why we re-use bgwriter-related code for this is that the bgwriter
* and checkpointer used to be just one process. It's probably not
* worth the trouble to split the stats support into two independent
* stats message types.)
*/
pgstat_send_bgwriter();
/*
* Sleep until we are signaled or it's time for another checkpoint or
* xlog file switch.
*/
now = (pg_time_t) time(NULL);
elapsed_secs = now - last_checkpoint_time;
if (elapsed_secs >= CheckPointTimeout)
continue; /* no sleep for us ... */
cur_timeout = CheckPointTimeout - elapsed_secs;
if (XLogArchiveTimeout > 0 && !RecoveryInProgress())
{
elapsed_secs = now - last_xlog_switch_time;
if (elapsed_secs >= XLogArchiveTimeout)
continue; /* no sleep for us ... */
cur_timeout = Min(cur_timeout, XLogArchiveTimeout - elapsed_secs);
}
rc = WaitLatch(MyLatch,
WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH,
cur_timeout * 1000L /* convert to ms */,
WAIT_EVENT_CHECKPOINTER_MAIN);
/*
* Emergency bailout if postmaster has died. This is to avoid the
* necessity for manual cleanup of all postmaster children.
*/
if (rc & WL_POSTMASTER_DEATH)
exit(1);
}
}
/*
* Main entry point for walwriter process
*
* This is invoked from AuxiliaryProcessMain, which has already created the
* basic execution environment, but not enabled signals yet.
*/
void
WalWriterMain(void)
{
sigjmp_buf local_sigjmp_buf;
MemoryContext walwriter_context;
int left_till_hibernate;
bool hibernating;
/*
* Properly accept or ignore signals the postmaster might send us
*
* We have no particular use for SIGINT at the moment, but seems
* reasonable to treat like SIGTERM.
*/
pqsignal(SIGHUP, WalSigHupHandler); /* set flag to read config file */
pqsignal(SIGINT, WalShutdownHandler); /* request shutdown */
pqsignal(SIGTERM, WalShutdownHandler); /* request shutdown */
pqsignal(SIGQUIT, wal_quickdie); /* hard crash time */
pqsignal(SIGALRM, SIG_IGN);
pqsignal(SIGPIPE, SIG_IGN);
pqsignal(SIGUSR1, walwriter_sigusr1_handler);
pqsignal(SIGUSR2, SIG_IGN); /* not used */
/*
* Reset some signals that are accepted by postmaster but not here
*/
pqsignal(SIGCHLD, SIG_DFL);
pqsignal(SIGTTIN, SIG_DFL);
pqsignal(SIGTTOU, SIG_DFL);
pqsignal(SIGCONT, SIG_DFL);
pqsignal(SIGWINCH, SIG_DFL);
/* We allow SIGQUIT (quickdie) at all times */
sigdelset(&BlockSig, SIGQUIT);
/*
* Create a resource owner to keep track of our resources (not clear that
* we need this, but may as well have one).
*/
CurrentResourceOwner = ResourceOwnerCreate(NULL, "Wal Writer");
/*
* Create a memory context that we will do all our work in. We do this so
* that we can reset the context during error recovery and thereby avoid
* possible memory leaks. Formerly this code just ran in
* TopMemoryContext, but resetting that would be a really bad idea.
*/
walwriter_context = AllocSetContextCreate(TopMemoryContext,
"Wal Writer",
ALLOCSET_DEFAULT_SIZES);
MemoryContextSwitchTo(walwriter_context);
/*
* If an exception is encountered, processing resumes here.
*
* This code is heavily based on bgwriter.c, q.v.
*/
if (sigsetjmp(local_sigjmp_buf, 1) != 0)
{
/* Since not using PG_TRY, must reset error stack by hand */
error_context_stack = NULL;
/* Prevent interrupts while cleaning up */
HOLD_INTERRUPTS();
/* Report the error to the server log */
EmitErrorReport();
/*
* These operations are really just a minimal subset of
* AbortTransaction(). We don't have very many resources to worry
* about in walwriter, but we do have LWLocks, and perhaps buffers?
*/
LWLockReleaseAll();
ConditionVariableCancelSleep();
pgstat_report_wait_end();
AbortBufferIO();
UnlockBuffers();
/* buffer pins are released here: */
ResourceOwnerRelease(CurrentResourceOwner,
RESOURCE_RELEASE_BEFORE_LOCKS,
false, true);
/* we needn't bother with the other ResourceOwnerRelease phases */
AtEOXact_Buffers(false);
AtEOXact_SMgr();
AtEOXact_Files();
AtEOXact_HashTables(false);
/*
* Now return to normal top-level context and clear ErrorContext for
* next time.
*/
MemoryContextSwitchTo(walwriter_context);
FlushErrorState();
/* Flush any leaked data in the top-level context */
MemoryContextResetAndDeleteChildren(walwriter_context);
/* Now we can allow interrupts again */
RESUME_INTERRUPTS();
/*
* Sleep at least 1 second after any error. A write error is likely
* to be repeated, and we don't want to be filling the error logs as
* fast as we can.
*/
pg_usleep(1000000L);
/*
* Close all open files after any error. This is helpful on Windows,
* where holding deleted files open causes various strange errors.
* It's not clear we need it elsewhere, but shouldn't hurt.
*/
smgrcloseall();
}
/* We can now handle ereport(ERROR) */
PG_exception_stack = &local_sigjmp_buf;
/*
* Unblock signals (they were blocked when the postmaster forked us)
*/
PG_SETMASK(&UnBlockSig);
/*
* Reset hibernation state after any error.
*/
left_till_hibernate = LOOPS_UNTIL_HIBERNATE;
hibernating = false;
SetWalWriterSleeping(false);
/*
* Advertise our latch that backends can use to wake us up while we're
* sleeping.
*/
ProcGlobal->walwriterLatch = &MyProc->procLatch;
/*
* Loop forever
*/
for (;;)
{
long cur_timeout;
int rc;
/*
* Advertise whether we might hibernate in this cycle. We do this
* before resetting the latch to ensure that any async commits will
* see the flag set if they might possibly need to wake us up, and
* that we won't miss any signal they send us. (If we discover work
* to do in the last cycle before we would hibernate, the global flag
* will be set unnecessarily, but little harm is done.) But avoid
* touching the global flag if it doesn't need to change.
*/
if (hibernating != (left_till_hibernate <= 1))
{
hibernating = (left_till_hibernate <= 1);
SetWalWriterSleeping(hibernating);
}
/* Clear any already-pending wakeups */
ResetLatch(MyLatch);
/*
* Process any requests or signals received recently.
*/
if (got_SIGHUP)
{
got_SIGHUP = false;
ProcessConfigFile(PGC_SIGHUP);
}
if (shutdown_requested)
{
/* Normal exit from the walwriter is here */
proc_exit(0); /* done */
}
/*
* Do what we're here for; then, if XLogBackgroundFlush() found useful
* work to do, reset hibernation counter.
*/
if (XLogBackgroundFlush())
left_till_hibernate = LOOPS_UNTIL_HIBERNATE;
else if (left_till_hibernate > 0)
left_till_hibernate--;
/*
* Sleep until we are signaled or WalWriterDelay has elapsed. If we
* haven't done anything useful for quite some time, lengthen the
* sleep time so as to reduce the server's idle power consumption.
*/
if (left_till_hibernate > 0)
cur_timeout = WalWriterDelay; /* in ms */
else
cur_timeout = WalWriterDelay * HIBERNATE_FACTOR;
rc = WaitLatch(MyLatch,
WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH,
cur_timeout,
WAIT_EVENT_WAL_WRITER_MAIN);
/*
* Emergency bailout if postmaster has died. This is to avoid the
* necessity for manual cleanup of all postmaster children.
*/
if (rc & WL_POSTMASTER_DEATH)
exit(1);
}
}
/*
* Main entry point for bgwriter process
*
* This is invoked from BootstrapMain, which has already created the basic
* execution environment, but not enabled signals yet.
*/
void
BackgroundWriterMain(void)
{
sigjmp_buf local_sigjmp_buf;
MemoryContext bgwriter_context;
am_bg_writer = true;
/*
* If possible, make this process a group leader, so that the postmaster
* can signal any child processes too. (bgwriter probably never has any
* child processes, but for consistency we make all postmaster child
* processes do this.)
*/
#ifdef HAVE_SETSID
if (setsid() < 0)
elog(FATAL, "setsid() failed: %m");
#endif
/*
* Properly accept or ignore signals the postmaster might send us
*
* SIGUSR1 is presently unused; keep it spare in case someday we want this
* process to participate in ProcSignal signalling.
*/
pqsignal(SIGHUP, BgSigHupHandler); /* set flag to read config file */
pqsignal(SIGINT, SIG_IGN); /* as of 9.2 no longer requests checkpoint */
pqsignal(SIGTERM, ReqShutdownHandler); /* shutdown */
pqsignal(SIGQUIT, bg_quickdie); /* hard crash time */
pqsignal(SIGALRM, SIG_IGN);
pqsignal(SIGPIPE, SIG_IGN);
pqsignal(SIGUSR1, SIG_IGN); /* reserve for ProcSignal */
pqsignal(SIGUSR2, SIG_IGN);
/*
* Reset some signals that are accepted by postmaster but not here
*/
pqsignal(SIGCHLD, SIG_DFL);
pqsignal(SIGTTIN, SIG_DFL);
pqsignal(SIGTTOU, SIG_DFL);
pqsignal(SIGCONT, SIG_DFL);
pqsignal(SIGWINCH, SIG_DFL);
/* We allow SIGQUIT (quickdie) at all times */
sigdelset(&BlockSig, SIGQUIT);
/*
* Create a resource owner to keep track of our resources (currently only
* buffer pins).
*/
CurrentResourceOwner = ResourceOwnerCreate(NULL, "Background Writer");
/*
* Create a memory context that we will do all our work in. We do this so
* that we can reset the context during error recovery and thereby avoid
* possible memory leaks. Formerly this code just ran in
* TopMemoryContext, but resetting that would be a really bad idea.
*/
bgwriter_context = AllocSetContextCreate(TopMemoryContext,
"Background Writer",
ALLOCSET_DEFAULT_MINSIZE,
ALLOCSET_DEFAULT_INITSIZE,
ALLOCSET_DEFAULT_MAXSIZE);
MemoryContextSwitchTo(bgwriter_context);
/*
* If an exception is encountered, processing resumes here.
*
* See notes in postgres.c about the design of this coding.
*/
if (sigsetjmp(local_sigjmp_buf, 1) != 0)
{
/* Since not using PG_TRY, must reset error stack by hand */
error_context_stack = NULL;
/* Prevent interrupts while cleaning up */
HOLD_INTERRUPTS();
/* Report the error to the server log */
EmitErrorReport();
/*
* These operations are really just a minimal subset of
* AbortTransaction(). We don't have very many resources to worry
* about in bgwriter, but we do have LWLocks, buffers, and temp files.
*/
LWLockReleaseAll();
AbortBufferIO();
UnlockBuffers();
/* buffer pins are released here: */
ResourceOwnerRelease(CurrentResourceOwner,
RESOURCE_RELEASE_BEFORE_LOCKS,
false, true);
/* we needn't bother with the other ResourceOwnerRelease phases */
AtEOXact_Buffers(false);
AtEOXact_Files();
AtEOXact_HashTables(false);
/*
* Now return to normal top-level context and clear ErrorContext for
* next time.
*/
MemoryContextSwitchTo(bgwriter_context);
FlushErrorState();
/* Flush any leaked data in the top-level context */
MemoryContextResetAndDeleteChildren(bgwriter_context);
/* Now we can allow interrupts again */
RESUME_INTERRUPTS();
/*
* Sleep at least 1 second after any error. A write error is likely
* to be repeated, and we don't want to be filling the error logs as
* fast as we can.
*/
pg_usleep(1000000L);
/*
* Close all open files after any error. This is helpful on Windows,
* where holding deleted files open causes various strange errors.
* It's not clear we need it elsewhere, but shouldn't hurt.
*/
smgrcloseall();
}
/* We can now handle ereport(ERROR) */
PG_exception_stack = &local_sigjmp_buf;
/*
* Unblock signals (they were blocked when the postmaster forked us)
*/
PG_SETMASK(&UnBlockSig);
/*
* Use the recovery target timeline ID during recovery
*/
if (RecoveryInProgress())
ThisTimeLineID = GetRecoveryTargetTLI();
/*
* Loop forever
*/
for (;;)
{
/*
* Emergency bailout if postmaster has died. This is to avoid the
* necessity for manual cleanup of all postmaster children.
*/
if (!PostmasterIsAlive())
exit(1);
if (got_SIGHUP)
{
got_SIGHUP = false;
ProcessConfigFile(PGC_SIGHUP);
/* update global shmem state for sync rep */
}
if (shutdown_requested)
{
/*
* From here on, elog(ERROR) should end with exit(1), not send
* control back to the sigsetjmp block above
*/
ExitOnAnyError = true;
/* Normal exit from the bgwriter is here */
proc_exit(0); /* done */
}
/*
* Do one cycle of dirty-buffer writing.
*/
BgBufferSync();
/* Nap for the configured time. */
BgWriterNap();
}
}
/*
* Helper function for the various SQL callable logical decoding functions.
*/
static Datum
pg_logical_slot_get_changes_guts(FunctionCallInfo fcinfo, bool confirm, bool binary)
{
Name name = PG_GETARG_NAME(0);
XLogRecPtr upto_lsn;
int32 upto_nchanges;
ReturnSetInfo *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo;
MemoryContext per_query_ctx;
MemoryContext oldcontext;
XLogRecPtr end_of_wal;
XLogRecPtr startptr;
LogicalDecodingContext *ctx;
ResourceOwner old_resowner = CurrentResourceOwner;
ArrayType *arr;
Size ndim;
List *options = NIL;
DecodingOutputState *p;
if (PG_ARGISNULL(1))
upto_lsn = InvalidXLogRecPtr;
else
upto_lsn = PG_GETARG_LSN(1);
if (PG_ARGISNULL(2))
upto_nchanges = InvalidXLogRecPtr;
else
upto_nchanges = PG_GETARG_INT32(2);
/* check to see if caller supports us returning a tuplestore */
if (rsinfo == NULL || !IsA(rsinfo, ReturnSetInfo))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("set-valued function called in context that cannot accept a set")));
if (!(rsinfo->allowedModes & SFRM_Materialize))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("materialize mode required, but it is not allowed in this context")));
/* state to write output to */
p = palloc0(sizeof(DecodingOutputState));
p->binary_output = binary;
/* Build a tuple descriptor for our result type */
if (get_call_result_type(fcinfo, NULL, &p->tupdesc) != TYPEFUNC_COMPOSITE)
elog(ERROR, "return type must be a row type");
check_permissions();
CheckLogicalDecodingRequirements();
arr = PG_GETARG_ARRAYTYPE_P(3);
ndim = ARR_NDIM(arr);
per_query_ctx = rsinfo->econtext->ecxt_per_query_memory;
oldcontext = MemoryContextSwitchTo(per_query_ctx);
if (ndim > 1)
{
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("array must be one-dimensional")));
}
else if (array_contains_nulls(arr))
{
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("array must not contain nulls")));
}
else if (ndim == 1)
{
int nelems;
Datum *datum_opts;
int i;
Assert(ARR_ELEMTYPE(arr) == TEXTOID);
deconstruct_array(arr, TEXTOID, -1, false, 'i',
&datum_opts, NULL, &nelems);
if (nelems % 2 != 0)
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("array must have even number of elements")));
for (i = 0; i < nelems; i += 2)
{
char *name = TextDatumGetCString(datum_opts[i]);
char *opt = TextDatumGetCString(datum_opts[i + 1]);
options = lappend(options, makeDefElem(name, (Node *) makeString(opt)));
}
}
p->tupstore = tuplestore_begin_heap(true, false, work_mem);
rsinfo->returnMode = SFRM_Materialize;
rsinfo->setResult = p->tupstore;
rsinfo->setDesc = p->tupdesc;
/* compute the current end-of-wal */
if (!RecoveryInProgress())
end_of_wal = GetFlushRecPtr();
else
end_of_wal = GetXLogReplayRecPtr(NULL);
CheckLogicalDecodingRequirements();
ReplicationSlotAcquire(NameStr(*name));
PG_TRY();
{
ctx = CreateDecodingContext(InvalidXLogRecPtr,
options,
logical_read_local_xlog_page,
LogicalOutputPrepareWrite,
LogicalOutputWrite);
MemoryContextSwitchTo(oldcontext);
/*
* Check whether the output pluggin writes textual output if that's
* what we need.
*/
if (!binary &&
ctx->options.output_type != OUTPUT_PLUGIN_TEXTUAL_OUTPUT)
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("output plugin cannot produce binary output")));
ctx->output_writer_private = p;
startptr = MyReplicationSlot->data.restart_lsn;
CurrentResourceOwner = ResourceOwnerCreate(CurrentResourceOwner, "logical decoding");
/* invalidate non-timetravel entries */
InvalidateSystemCaches();
while ((startptr != InvalidXLogRecPtr && startptr < end_of_wal) ||
(ctx->reader->EndRecPtr && ctx->reader->EndRecPtr < end_of_wal))
{
XLogRecord *record;
char *errm = NULL;
record = XLogReadRecord(ctx->reader, startptr, &errm);
if (errm)
elog(ERROR, "%s", errm);
startptr = InvalidXLogRecPtr;
/*
* The {begin_txn,change,commit_txn}_wrapper callbacks above will
* store the description into our tuplestore.
*/
if (record != NULL)
LogicalDecodingProcessRecord(ctx, record);
/* check limits */
if (upto_lsn != InvalidXLogRecPtr &&
upto_lsn <= ctx->reader->EndRecPtr)
break;
if (upto_nchanges != 0 &&
upto_nchanges <= p->returned_rows)
break;
}
}
PG_CATCH();
{
/* clear all timetravel entries */
InvalidateSystemCaches();
PG_RE_THROW();
}
PG_END_TRY();
tuplestore_donestoring(tupstore);
CurrentResourceOwner = old_resowner;
/*
* Next time, start where we left off. (Hunting things, the family
* business..)
*/
if (ctx->reader->EndRecPtr != InvalidXLogRecPtr && confirm)
LogicalConfirmReceivedLocation(ctx->reader->EndRecPtr);
/* free context, call shutdown callback */
FreeDecodingContext(ctx);
ReplicationSlotRelease();
InvalidateSystemCaches();
return (Datum) 0;
}
/*
* Main entry point for bgwriter process
*
* This is invoked from BootstrapMain, which has already created the basic
* execution environment, but not enabled signals yet.
*/
void
BackgroundWriterMain(void)
{
sigjmp_buf local_sigjmp_buf;
MemoryContext bgwriter_context;
BgWriterShmem->bgwriter_pid = MyProcPid;
am_bg_writer = true;
/*
* If possible, make this process a group leader, so that the postmaster
* can signal any child processes too. (bgwriter probably never has any
* child processes, but for consistency we make all postmaster child
* processes do this.)
*/
#ifdef HAVE_SETSID
if (setsid() < 0)
elog(FATAL, "setsid() failed: %m");
#endif
/*
* Properly accept or ignore signals the postmaster might send us
*
* Note: we deliberately ignore SIGTERM, because during a standard Unix
* system shutdown cycle, init will SIGTERM all processes at once. We
* want to wait for the backends to exit, whereupon the postmaster will
* tell us it's okay to shut down (via SIGUSR2).
*
* SIGUSR1 is presently unused; keep it spare in case someday we want this
* process to participate in sinval messaging.
*/
pqsignal(SIGHUP, BgSigHupHandler); /* set flag to read config file */
pqsignal(SIGINT, ReqCheckpointHandler); /* request checkpoint */
pqsignal(SIGTERM, SIG_IGN); /* ignore SIGTERM */
pqsignal(SIGQUIT, bg_quickdie); /* hard crash time */
pqsignal(SIGALRM, SIG_IGN);
pqsignal(SIGPIPE, SIG_IGN);
pqsignal(SIGUSR1, SIG_IGN); /* reserve for sinval */
pqsignal(SIGUSR2, ReqShutdownHandler); /* request shutdown */
/*
* Reset some signals that are accepted by postmaster but not here
*/
pqsignal(SIGCHLD, SIG_DFL);
pqsignal(SIGTTIN, SIG_DFL);
pqsignal(SIGTTOU, SIG_DFL);
pqsignal(SIGCONT, SIG_DFL);
pqsignal(SIGWINCH, SIG_DFL);
/* We allow SIGQUIT (quickdie) at all times */
#ifdef HAVE_SIGPROCMASK
sigdelset(&BlockSig, SIGQUIT);
#else
BlockSig &= ~(sigmask(SIGQUIT));
#endif
/*
* Initialize so that first time-driven event happens at the correct time.
*/
last_checkpoint_time = last_xlog_switch_time = time(NULL);
/*
* Create a resource owner to keep track of our resources (currently only
* buffer pins).
*/
CurrentResourceOwner = ResourceOwnerCreate(NULL, "Background Writer");
/*
* Create a memory context that we will do all our work in. We do this so
* that we can reset the context during error recovery and thereby avoid
* possible memory leaks. Formerly this code just ran in
* TopMemoryContext, but resetting that would be a really bad idea.
*/
bgwriter_context = AllocSetContextCreate(TopMemoryContext,
"Background Writer",
ALLOCSET_DEFAULT_MINSIZE,
ALLOCSET_DEFAULT_INITSIZE,
ALLOCSET_DEFAULT_MAXSIZE);
MemoryContextSwitchTo(bgwriter_context);
/*
* If an exception is encountered, processing resumes here.
*
* See notes in postgres.c about the design of this coding.
*/
if (sigsetjmp(local_sigjmp_buf, 1) != 0)
{
/* Since not using PG_TRY, must reset error stack by hand */
error_context_stack = NULL;
/* Prevent interrupts while cleaning up */
HOLD_INTERRUPTS();
/* Report the error to the server log */
EmitErrorReport();
/*
* These operations are really just a minimal subset of
* AbortTransaction(). We don't have very many resources to worry
* about in bgwriter, but we do have LWLocks, buffers, and temp files.
*/
LWLockReleaseAll();
AbortBufferIO();
UnlockBuffers();
/* buffer pins are released here: */
ResourceOwnerRelease(CurrentResourceOwner,
RESOURCE_RELEASE_BEFORE_LOCKS,
false, true);
/* we needn't bother with the other ResourceOwnerRelease phases */
AtEOXact_Buffers(false);
AtEOXact_Files();
AtEOXact_HashTables(false);
/* Warn any waiting backends that the checkpoint failed. */
if (ckpt_active)
{
/* use volatile pointer to prevent code rearrangement */
volatile BgWriterShmemStruct *bgs = BgWriterShmem;
SpinLockAcquire(&bgs->ckpt_lck);
bgs->ckpt_failed++;
bgs->ckpt_done = bgs->ckpt_started;
SpinLockRelease(&bgs->ckpt_lck);
ckpt_active = false;
}
/*
* Now return to normal top-level context and clear ErrorContext for
* next time.
*/
MemoryContextSwitchTo(bgwriter_context);
FlushErrorState();
/* Flush any leaked data in the top-level context */
MemoryContextResetAndDeleteChildren(bgwriter_context);
/* Now we can allow interrupts again */
RESUME_INTERRUPTS();
/*
* Sleep at least 1 second after any error. A write error is likely
* to be repeated, and we don't want to be filling the error logs as
* fast as we can.
*/
pg_usleep(1000000L);
/*
* Close all open files after any error. This is helpful on Windows,
* where holding deleted files open causes various strange errors.
* It's not clear we need it elsewhere, but shouldn't hurt.
*/
smgrcloseall();
}
/* We can now handle ereport(ERROR) */
PG_exception_stack = &local_sigjmp_buf;
/*
* Unblock signals (they were blocked when the postmaster forked us)
*/
PG_SETMASK(&UnBlockSig);
/*
* Loop forever
*/
for (;;)
{
bool do_checkpoint = false;
int flags = 0;
time_t now;
int elapsed_secs;
/*
* Emergency bailout if postmaster has died. This is to avoid the
* necessity for manual cleanup of all postmaster children.
*/
if (!PostmasterIsAlive(true))
exit(1);
/*
* Process any requests or signals received recently.
*/
AbsorbFsyncRequests();
if (got_SIGHUP)
{
got_SIGHUP = false;
ProcessConfigFile(PGC_SIGHUP);
}
if (checkpoint_requested)
{
checkpoint_requested = false;
do_checkpoint = true;
BgWriterStats.m_requested_checkpoints++;
}
if (shutdown_requested)
{
/*
* From here on, elog(ERROR) should end with exit(1), not send
* control back to the sigsetjmp block above
*/
ExitOnAnyError = true;
/* Close down the database */
ShutdownXLOG(0, 0);
DumpFreeSpaceMap(0, 0);
/* Normal exit from the bgwriter is here */
proc_exit(0); /* done */
}
/*
* Force a checkpoint if too much time has elapsed since the last one.
* Note that we count a timed checkpoint in stats only when this
* occurs without an external request, but we set the CAUSE_TIME flag
* bit even if there is also an external request.
*/
now = time(NULL);
elapsed_secs = now - last_checkpoint_time;
if (elapsed_secs >= CheckPointTimeout)
{
if (!do_checkpoint)
BgWriterStats.m_timed_checkpoints++;
do_checkpoint = true;
flags |= CHECKPOINT_CAUSE_TIME;
}
/*
* Do a checkpoint if requested, otherwise do one cycle of
* dirty-buffer writing.
*/
if (do_checkpoint)
{
/* use volatile pointer to prevent code rearrangement */
volatile BgWriterShmemStruct *bgs = BgWriterShmem;
/*
* Atomically fetch the request flags to figure out what kind of a
* checkpoint we should perform, and increase the started-counter
* to acknowledge that we've started a new checkpoint.
*/
SpinLockAcquire(&bgs->ckpt_lck);
flags |= bgs->ckpt_flags;
bgs->ckpt_flags = 0;
bgs->ckpt_started++;
SpinLockRelease(&bgs->ckpt_lck);
/*
* We will warn if (a) too soon since last checkpoint (whatever
* caused it) and (b) somebody set the CHECKPOINT_CAUSE_XLOG flag
* since the last checkpoint start. Note in particular that this
* implementation will not generate warnings caused by
* CheckPointTimeout < CheckPointWarning.
*/
if ((flags & CHECKPOINT_CAUSE_XLOG) &&
elapsed_secs < CheckPointWarning)
ereport(LOG,
(errmsg("checkpoints are occurring too frequently (%d seconds apart)",
elapsed_secs),
errhint("Consider increasing the configuration parameter \"checkpoint_segments\".")));
/*
* Initialize bgwriter-private variables used during checkpoint.
*/
ckpt_active = true;
ckpt_start_recptr = GetInsertRecPtr();
ckpt_start_time = now;
ckpt_cached_elapsed = 0;
/*
* Do the checkpoint.
*/
CreateCheckPoint(flags);
/*
* After any checkpoint, close all smgr files. This is so we
* won't hang onto smgr references to deleted files indefinitely.
*/
smgrcloseall();
/*
* Indicate checkpoint completion to any waiting backends.
*/
SpinLockAcquire(&bgs->ckpt_lck);
bgs->ckpt_done = bgs->ckpt_started;
SpinLockRelease(&bgs->ckpt_lck);
ckpt_active = false;
/*
* Note we record the checkpoint start time not end time as
* last_checkpoint_time. This is so that time-driven checkpoints
* happen at a predictable spacing.
*/
last_checkpoint_time = now;
}
else
BgBufferSync();
/* Check for archive_timeout and switch xlog files if necessary. */
CheckArchiveTimeout();
/* Nap for the configured time. */
BgWriterNap();
}
}
/**
* This method is called after fork of the sweeper process. It sets up signal
* handlers and does initialization that is required by a postgres backend.
*/
NON_EXEC_STATIC void
BackoffSweeperMain(int argc, char *argv[])
{
sigjmp_buf local_sigjmp_buf;
IsUnderPostmaster = true;
isSweeperProcess = true;
/* Stay away from PMChildSlot */
MyPMChildSlot = -1;
/* reset MyProcPid */
MyProcPid = getpid();
/* Lose the postmaster's on-exit routines */
on_exit_reset();
/* Identify myself via ps */
init_ps_display("sweeper process", "", "", "");
SetProcessingMode(InitProcessing);
/*
* Set up signal handlers. We operate on databases much like a regular
* backend, so we use the same signal handling. See equivalent code in
* tcop/postgres.c.
*/
pqsignal(SIGHUP, SIG_IGN);
pqsignal(SIGINT, SIG_IGN);
pqsignal(SIGALRM, SIG_IGN);
pqsignal(SIGPIPE, SIG_IGN);
pqsignal(SIGUSR1, SIG_IGN);
pqsignal(SIGTERM, die);
pqsignal(SIGQUIT, quickdie);
pqsignal(SIGUSR2, BackoffRequestShutdown);
pqsignal(SIGFPE, FloatExceptionHandler);
pqsignal(SIGCHLD, SIG_DFL);
/*
* Copied from bgwriter
*/
CurrentResourceOwner = ResourceOwnerCreate(NULL, "Sweeper process");
/* Early initialization */
BaseInit();
/* See InitPostgres()... */
InitProcess();
SetProcessingMode(NormalProcessing);
/*
* If an exception is encountered, processing resumes here.
*
* See notes in postgres.c about the design of this coding.
*/
if (sigsetjmp(local_sigjmp_buf, 1) != 0)
{
/* Prevents interrupts while cleaning up */
HOLD_INTERRUPTS();
/* Report the error to the server log */
EmitErrorReport();
/*
* We can now go away. Note that because we'll call InitProcess, a
* callback will be registered to do ProcKill, which will clean up
* necessary state.
*/
proc_exit(0);
}
/* We can now handle ereport(ERROR) */
PG_exception_stack = &local_sigjmp_buf;
PG_SETMASK(&UnBlockSig);
MyBackendId = InvalidBackendId;
/* main loop */
BackoffSweeperLoop();
/* One iteration done, go away */
proc_exit(0);
}
/*
* Main entry point for walwriter process
*
* This is invoked from BootstrapMain, which has already created the basic
* execution environment, but not enabled signals yet.
*/
void
WalWriterMain(void)
{
sigjmp_buf local_sigjmp_buf;
MemoryContext walwriter_context;
/*
* If possible, make this process a group leader, so that the postmaster
* can signal any child processes too. (walwriter probably never has any
* child processes, but for consistency we make all postmaster child
* processes do this.)
*/
#ifdef HAVE_SETSID
if (setsid() < 0)
elog(FATAL, "setsid() failed: %m");
#endif
/*
* Properly accept or ignore signals the postmaster might send us
*
* We have no particular use for SIGINT at the moment, but seems
* reasonable to treat like SIGTERM.
*/
pqsignal(SIGHUP, WalSigHupHandler); /* set flag to read config file */
pqsignal(SIGINT, WalShutdownHandler); /* request shutdown */
pqsignal(SIGTERM, WalShutdownHandler); /* request shutdown */
pqsignal(SIGQUIT, wal_quickdie); /* hard crash time */
pqsignal(SIGALRM, SIG_IGN);
pqsignal(SIGPIPE, SIG_IGN);
pqsignal(SIGUSR1, SIG_IGN); /* reserve for ProcSignal */
pqsignal(SIGUSR2, SIG_IGN); /* not used */
/*
* Reset some signals that are accepted by postmaster but not here
*/
pqsignal(SIGCHLD, SIG_DFL);
pqsignal(SIGTTIN, SIG_DFL);
pqsignal(SIGTTOU, SIG_DFL);
pqsignal(SIGCONT, SIG_DFL);
pqsignal(SIGWINCH, SIG_DFL);
/* We allow SIGQUIT (quickdie) at all times */
sigdelset(&BlockSig, SIGQUIT);
/*
* Create a resource owner to keep track of our resources (not clear that
* we need this, but may as well have one).
*/
CurrentResourceOwner = ResourceOwnerCreate(NULL, "Wal Writer");
/*
* Create a memory context that we will do all our work in. We do this so
* that we can reset the context during error recovery and thereby avoid
* possible memory leaks. Formerly this code just ran in
* TopMemoryContext, but resetting that would be a really bad idea.
*/
walwriter_context = AllocSetContextCreate(TopMemoryContext,
"Wal Writer",
ALLOCSET_DEFAULT_MINSIZE,
ALLOCSET_DEFAULT_INITSIZE,
ALLOCSET_DEFAULT_MAXSIZE);
MemoryContextSwitchTo(walwriter_context);
/*
* If an exception is encountered, processing resumes here.
*
* This code is heavily based on bgwriter.c, q.v.
*/
if (sigsetjmp(local_sigjmp_buf, 1) != 0)
{
/* Since not using PG_TRY, must reset error stack by hand */
error_context_stack = NULL;
/* Prevent interrupts while cleaning up */
HOLD_INTERRUPTS();
/* Report the error to the server log */
EmitErrorReport();
/*
* These operations are really just a minimal subset of
* AbortTransaction(). We don't have very many resources to worry
* about in walwriter, but we do have LWLocks, and perhaps buffers?
*/
LWLockReleaseAll();
AbortBufferIO();
UnlockBuffers();
/* buffer pins are released here: */
ResourceOwnerRelease(CurrentResourceOwner,
RESOURCE_RELEASE_BEFORE_LOCKS,
false, true);
/* we needn't bother with the other ResourceOwnerRelease phases */
AtEOXact_Buffers(false);
AtEOXact_Files();
AtEOXact_HashTables(false);
/*
* Now return to normal top-level context and clear ErrorContext for
* next time.
*/
MemoryContextSwitchTo(walwriter_context);
FlushErrorState();
/* Flush any leaked data in the top-level context */
MemoryContextResetAndDeleteChildren(walwriter_context);
/* Now we can allow interrupts again */
RESUME_INTERRUPTS();
/*
* Sleep at least 1 second after any error. A write error is likely
* to be repeated, and we don't want to be filling the error logs as
* fast as we can.
*/
pg_usleep(1000000L);
/*
* Close all open files after any error. This is helpful on Windows,
* where holding deleted files open causes various strange errors.
* It's not clear we need it elsewhere, but shouldn't hurt.
*/
smgrcloseall();
}
/* We can now handle ereport(ERROR) */
PG_exception_stack = &local_sigjmp_buf;
/*
* Unblock signals (they were blocked when the postmaster forked us)
*/
PG_SETMASK(&UnBlockSig);
/*
* Loop forever
*/
for (;;)
{
long udelay;
/*
* Emergency bailout if postmaster has died. This is to avoid the
* necessity for manual cleanup of all postmaster children.
*/
if (!PostmasterIsAlive(true))
exit(1);
/*
* Process any requests or signals received recently.
*/
if (got_SIGHUP)
{
got_SIGHUP = false;
ProcessConfigFile(PGC_SIGHUP);
}
if (shutdown_requested)
{
/* Normal exit from the walwriter is here */
proc_exit(0); /* done */
}
/*
* Do what we're here for...
*/
XLogBackgroundFlush();
/*
* Delay until time to do something more, but fall out of delay
* reasonably quickly if signaled.
*/
udelay = WalWriterDelay * 1000L;
while (udelay > 999999L)
{
if (got_SIGHUP || shutdown_requested)
break;
pg_usleep(1000000L);
udelay -= 1000000L;
}
if (!(got_SIGHUP || shutdown_requested))
pg_usleep(udelay);
}
}
/*
* Background worker entrypoint.
*
* This is intended to demonstrate how a background worker can be used to
* facilitate a parallel computation. Most of the logic here is fairly
* boilerplate stuff, designed to attach to the shared memory segment,
* notify the user backend that we're alive, and so on. The
* application-specific bits of logic that you'd replace for your own worker
* are attach_to_queues() and copy_messages().
*/
void
test_shm_mq_main(Datum main_arg)
{
dsm_segment *seg;
shm_toc *toc;
shm_mq_handle *inqh;
shm_mq_handle *outqh;
volatile test_shm_mq_header *hdr;
int myworkernumber;
PGPROC *registrant;
/*
* Establish signal handlers.
*
* We want CHECK_FOR_INTERRUPTS() to kill off this worker process just as
* it would a normal user backend. To make that happen, we establish a
* signal handler that is a stripped-down version of die(). We don't have
* any equivalent of the backend's command-read loop, where interrupts can
* be processed immediately, so make sure ImmediateInterruptOK is turned
* off.
*/
pqsignal(SIGTERM, handle_sigterm);
ImmediateInterruptOK = false;
BackgroundWorkerUnblockSignals();
/*
* Connect to the dynamic shared memory segment.
*
* The backend that registered this worker passed us the ID of a shared
* memory segment to which we must attach for further instructions. In
* order to attach to dynamic shared memory, we need a resource owner.
* Once we've mapped the segment in our address space, attach to the table
* of contents so we can locate the various data structures we'll need to
* find within the segment.
*/
CurrentResourceOwner = ResourceOwnerCreate(NULL, "test_shm_mq worker");
seg = dsm_attach(DatumGetInt32(main_arg));
if (seg == NULL)
ereport(ERROR,
(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
errmsg("unable to map dynamic shared memory segment")));
toc = shm_toc_attach(PG_TEST_SHM_MQ_MAGIC, dsm_segment_address(seg));
if (toc == NULL)
ereport(ERROR,
(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
errmsg("bad magic number in dynamic shared memory segment")));
/*
* Acquire a worker number.
*
* By convention, the process registering this background worker should
* have stored the control structure at key 0. We look up that key to
* find it. Our worker number gives our identity: there may be just one
* worker involved in this parallel operation, or there may be many.
*/
hdr = shm_toc_lookup(toc, 0);
SpinLockAcquire(&hdr->mutex);
myworkernumber = ++hdr->workers_attached;
SpinLockRelease(&hdr->mutex);
if (myworkernumber > hdr->workers_total)
ereport(ERROR,
(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
errmsg("too many message queue testing workers already")));
/*
* Attach to the appropriate message queues.
*/
attach_to_queues(seg, toc, myworkernumber, &inqh, &outqh);
/*
* Indicate that we're fully initialized and ready to begin the main part
* of the parallel operation.
*
* Once we signal that we're ready, the user backend is entitled to assume
* that our on_dsm_detach callbacks will fire before we disconnect from
* the shared memory segment and exit. Generally, that means we must have
* attached to all relevant dynamic shared memory data structures by now.
*/
SpinLockAcquire(&hdr->mutex);
++hdr->workers_ready;
SpinLockRelease(&hdr->mutex);
registrant = BackendPidGetProc(MyBgworkerEntry->bgw_notify_pid);
if (registrant == NULL)
{
elog(DEBUG1, "registrant backend has exited prematurely");
proc_exit(1);
}
SetLatch(®istrant->procLatch);
/* Do the work. */
copy_messages(inqh, outqh);
/*
* We're done. Explicitly detach the shared memory segment so that we
* don't get a resource leak warning at commit time. This will fire any
* on_dsm_detach callbacks we've registered, as well. Once that's done,
* we can go ahead and exit.
*/
dsm_detach(seg);
proc_exit(1);
}
void
worker_test_main(Datum main_arg)
{
dsm_segment *seg;
volatile test_shm_mq_header *hdr;
PGPROC *registrant;
pqsignal(SIGHUP, handle_sighup);
pqsignal(SIGTERM, handle_sigterm);
BackgroundWorkerUnblockSignals();
printf("worker_test_main: %d\n", DatumGetInt32(main_arg));
CurrentResourceOwner = ResourceOwnerCreate(NULL, "worker test");
seg = dsm_attach(DatumGetInt32(main_arg));
if (seg == NULL)
ereport(ERROR,
(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
errmsg("unable to map dynamic shared memory segment")));
hdr = dsm_segment_address(seg);
/* 開始 */
SpinLockAcquire(&hdr->mutex);
hdr->workers_ready++;
hdr->workers_attached++;
SpinLockRelease(&hdr->mutex);
registrant = BackendPidGetProc(MyBgworkerEntry->bgw_notify_pid);
if (registrant == NULL)
{
elog(DEBUG1, "registrant backend has exited prematurely");
proc_exit(1);
}
SetLatch(®istrant->procLatch);
/* Do the work */
BackgroundWorkerInitializeConnection(hdr->dbname, NULL);
printf("DSM: %p\n", dsm_segment_address);
#if 0
SetCurrentStatementStartTimestamp();
StartTransactionCommand();
SPI_connect();
PushActiveSnapshot(GetTransactionSnapshot());
pgstat_report_activity(STATE_RUNNING, "initializing spi_worker schema");
SPI_finish();
PopActiveSnapshot();
CommitTransactionCommand();
pgstat_report_activity(STATE_IDLE, NULL);
#endif
dsm_detach(seg);
proc_exit(0);
}
/*
* Helper function for the various SQL callable logical decoding functions.
*/
static Datum
pg_logical_slot_get_changes_guts(FunctionCallInfo fcinfo, bool confirm, bool binary)
{
Name name;
XLogRecPtr upto_lsn;
int32 upto_nchanges;
ReturnSetInfo *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo;
MemoryContext per_query_ctx;
MemoryContext oldcontext;
XLogRecPtr end_of_wal;
XLogRecPtr startptr;
LogicalDecodingContext *ctx;
ResourceOwner old_resowner = CurrentResourceOwner;
ArrayType *arr;
Size ndim;
List *options = NIL;
DecodingOutputState *p;
check_permissions();
CheckLogicalDecodingRequirements();
if (PG_ARGISNULL(0))
ereport(ERROR,
(errcode(ERRCODE_NULL_VALUE_NOT_ALLOWED),
errmsg("slot name must not be null")));
name = PG_GETARG_NAME(0);
if (PG_ARGISNULL(1))
upto_lsn = InvalidXLogRecPtr;
else
upto_lsn = PG_GETARG_LSN(1);
if (PG_ARGISNULL(2))
upto_nchanges = InvalidXLogRecPtr;
else
upto_nchanges = PG_GETARG_INT32(2);
if (PG_ARGISNULL(3))
ereport(ERROR,
(errcode(ERRCODE_NULL_VALUE_NOT_ALLOWED),
errmsg("options array must not be null")));
arr = PG_GETARG_ARRAYTYPE_P(3);
/* check to see if caller supports us returning a tuplestore */
if (rsinfo == NULL || !IsA(rsinfo, ReturnSetInfo))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("set-valued function called in context that cannot accept a set")));
if (!(rsinfo->allowedModes & SFRM_Materialize))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("materialize mode required, but it is not allowed in this context")));
/* state to write output to */
p = palloc0(sizeof(DecodingOutputState));
p->binary_output = binary;
/* Build a tuple descriptor for our result type */
if (get_call_result_type(fcinfo, NULL, &p->tupdesc) != TYPEFUNC_COMPOSITE)
elog(ERROR, "return type must be a row type");
per_query_ctx = rsinfo->econtext->ecxt_per_query_memory;
oldcontext = MemoryContextSwitchTo(per_query_ctx);
/* Deconstruct options array */
ndim = ARR_NDIM(arr);
if (ndim > 1)
{
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("array must be one-dimensional")));
}
else if (array_contains_nulls(arr))
{
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("array must not contain nulls")));
}
else if (ndim == 1)
{
int nelems;
Datum *datum_opts;
int i;
Assert(ARR_ELEMTYPE(arr) == TEXTOID);
deconstruct_array(arr, TEXTOID, -1, false, 'i',
&datum_opts, NULL, &nelems);
if (nelems % 2 != 0)
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("array must have even number of elements")));
for (i = 0; i < nelems; i += 2)
{
char *name = TextDatumGetCString(datum_opts[i]);
char *opt = TextDatumGetCString(datum_opts[i + 1]);
options = lappend(options, makeDefElem(name, (Node *) makeString(opt), -1));
}
}
p->tupstore = tuplestore_begin_heap(true, false, work_mem);
rsinfo->returnMode = SFRM_Materialize;
rsinfo->setResult = p->tupstore;
rsinfo->setDesc = p->tupdesc;
/*
* Compute the current end-of-wal and maintain ThisTimeLineID.
* RecoveryInProgress() will update ThisTimeLineID on promotion.
*/
if (!RecoveryInProgress())
end_of_wal = GetFlushRecPtr();
else
end_of_wal = GetXLogReplayRecPtr(&ThisTimeLineID);
ReplicationSlotAcquire(NameStr(*name), true);
PG_TRY();
{
/* restart at slot's confirmed_flush */
ctx = CreateDecodingContext(InvalidXLogRecPtr,
options,
false,
logical_read_local_xlog_page,
LogicalOutputPrepareWrite,
LogicalOutputWrite, NULL);
MemoryContextSwitchTo(oldcontext);
/*
* Check whether the output plugin writes textual output if that's
* what we need.
*/
if (!binary &&
ctx->options.output_type !=OUTPUT_PLUGIN_TEXTUAL_OUTPUT)
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("logical decoding output plugin \"%s\" produces binary output, but function \"%s\" expects textual data",
NameStr(MyReplicationSlot->data.plugin),
format_procedure(fcinfo->flinfo->fn_oid))));
ctx->output_writer_private = p;
/*
* Decoding of WAL must start at restart_lsn so that the entirety of
* xacts that committed after the slot's confirmed_flush can be
* accumulated into reorder buffers.
*/
startptr = MyReplicationSlot->data.restart_lsn;
CurrentResourceOwner = ResourceOwnerCreate(CurrentResourceOwner, "logical decoding");
/* invalidate non-timetravel entries */
InvalidateSystemCaches();
/* Decode until we run out of records */
while ((startptr != InvalidXLogRecPtr && startptr < end_of_wal) ||
(ctx->reader->EndRecPtr != InvalidXLogRecPtr && ctx->reader->EndRecPtr < end_of_wal))
{
XLogRecord *record;
char *errm = NULL;
record = XLogReadRecord(ctx->reader, startptr, &errm);
if (errm)
elog(ERROR, "%s", errm);
/*
* Now that we've set up the xlog reader state, subsequent calls
* pass InvalidXLogRecPtr to say "continue from last record"
*/
startptr = InvalidXLogRecPtr;
/*
* The {begin_txn,change,commit_txn}_wrapper callbacks above will
* store the description into our tuplestore.
*/
if (record != NULL)
LogicalDecodingProcessRecord(ctx, ctx->reader);
/* check limits */
if (upto_lsn != InvalidXLogRecPtr &&
upto_lsn <= ctx->reader->EndRecPtr)
break;
if (upto_nchanges != 0 &&
upto_nchanges <= p->returned_rows)
break;
CHECK_FOR_INTERRUPTS();
}
tuplestore_donestoring(tupstore);
CurrentResourceOwner = old_resowner;
/*
* Next time, start where we left off. (Hunting things, the family
* business..)
*/
if (ctx->reader->EndRecPtr != InvalidXLogRecPtr && confirm)
{
LogicalConfirmReceivedLocation(ctx->reader->EndRecPtr);
/*
* If only the confirmed_flush_lsn has changed the slot won't get
* marked as dirty by the above. Callers on the walsender
* interface are expected to keep track of their own progress and
* don't need it written out. But SQL-interface users cannot
* specify their own start positions and it's harder for them to
* keep track of their progress, so we should make more of an
* effort to save it for them.
*
* Dirty the slot so it's written out at the next checkpoint.
* We'll still lose its position on crash, as documented, but it's
* better than always losing the position even on clean restart.
*/
ReplicationSlotMarkDirty();
}
/* free context, call shutdown callback */
FreeDecodingContext(ctx);
ReplicationSlotRelease();
InvalidateSystemCaches();
}
PG_CATCH();
{
/* clear all timetravel entries */
InvalidateSystemCaches();
PG_RE_THROW();
}
PG_END_TRY();
return (Datum) 0;
}
void
InitResQueues(void)
{
HeapTuple tuple;
int numQueues = 0;
bool queuesok = true;
cqContext *pcqCtx;
cqContext cqc;
Assert(ResScheduler);
/*
* Need a resource owner to keep the heapam code happy.
*/
Assert(CurrentResourceOwner == NULL);
ResourceOwner owner = ResourceOwnerCreate(NULL, "InitQueues");
CurrentResourceOwner = owner;
/**
* The resqueue shared mem initialization must be serialized. Only the first session
* should do the init.
* Serialization is done the ResQueueLock LW_EXCLUSIVE. However, we must obtain all DB
* lock before obtaining LWlock.
* So, we must have obtained ResQueueRelationId and ResQueueCapabilityRelationId lock
* first.
*/
Relation relResqueue = heap_open(ResQueueRelationId, AccessShareLock);
LockRelationOid(ResQueueCapabilityRelationId, RowExclusiveLock);
LWLockAcquire(ResQueueLock, LW_EXCLUSIVE);
if (ResScheduler->num_queues > 0)
{
/* Hash table has already been loaded */
LWLockRelease(ResQueueLock);
UnlockRelationOid(ResQueueCapabilityRelationId, RowExclusiveLock);
heap_close(relResqueue, AccessShareLock);
CurrentResourceOwner = NULL;
ResourceOwnerDelete(owner);
return;
}
/* XXX XXX: should this be rowexclusive ? */
pcqCtx = caql_beginscan(
caql_indexOK(
caql_addrel(cqclr(&cqc), relResqueue),
false),
cql("SELECT * FROM pg_resqueue ", NULL));
while (HeapTupleIsValid(tuple = caql_getnext(pcqCtx)))
{
Form_pg_resqueue queueform;
Oid queueid;
bool overcommit;
float4 ignorelimit;
Cost thresholds[NUM_RES_LIMIT_TYPES];
char *queuename;
numQueues++;
queueform = (Form_pg_resqueue) GETSTRUCT(tuple);
queueid = HeapTupleGetOid(tuple);
queuename = NameStr(queueform->rsqname);
thresholds[RES_COUNT_LIMIT] = queueform->rsqcountlimit;
thresholds[RES_COST_LIMIT] = queueform->rsqcostlimit;
thresholds[RES_MEMORY_LIMIT] = ResourceQueueGetMemoryLimit(queueid);
overcommit = queueform->rsqovercommit;
ignorelimit = queueform->rsqignorecostlimit;
queuesok = ResCreateQueue(queueid, thresholds, overcommit, ignorelimit);
if (!queuesok)
{
/** Break out of loop. Close relations, relinquish LWLock and then error out */
break;
}
}
caql_endscan(pcqCtx);
LWLockRelease(ResQueueLock);
UnlockRelationOid(ResQueueCapabilityRelationId, RowExclusiveLock);
heap_close(relResqueue, AccessShareLock);
if (!queuesok)
ereport(PANIC,
(errcode(ERRCODE_INSUFFICIENT_RESOURCES),
errmsg("insufficient resource queues available"),
errhint("Increase max_resource_queues to %d.", numQueues)));
elog(LOG,"initialized %d resource queues", numQueues);
CurrentResourceOwner = NULL;
ResourceOwnerDelete(owner);
return;
}
/* Main entry point for walreceiver process */
void
WalReceiverMain(void)
{
char conninfo[MAXCONNINFO];
XLogRecPtr startpoint;
/* use volatile pointer to prevent code rearrangement */
volatile WalRcvData *walrcv = WalRcv;
am_walreceiver = true;
/*
* WalRcv should be set up already (if we are a backend, we inherit this
* by fork() or EXEC_BACKEND mechanism from the postmaster).
*/
Assert(walrcv != NULL);
/*
* Mark walreceiver as running in shared memory.
*
* Do this as early as possible, so that if we fail later on, we'll set
* state to STOPPED. If we die before this, the startup process will keep
* waiting for us to start up, until it times out.
*/
SpinLockAcquire(&walrcv->mutex);
Assert(walrcv->pid == 0);
switch (walrcv->walRcvState)
{
case WALRCV_STOPPING:
/* If we've already been requested to stop, don't start up. */
walrcv->walRcvState = WALRCV_STOPPED;
/* fall through */
case WALRCV_STOPPED:
SpinLockRelease(&walrcv->mutex);
proc_exit(1);
break;
case WALRCV_STARTING:
/* The usual case */
break;
case WALRCV_RUNNING:
/* Shouldn't happen */
elog(PANIC, "walreceiver still running according to shared memory state");
}
/* Advertise our PID so that the startup process can kill us */
walrcv->pid = MyProcPid;
walrcv->walRcvState = WALRCV_RUNNING;
/* Fetch information required to start streaming */
strlcpy(conninfo, (char *) walrcv->conninfo, MAXCONNINFO);
startpoint = walrcv->receivedUpto;
SpinLockRelease(&walrcv->mutex);
/* Arrange to clean up at walreceiver exit */
on_shmem_exit(WalRcvDie, 0);
/*
* If possible, make this process a group leader, so that the postmaster
* can signal any child processes too. (walreceiver probably never has
* any child processes, but for consistency we make all postmaster child
* processes do this.)
*/
#ifdef HAVE_SETSID
if (setsid() < 0)
elog(FATAL, "setsid() failed: %m");
#endif
/* Properly accept or ignore signals the postmaster might send us */
pqsignal(SIGHUP, WalRcvSigHupHandler); /* set flag to read config
* file */
pqsignal(SIGINT, SIG_IGN);
pqsignal(SIGTERM, WalRcvShutdownHandler); /* request shutdown */
pqsignal(SIGQUIT, WalRcvQuickDieHandler); /* hard crash time */
pqsignal(SIGALRM, SIG_IGN);
pqsignal(SIGPIPE, SIG_IGN);
pqsignal(SIGUSR1, SIG_IGN);
pqsignal(SIGUSR2, SIG_IGN);
/* Reset some signals that are accepted by postmaster but not here */
pqsignal(SIGCHLD, SIG_DFL);
pqsignal(SIGTTIN, SIG_DFL);
pqsignal(SIGTTOU, SIG_DFL);
pqsignal(SIGCONT, SIG_DFL);
pqsignal(SIGWINCH, SIG_DFL);
/* We allow SIGQUIT (quickdie) at all times */
sigdelset(&BlockSig, SIGQUIT);
/* Load the libpq-specific functions */
load_file("libpqwalreceiver", false);
if (walrcv_connect == NULL || walrcv_receive == NULL ||
walrcv_disconnect == NULL)
elog(ERROR, "libpqwalreceiver didn't initialize correctly");
/*
* Create a resource owner to keep track of our resources (not clear that
* we need this, but may as well have one).
*/
CurrentResourceOwner = ResourceOwnerCreate(NULL, "Wal Receiver");
/* Unblock signals (they were blocked when the postmaster forked us) */
PG_SETMASK(&UnBlockSig);
/* Establish the connection to the primary for XLOG streaming */
EnableWalRcvImmediateExit();
walrcv_connect(conninfo, startpoint);
DisableWalRcvImmediateExit();
/* Loop until end-of-streaming or error */
for (;;)
{
unsigned char type;
char *buf;
int len;
/*
* Emergency bailout if postmaster has died. This is to avoid the
* necessity for manual cleanup of all postmaster children.
*/
if (!PostmasterIsAlive(true))
exit(1);
/*
* Exit walreceiver if we're not in recovery. This should not happen,
* but cross-check the status here.
*/
if (!RecoveryInProgress())
ereport(FATAL,
(errmsg("cannot continue WAL streaming, recovery has already ended")));
/* Process any requests or signals received recently */
ProcessWalRcvInterrupts();
if (got_SIGHUP)
{
got_SIGHUP = false;
ProcessConfigFile(PGC_SIGHUP);
}
/* Wait a while for data to arrive */
if (walrcv_receive(NAPTIME_PER_CYCLE, &type, &buf, &len))
{
/* Accept the received data, and process it */
XLogWalRcvProcessMsg(type, buf, len);
/* Receive any more data we can without sleeping */
while (walrcv_receive(0, &type, &buf, &len))
XLogWalRcvProcessMsg(type, buf, len);
/*
* If we've written some records, flush them to disk and let the
* startup process know about them.
*/
XLogWalRcvFlush();
}
}
}
/*
* Main entry point for bgwriter process
*
* This is invoked from AuxiliaryProcessMain, which has already created the
* basic execution environment, but not enabled signals yet.
*/
void
BackgroundWriterMain(void)
{
sigjmp_buf local_sigjmp_buf;
MemoryContext bgwriter_context;
bool prev_hibernate;
/*
* Properly accept or ignore signals the postmaster might send us.
*
* bgwriter doesn't participate in ProcSignal signalling, but a SIGUSR1
* handler is still needed for latch wakeups.
*/
pqsignal(SIGHUP, BgSigHupHandler); /* set flag to read config file */
pqsignal(SIGINT, SIG_IGN);
pqsignal(SIGTERM, ReqShutdownHandler); /* shutdown */
pqsignal(SIGQUIT, bg_quickdie); /* hard crash time */
pqsignal(SIGALRM, SIG_IGN);
pqsignal(SIGPIPE, SIG_IGN);
pqsignal(SIGUSR1, bgwriter_sigusr1_handler);
pqsignal(SIGUSR2, SIG_IGN);
/*
* Reset some signals that are accepted by postmaster but not here
*/
pqsignal(SIGCHLD, SIG_DFL);
pqsignal(SIGTTIN, SIG_DFL);
pqsignal(SIGTTOU, SIG_DFL);
pqsignal(SIGCONT, SIG_DFL);
pqsignal(SIGWINCH, SIG_DFL);
/* We allow SIGQUIT (quickdie) at all times */
sigdelset(&BlockSig, SIGQUIT);
/*
* Create a resource owner to keep track of our resources (currently only
* buffer pins).
*/
CurrentResourceOwner = ResourceOwnerCreate(NULL, "Background Writer");
/*
* We just started, assume there has been either a shutdown or
* end-of-recovery snapshot.
*/
last_snapshot_ts = GetCurrentTimestamp();
/*
* Create a memory context that we will do all our work in. We do this so
* that we can reset the context during error recovery and thereby avoid
* possible memory leaks. Formerly this code just ran in
* TopMemoryContext, but resetting that would be a really bad idea.
*/
bgwriter_context = AllocSetContextCreate(TopMemoryContext,
"Background Writer",
ALLOCSET_DEFAULT_MINSIZE,
ALLOCSET_DEFAULT_INITSIZE,
ALLOCSET_DEFAULT_MAXSIZE);
MemoryContextSwitchTo(bgwriter_context);
/*
* If an exception is encountered, processing resumes here.
*
* See notes in postgres.c about the design of this coding.
*/
if (sigsetjmp(local_sigjmp_buf, 1) != 0)
{
/* Since not using PG_TRY, must reset error stack by hand */
error_context_stack = NULL;
/* Prevent interrupts while cleaning up */
HOLD_INTERRUPTS();
/* Report the error to the server log */
EmitErrorReport();
/*
* These operations are really just a minimal subset of
* AbortTransaction(). We don't have very many resources to worry
* about in bgwriter, but we do have LWLocks, buffers, and temp files.
*/
LWLockReleaseAll();
AbortBufferIO();
UnlockBuffers();
/* buffer pins are released here: */
ResourceOwnerRelease(CurrentResourceOwner,
RESOURCE_RELEASE_BEFORE_LOCKS,
false, true);
/* we needn't bother with the other ResourceOwnerRelease phases */
AtEOXact_Buffers(false);
AtEOXact_SMgr();
AtEOXact_Files();
AtEOXact_HashTables(false);
/*
* Now return to normal top-level context and clear ErrorContext for
* next time.
*/
MemoryContextSwitchTo(bgwriter_context);
FlushErrorState();
/* Flush any leaked data in the top-level context */
MemoryContextResetAndDeleteChildren(bgwriter_context);
/* Now we can allow interrupts again */
RESUME_INTERRUPTS();
/*
* Sleep at least 1 second after any error. A write error is likely
* to be repeated, and we don't want to be filling the error logs as
* fast as we can.
*/
pg_usleep(1000000L);
/*
* Close all open files after any error. This is helpful on Windows,
* where holding deleted files open causes various strange errors.
* It's not clear we need it elsewhere, but shouldn't hurt.
*/
smgrcloseall();
/* Report wait end here, when there is no further possibility of wait */
pgstat_report_wait_end();
}
/* We can now handle ereport(ERROR) */
PG_exception_stack = &local_sigjmp_buf;
/*
* Unblock signals (they were blocked when the postmaster forked us)
*/
PG_SETMASK(&UnBlockSig);
/*
* Reset hibernation state after any error.
*/
prev_hibernate = false;
/*
* Loop forever
*/
for (;;)
{
bool can_hibernate;
int rc;
/* Clear any already-pending wakeups */
ResetLatch(MyLatch);
if (got_SIGHUP)
{
got_SIGHUP = false;
ProcessConfigFile(PGC_SIGHUP);
}
if (shutdown_requested)
{
/*
* From here on, elog(ERROR) should end with exit(1), not send
* control back to the sigsetjmp block above
*/
ExitOnAnyError = true;
/* Normal exit from the bgwriter is here */
proc_exit(0); /* done */
}
/*
* Do one cycle of dirty-buffer writing.
*/
can_hibernate = BgBufferSync();
/*
* Send off activity statistics to the stats collector
*/
pgstat_send_bgwriter();
if (FirstCallSinceLastCheckpoint())
{
/*
* After any checkpoint, close all smgr files. This is so we
* won't hang onto smgr references to deleted files indefinitely.
*/
smgrcloseall();
}
/*
* Log a new xl_running_xacts every now and then so replication can
* get into a consistent state faster (think of suboverflowed
* snapshots) and clean up resources (locks, KnownXids*) more
* frequently. The costs of this are relatively low, so doing it 4
* times (LOG_SNAPSHOT_INTERVAL_MS) a minute seems fine.
*
* We assume the interval for writing xl_running_xacts is
* significantly bigger than BgWriterDelay, so we don't complicate the
* overall timeout handling but just assume we're going to get called
* often enough even if hibernation mode is active. It's not that
* important that log_snap_interval_ms is met strictly. To make sure
* we're not waking the disk up unnecessarily on an idle system we
* check whether there has been any WAL inserted since the last time
* we've logged a running xacts.
*
* We do this logging in the bgwriter as its the only process that is
* run regularly and returns to its mainloop all the time. E.g.
* Checkpointer, when active, is barely ever in its mainloop and thus
* makes it hard to log regularly.
*/
if (XLogStandbyInfoActive() && !RecoveryInProgress())
{
TimestampTz timeout = 0;
TimestampTz now = GetCurrentTimestamp();
timeout = TimestampTzPlusMilliseconds(last_snapshot_ts,
LOG_SNAPSHOT_INTERVAL_MS);
/*
* only log if enough time has passed and some xlog record has
* been inserted.
*/
if (now >= timeout &&
last_snapshot_lsn != GetXLogInsertRecPtr())
{
last_snapshot_lsn = LogStandbySnapshot();
last_snapshot_ts = now;
}
}
/*
* Sleep until we are signaled or BgWriterDelay has elapsed.
*
* Note: the feedback control loop in BgBufferSync() expects that we
* will call it every BgWriterDelay msec. While it's not critical for
* correctness that that be exact, the feedback loop might misbehave
* if we stray too far from that. Hence, avoid loading this process
* down with latch events that are likely to happen frequently during
* normal operation.
*/
rc = WaitLatch(MyLatch,
WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH,
BgWriterDelay /* ms */ );
/*
* If no latch event and BgBufferSync says nothing's happening, extend
* the sleep in "hibernation" mode, where we sleep for much longer
* than bgwriter_delay says. Fewer wakeups save electricity. When a
* backend starts using buffers again, it will wake us up by setting
* our latch. Because the extra sleep will persist only as long as no
* buffer allocations happen, this should not distort the behavior of
* BgBufferSync's control loop too badly; essentially, it will think
* that the system-wide idle interval didn't exist.
*
* There is a race condition here, in that a backend might allocate a
* buffer between the time BgBufferSync saw the alloc count as zero
* and the time we call StrategyNotifyBgWriter. While it's not
* critical that we not hibernate anyway, we try to reduce the odds of
* that by only hibernating when BgBufferSync says nothing's happening
* for two consecutive cycles. Also, we mitigate any possible
* consequences of a missed wakeup by not hibernating forever.
*/
if (rc == WL_TIMEOUT && can_hibernate && prev_hibernate)
{
/* Ask for notification at next buffer allocation */
StrategyNotifyBgWriter(MyProc->pgprocno);
/* Sleep ... */
rc = WaitLatch(MyLatch,
WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH,
BgWriterDelay * HIBERNATE_FACTOR);
/* Reset the notification request in case we timed out */
StrategyNotifyBgWriter(-1);
}
/*
* Emergency bailout if postmaster has died. This is to avoid the
* necessity for manual cleanup of all postmaster children.
*/
if (rc & WL_POSTMASTER_DEATH)
exit(1);
prev_hibernate = can_hibernate;
}
}
/*
* ContinuousQueryWorkerStartup
*
* Launches a CQ worker, which continuously generates partial query results to send
* back to the combiner process.
*/
void
ContinuousQueryWorkerRun(Portal portal, ContinuousViewState *state, QueryDesc *queryDesc, ResourceOwner owner)
{
EState *estate = NULL;
DestReceiver *dest;
CmdType operation;
MemoryContext oldcontext;
int timeoutms = state->maxwaitms;
MemoryContext runcontext;
CQProcEntry *entry = GetCQProcEntry(MyCQId);
ResourceOwner cqowner = ResourceOwnerCreate(NULL, "CQResourceOwner");
bool savereadonly = XactReadOnly;
cq_stat_initialize(state->viewid, MyProcPid);
dest = CreateDestReceiver(DestCombiner);
SetCombinerDestReceiverParams(dest, MyCQId);
/* workers only need read-only transactions */
XactReadOnly = true;
runcontext = AllocSetContextCreate(TopMemoryContext, "CQRunContext",
ALLOCSET_DEFAULT_MINSIZE,
ALLOCSET_DEFAULT_INITSIZE,
ALLOCSET_DEFAULT_MAXSIZE);
elog(LOG, "\"%s\" worker %d running", queryDesc->plannedstmt->cq_target->relname, MyProcPid);
MarkWorkerAsRunning(MyCQId, MyWorkerId);
pgstat_report_activity(STATE_RUNNING, queryDesc->sourceText);
TupleBufferInitLatch(WorkerTupleBuffer, MyCQId, MyWorkerId, &MyProc->procLatch);
oldcontext = MemoryContextSwitchTo(runcontext);
retry:
PG_TRY();
{
bool xact_commit = true;
TimestampTz last_process = GetCurrentTimestamp();
TimestampTz last_commit = GetCurrentTimestamp();
start_executor(queryDesc, runcontext, cqowner);
CurrentResourceOwner = cqowner;
estate = queryDesc->estate;
operation = queryDesc->operation;
/*
* Initialize context that lives for the duration of a single iteration
* of the main worker loop
*/
CQExecutionContext = AllocSetContextCreate(estate->es_query_cxt, "CQExecutionContext",
ALLOCSET_DEFAULT_MINSIZE,
ALLOCSET_DEFAULT_INITSIZE,
ALLOCSET_DEFAULT_MAXSIZE);
estate->es_lastoid = InvalidOid;
/*
* Startup combiner receiver
*/
(*dest->rStartup) (dest, operation, queryDesc->tupDesc);
for (;;)
{
if (!TupleBufferHasUnreadSlots())
{
if (TimestampDifferenceExceeds(last_process, GetCurrentTimestamp(), state->emptysleepms))
{
/* force stats flush */
cq_stat_report(true);
pgstat_report_activity(STATE_IDLE, queryDesc->sourceText);
TupleBufferWait(WorkerTupleBuffer, MyCQId, MyWorkerId);
pgstat_report_activity(STATE_RUNNING, queryDesc->sourceText);
}
else
pg_usleep(Min(WAIT_SLEEP_MS, state->emptysleepms) * 1000);
}
TupleBufferResetNotify(WorkerTupleBuffer, MyCQId, MyWorkerId);
if (xact_commit)
StartTransactionCommand();
set_snapshot(estate, cqowner);
CurrentResourceOwner = cqowner;
MemoryContextSwitchTo(estate->es_query_cxt);
estate->es_processed = 0;
estate->es_filtered = 0;
/*
* Run plan on a microbatch
*/
ExecutePlan(estate, queryDesc->planstate, operation,
true, 0, timeoutms, ForwardScanDirection, dest);
IncrementCQExecutions(1);
TupleBufferClearPinnedSlots();
if (state->long_xact)
{
if (TimestampDifferenceExceeds(last_commit, GetCurrentTimestamp(), LONG_RUNNING_XACT_DURATION))
xact_commit = true;
else
xact_commit = false;
}
unset_snapshot(estate, cqowner);
if (xact_commit)
{
CommitTransactionCommand();
last_commit = GetCurrentTimestamp();
}
MemoryContextResetAndDeleteChildren(CQExecutionContext);
MemoryContextSwitchTo(runcontext);
CurrentResourceOwner = cqowner;
if (estate->es_processed || estate->es_filtered)
{
/*
* If the CV query is such that the select does not return any tuples
* ex: select id where id=99; and id=99 does not exist, then this reset
* will fail. What will happen is that the worker will block at the latch for every
* allocated slot, TILL a cv returns a non-zero tuple, at which point
* the worker will resume a simple sleep for the threshold time.
*/
last_process = GetCurrentTimestamp();
/*
* Send stats to the collector
*/
cq_stat_report(false);
}
/* Has the CQ been deactivated? */
if (!entry->active)
{
if (ActiveSnapshotSet())
unset_snapshot(estate, cqowner);
if (IsTransactionState())
CommitTransactionCommand();
break;
}
}
CurrentResourceOwner = cqowner;
/*
* The cleanup functions below expect these things to be registered
*/
RegisterSnapshotOnOwner(estate->es_snapshot, cqowner);
RegisterSnapshotOnOwner(queryDesc->snapshot, cqowner);
RegisterSnapshotOnOwner(queryDesc->crosscheck_snapshot, cqowner);
/* cleanup */
ExecutorFinish(queryDesc);
ExecutorEnd(queryDesc);
FreeQueryDesc(queryDesc);
}
PG_CATCH();
{
EmitErrorReport();
FlushErrorState();
/* Since the worker is read-only, we can simply commit the transaction. */
if (ActiveSnapshotSet())
unset_snapshot(estate, cqowner);
if (IsTransactionState())
CommitTransactionCommand();
TupleBufferUnpinAllPinnedSlots();
TupleBufferClearReaders();
/* This resets the es_query_ctx and in turn the CQExecutionContext */
MemoryContextResetAndDeleteChildren(runcontext);
IncrementCQErrors(1);
if (continuous_query_crash_recovery)
goto retry;
}
PG_END_TRY();
(*dest->rShutdown) (dest);
MemoryContextSwitchTo(oldcontext);
MemoryContextDelete(runcontext);
XactReadOnly = savereadonly;
/*
* Remove proc-level stats
*/
cq_stat_report(true);
cq_stat_send_purge(state->viewid, MyProcPid, CQ_STAT_WORKER);
CurrentResourceOwner = owner;
}
/*
* FtsProbeMain
*/
NON_EXEC_STATIC void
ftsMain(int argc, char *argv[])
{
sigjmp_buf local_sigjmp_buf;
char *fullpath;
IsUnderPostmaster = true;
am_ftsprobe = true;
/* Stay away from PMChildSlot */
MyPMChildSlot = -1;
/* reset MyProcPid */
MyProcPid = getpid();
/* Lose the postmaster's on-exit routines */
on_exit_reset();
/* Identify myself via ps */
init_ps_display("ftsprobe process", "", "", "");
SetProcessingMode(InitProcessing);
/*
* reread postgresql.conf if requested
*/
pqsignal(SIGHUP, sigHupHandler);
/*
* Presently, SIGINT will lead to autovacuum shutdown, because that's how
* we handle ereport(ERROR). It could be improved however.
*/
pqsignal(SIGINT, ReqFtsFullScan); /* request full-scan */
pqsignal(SIGTERM, die);
pqsignal(SIGQUIT, quickdie); /* we don't do any ftsprobe specific cleanup, just use the standard. */
pqsignal(SIGALRM, handle_sig_alarm);
pqsignal(SIGPIPE, SIG_IGN);
pqsignal(SIGUSR1, procsignal_sigusr1_handler);
/* We don't listen for async notifies */
pqsignal(SIGUSR2, RequestShutdown);
pqsignal(SIGFPE, FloatExceptionHandler);
pqsignal(SIGCHLD, SIG_DFL);
/*
* Copied from bgwriter
*/
CurrentResourceOwner = ResourceOwnerCreate(NULL, "FTS Probe");
/* Early initialization */
BaseInit();
/* See InitPostgres()... */
InitProcess();
InitBufferPoolBackend();
InitXLOGAccess();
SetProcessingMode(NormalProcessing);
/*
* If an exception is encountered, processing resumes here.
*
* See notes in postgres.c about the design of this coding.
*/
if (sigsetjmp(local_sigjmp_buf, 1) != 0)
{
/* Prevents interrupts while cleaning up */
HOLD_INTERRUPTS();
/* Report the error to the server log */
EmitErrorReport();
/*
* We can now go away. Note that because we'll call InitProcess, a
* callback will be registered to do ProcKill, which will clean up
* necessary state.
*/
proc_exit(0);
}
/* We can now handle ereport(ERROR) */
PG_exception_stack = &local_sigjmp_buf;
PG_SETMASK(&UnBlockSig);
/*
* Add my PGPROC struct to the ProcArray.
*
* Once I have done this, I am visible to other backends!
*/
InitProcessPhase2();
/*
* Initialize my entry in the shared-invalidation manager's array of
* per-backend data.
*
* Sets up MyBackendId, a unique backend identifier.
*/
MyBackendId = InvalidBackendId;
SharedInvalBackendInit(false);
if (MyBackendId > MaxBackends || MyBackendId <= 0)
elog(FATAL, "bad backend id: %d", MyBackendId);
/*
* bufmgr needs another initialization call too
*/
InitBufferPoolBackend();
/* heap access requires the rel-cache */
RelationCacheInitialize();
InitCatalogCache();
/*
* It's now possible to do real access to the system catalogs.
*
* Load relcache entries for the system catalogs. This must create at
* least the minimum set of "nailed-in" cache entries.
*/
RelationCacheInitializePhase2();
/*
* In order to access the catalog, we need a database, and a
* tablespace; our access to the heap is going to be slightly
* limited, so we'll just use some defaults.
*/
if (!FindMyDatabase(probeDatabase, &MyDatabaseId, &MyDatabaseTableSpace))
ereport(FATAL,
(errcode(ERRCODE_UNDEFINED_DATABASE),
errmsg("database \"%s\" does not exit", probeDatabase)));
/* Now we can mark our PGPROC entry with the database ID */
/* (We assume this is an atomic store so no lock is needed) */
MyProc->databaseId = MyDatabaseId;
fullpath = GetDatabasePath(MyDatabaseId, MyDatabaseTableSpace);
SetDatabasePath(fullpath);
RelationCacheInitializePhase3();
/* shmem: publish probe pid */
ftsProbeInfo->fts_probePid = MyProcPid;
/* main loop */
FtsLoop();
/* One iteration done, go away */
proc_exit(0);
}
