/*
* Load a single slot from disk into memory.
*/
static void
RestoreSlotFromDisk(const char *name)
{
ReplicationSlotOnDisk cp;
int i;
char path[MAXPGPATH + 22];
int fd;
bool restored = false;
int readBytes;
pg_crc32c checksum;
/* no need to lock here, no concurrent access allowed yet */
/* delete temp file if it exists */
sprintf(path, "pg_replslot/%s/state.tmp", name);
if (unlink(path) < 0 && errno != ENOENT)
ereport(PANIC,
(errcode_for_file_access(),
errmsg("could not remove file \"%s\": %m", path)));
sprintf(path, "pg_replslot/%s/state", name);
elog(DEBUG1, "restoring replication slot from \"%s\"", path);
fd = OpenTransientFile(path, O_RDWR | PG_BINARY);
/*
* We do not need to handle this as we are rename()ing the directory into
* place only after we fsync()ed the state file.
*/
if (fd < 0)
ereport(PANIC,
(errcode_for_file_access(),
errmsg("could not open file \"%s\": %m", path)));
/*
* Sync state file before we're reading from it. We might have crashed
* while it wasn't synced yet and we shouldn't continue on that basis.
*/
pgstat_report_wait_start(WAIT_EVENT_REPLICATION_SLOT_RESTORE_SYNC);
if (pg_fsync(fd) != 0)
{
CloseTransientFile(fd);
ereport(PANIC,
(errcode_for_file_access(),
errmsg("could not fsync file \"%s\": %m",
path)));
}
pgstat_report_wait_end();
/* Also sync the parent directory */
START_CRIT_SECTION();
fsync_fname(path, true);
END_CRIT_SECTION();
/* read part of statefile that's guaranteed to be version independent */
pgstat_report_wait_start(WAIT_EVENT_REPLICATION_SLOT_READ);
readBytes = read(fd, &cp, ReplicationSlotOnDiskConstantSize);
pgstat_report_wait_end();
if (readBytes != ReplicationSlotOnDiskConstantSize)
{
int saved_errno = errno;
CloseTransientFile(fd);
errno = saved_errno;
ereport(PANIC,
(errcode_for_file_access(),
errmsg("could not read file \"%s\", read %d of %u: %m",
path, readBytes,
(uint32) ReplicationSlotOnDiskConstantSize)));
}
/* verify magic */
if (cp.magic != SLOT_MAGIC)
ereport(PANIC,
(errcode_for_file_access(),
errmsg("replication slot file \"%s\" has wrong magic number: %u instead of %u",
path, cp.magic, SLOT_MAGIC)));
/* verify version */
if (cp.version != SLOT_VERSION)
ereport(PANIC,
(errcode_for_file_access(),
errmsg("replication slot file \"%s\" has unsupported version %u",
path, cp.version)));
/* boundary check on length */
if (cp.length != ReplicationSlotOnDiskV2Size)
ereport(PANIC,
(errcode_for_file_access(),
errmsg("replication slot file \"%s\" has corrupted length %u",
path, cp.length)));
/* Now that we know the size, read the entire file */
pgstat_report_wait_start(WAIT_EVENT_REPLICATION_SLOT_READ);
readBytes = read(fd,
(char *) &cp + ReplicationSlotOnDiskConstantSize,
cp.length);
pgstat_report_wait_end();
if (readBytes != cp.length)
{
int saved_errno = errno;
CloseTransientFile(fd);
errno = saved_errno;
ereport(PANIC,
(errcode_for_file_access(),
errmsg("could not read file \"%s\", read %d of %u: %m",
path, readBytes, cp.length)));
}
CloseTransientFile(fd);
/* now verify the CRC */
INIT_CRC32C(checksum);
COMP_CRC32C(checksum,
(char *) &cp + SnapBuildOnDiskNotChecksummedSize,
SnapBuildOnDiskChecksummedSize);
FIN_CRC32C(checksum);
if (!EQ_CRC32C(checksum, cp.checksum))
ereport(PANIC,
(errmsg("checksum mismatch for replication slot file \"%s\": is %u, should be %u",
path, checksum, cp.checksum)));
/*
* If we crashed with an ephemeral slot active, don't restore but delete
* it.
*/
if (cp.slotdata.persistency != RS_PERSISTENT)
{
sprintf(path, "pg_replslot/%s", name);
if (!rmtree(path, true))
{
ereport(WARNING,
(errcode_for_file_access(),
errmsg("could not remove directory \"%s\"", path)));
}
fsync_fname("pg_replslot", true);
return;
}
/* nothing can be active yet, don't lock anything */
for (i = 0; i < max_replication_slots; i++)
{
ReplicationSlot *slot;
slot = &ReplicationSlotCtl->replication_slots[i];
if (slot->in_use)
continue;
/* restore the entire set of persistent data */
memcpy(&slot->data, &cp.slotdata,
sizeof(ReplicationSlotPersistentData));
/* initialize in memory state */
slot->effective_xmin = cp.slotdata.xmin;
slot->effective_catalog_xmin = cp.slotdata.catalog_xmin;
slot->candidate_catalog_xmin = InvalidTransactionId;
slot->candidate_xmin_lsn = InvalidXLogRecPtr;
slot->candidate_restart_lsn = InvalidXLogRecPtr;
slot->candidate_restart_valid = InvalidXLogRecPtr;
slot->in_use = true;
slot->active_pid = 0;
restored = true;
break;
}
if (!restored)
ereport(PANIC,
(errmsg("too many replication slots active before shutdown"),
errhint("Increase max_replication_slots and try again.")));
}
/*
* 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);
}
}
/*
* Physical write of a page from a buffer slot
*
* On failure, we cannot just ereport(ERROR) since caller has put state in
* shared memory that must be undone. So, we return FALSE and save enough
* info in static variables to let SlruReportIOError make the report.
*
* For now, assume it's not worth keeping a file pointer open across
* independent read/write operations. We do batch operations during
* SimpleLruFlush, though.
*
* fdata is NULL for a standalone write, pointer to open-file info during
* SimpleLruFlush.
*/
static bool
SlruPhysicalWritePage(SlruCtl ctl, int pageno, int slotno, SlruFlush fdata)
{
SlruShared shared = ctl->shared;
int segno = pageno / SLRU_PAGES_PER_SEGMENT;
int rpageno = pageno % SLRU_PAGES_PER_SEGMENT;
int offset = rpageno * BLCKSZ;
char path[MAXPGPATH];
int fd = -1;
/*
* Honor the write-WAL-before-data rule, if appropriate, so that we do not
* write out data before associated WAL records. This is the same action
* performed during FlushBuffer() in the main buffer manager.
*/
if (shared->group_lsn != NULL)
{
/*
* We must determine the largest async-commit LSN for the page. This
* is a bit tedious, but since this entire function is a slow path
* anyway, it seems better to do this here than to maintain a per-page
* LSN variable (which'd need an extra comparison in the
* transaction-commit path).
*/
XLogRecPtr max_lsn;
int lsnindex,
lsnoff;
lsnindex = slotno * shared->lsn_groups_per_page;
max_lsn = shared->group_lsn[lsnindex++];
for (lsnoff = 1; lsnoff < shared->lsn_groups_per_page; lsnoff++)
{
XLogRecPtr this_lsn = shared->group_lsn[lsnindex++];
if (max_lsn < this_lsn)
max_lsn = this_lsn;
}
if (!XLogRecPtrIsInvalid(max_lsn))
{
/*
* As noted above, elog(ERROR) is not acceptable here, so if
* XLogFlush were to fail, we must PANIC. This isn't much of a
* restriction because XLogFlush is just about all critical
* section anyway, but let's make sure.
*/
START_CRIT_SECTION();
XLogFlush(max_lsn);
END_CRIT_SECTION();
}
}
/*
* During a Flush, we may already have the desired file open.
*/
if (fdata)
{
int i;
for (i = 0; i < fdata->num_files; i++)
{
if (fdata->segno[i] == segno)
{
fd = fdata->fd[i];
break;
}
}
}
if (fd < 0)
{
/*
* If the file doesn't already exist, we should create it. It is
* possible for this to need to happen when writing a page that's not
* first in its segment; we assume the OS can cope with that. (Note:
* it might seem that it'd be okay to create files only when
* SimpleLruZeroPage is called for the first page of a segment.
* However, if after a crash and restart the REDO logic elects to
* replay the log from a checkpoint before the latest one, then it's
* possible that we will get commands to set transaction status of
* transactions that have already been truncated from the commit log.
* Easiest way to deal with that is to accept references to
* nonexistent files here and in SlruPhysicalReadPage.)
*
* Note: it is possible for more than one backend to be executing this
* code simultaneously for different pages of the same file. Hence,
* don't use O_EXCL or O_TRUNC or anything like that.
*/
SlruFileName(ctl, path, segno);
fd = OpenTransientFile(path, O_RDWR | O_CREAT | PG_BINARY,
S_IRUSR | S_IWUSR);
if (fd < 0)
{
slru_errcause = SLRU_OPEN_FAILED;
slru_errno = errno;
return false;
}
if (fdata)
{
if (fdata->num_files < MAX_FLUSH_BUFFERS)
{
fdata->fd[fdata->num_files] = fd;
fdata->segno[fdata->num_files] = segno;
fdata->num_files++;
}
else
{
/*
* In the unlikely event that we exceed MAX_FLUSH_BUFFERS,
* fall back to treating it as a standalone write.
*/
fdata = NULL;
}
}
}
if (lseek(fd, (off_t) offset, SEEK_SET) < 0)
{
slru_errcause = SLRU_SEEK_FAILED;
slru_errno = errno;
if (!fdata)
CloseTransientFile(fd);
return false;
}
errno = 0;
pgstat_report_wait_start(WAIT_EVENT_SLRU_WRITE);
if (write(fd, shared->page_buffer[slotno], BLCKSZ) != BLCKSZ)
{
pgstat_report_wait_end();
/* if write didn't set errno, assume problem is no disk space */
if (errno == 0)
errno = ENOSPC;
slru_errcause = SLRU_WRITE_FAILED;
slru_errno = errno;
if (!fdata)
CloseTransientFile(fd);
return false;
}
pgstat_report_wait_end();
/*
* If not part of Flush, need to fsync now. We assume this happens
* infrequently enough that it's not a performance issue.
*/
if (!fdata)
{
pgstat_report_wait_start(WAIT_EVENT_SLRU_SYNC);
if (ctl->do_fsync && pg_fsync(fd))
{
pgstat_report_wait_end();
slru_errcause = SLRU_FSYNC_FAILED;
slru_errno = errno;
CloseTransientFile(fd);
return false;
}
pgstat_report_wait_end();
if (CloseTransientFile(fd))
{
slru_errcause = SLRU_CLOSE_FAILED;
slru_errno = errno;
return false;
}
}
return true;
}
/*
* Shared functionality between saving and creating a replication slot.
*/
static void
SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel)
{
char tmppath[MAXPGPATH];
char path[MAXPGPATH];
int fd;
ReplicationSlotOnDisk cp;
bool was_dirty;
/* first check whether there's something to write out */
SpinLockAcquire(&slot->mutex);
was_dirty = slot->dirty;
slot->just_dirtied = false;
SpinLockRelease(&slot->mutex);
/* and don't do anything if there's nothing to write */
if (!was_dirty)
return;
LWLockAcquire(&slot->io_in_progress_lock, LW_EXCLUSIVE);
/* silence valgrind :( */
memset(&cp, 0, sizeof(ReplicationSlotOnDisk));
sprintf(tmppath, "%s/state.tmp", dir);
sprintf(path, "%s/state", dir);
fd = OpenTransientFile(tmppath, O_CREAT | O_EXCL | O_WRONLY | PG_BINARY);
if (fd < 0)
{
ereport(elevel,
(errcode_for_file_access(),
errmsg("could not create file \"%s\": %m",
tmppath)));
return;
}
cp.magic = SLOT_MAGIC;
INIT_CRC32C(cp.checksum);
cp.version = SLOT_VERSION;
cp.length = ReplicationSlotOnDiskV2Size;
SpinLockAcquire(&slot->mutex);
memcpy(&cp.slotdata, &slot->data, sizeof(ReplicationSlotPersistentData));
SpinLockRelease(&slot->mutex);
COMP_CRC32C(cp.checksum,
(char *) (&cp) + SnapBuildOnDiskNotChecksummedSize,
SnapBuildOnDiskChecksummedSize);
FIN_CRC32C(cp.checksum);
pgstat_report_wait_start(WAIT_EVENT_REPLICATION_SLOT_WRITE);
if ((write(fd, &cp, sizeof(cp))) != sizeof(cp))
{
int save_errno = errno;
pgstat_report_wait_end();
CloseTransientFile(fd);
errno = save_errno;
ereport(elevel,
(errcode_for_file_access(),
errmsg("could not write to file \"%s\": %m",
tmppath)));
return;
}
pgstat_report_wait_end();
/* fsync the temporary file */
pgstat_report_wait_start(WAIT_EVENT_REPLICATION_SLOT_SYNC);
if (pg_fsync(fd) != 0)
{
int save_errno = errno;
pgstat_report_wait_end();
CloseTransientFile(fd);
errno = save_errno;
ereport(elevel,
(errcode_for_file_access(),
errmsg("could not fsync file \"%s\": %m",
tmppath)));
return;
}
pgstat_report_wait_end();
CloseTransientFile(fd);
/* rename to permanent file, fsync file and directory */
if (rename(tmppath, path) != 0)
{
ereport(elevel,
(errcode_for_file_access(),
errmsg("could not rename file \"%s\" to \"%s\": %m",
tmppath, path)));
return;
}
/* Check CreateSlot() for the reasoning of using a crit. section. */
START_CRIT_SECTION();
fsync_fname(path, false);
fsync_fname(dir, true);
fsync_fname("pg_replslot", true);
END_CRIT_SECTION();
/*
* Successfully wrote, unset dirty bit, unless somebody dirtied again
* already.
*/
SpinLockAcquire(&slot->mutex);
if (!slot->just_dirtied)
slot->dirty = false;
SpinLockRelease(&slot->mutex);
LWLockRelease(&slot->io_in_progress_lock);
}
/*
* Flush dirty pages to disk during checkpoint or database shutdown
*/
void
SimpleLruFlush(SlruCtl ctl, bool allow_redirtied)
{
SlruShared shared = ctl->shared;
SlruFlushData fdata;
int slotno;
int pageno = 0;
int i;
bool ok;
/*
* Find and write dirty pages
*/
fdata.num_files = 0;
LWLockAcquire(shared->ControlLock, LW_EXCLUSIVE);
for (slotno = 0; slotno < shared->num_slots; slotno++)
{
SlruInternalWritePage(ctl, slotno, &fdata);
/*
* In some places (e.g. checkpoints), we cannot assert that the slot
* is clean now, since another process might have re-dirtied it
* already. That's okay.
*/
Assert(allow_redirtied ||
shared->page_status[slotno] == SLRU_PAGE_EMPTY ||
(shared->page_status[slotno] == SLRU_PAGE_VALID &&
!shared->page_dirty[slotno]));
}
LWLockRelease(shared->ControlLock);
/*
* Now fsync and close any files that were open
*/
ok = true;
for (i = 0; i < fdata.num_files; i++)
{
pgstat_report_wait_start(WAIT_EVENT_SLRU_FLUSH_SYNC);
if (ctl->do_fsync && pg_fsync(fdata.fd[i]))
{
slru_errcause = SLRU_FSYNC_FAILED;
slru_errno = errno;
pageno = fdata.segno[i] * SLRU_PAGES_PER_SEGMENT;
ok = false;
}
pgstat_report_wait_end();
if (CloseTransientFile(fdata.fd[i]))
{
slru_errcause = SLRU_CLOSE_FAILED;
slru_errno = errno;
pageno = fdata.segno[i] * SLRU_PAGES_PER_SEGMENT;
ok = false;
}
}
if (!ok)
SlruReportIOError(ctl, pageno, InvalidTransactionId);
}
/*
* Physical read of a (previously existing) page into a buffer slot
*
* On failure, we cannot just ereport(ERROR) since caller has put state in
* shared memory that must be undone. So, we return FALSE and save enough
* info in static variables to let SlruReportIOError make the report.
*
* For now, assume it's not worth keeping a file pointer open across
* read/write operations. We could cache one virtual file pointer ...
*/
static bool
SlruPhysicalReadPage(SlruCtl ctl, int pageno, int slotno)
{
SlruShared shared = ctl->shared;
int segno = pageno / SLRU_PAGES_PER_SEGMENT;
int rpageno = pageno % SLRU_PAGES_PER_SEGMENT;
int offset = rpageno * BLCKSZ;
char path[MAXPGPATH];
int fd;
SlruFileName(ctl, path, segno);
/*
* In a crash-and-restart situation, it's possible for us to receive
* commands to set the commit status of transactions whose bits are in
* already-truncated segments of the commit log (see notes in
* SlruPhysicalWritePage). Hence, if we are InRecovery, allow the case
* where the file doesn't exist, and return zeroes instead.
*/
fd = OpenTransientFile(path, O_RDWR | PG_BINARY, S_IRUSR | S_IWUSR);
if (fd < 0)
{
if (errno != ENOENT || !InRecovery)
{
slru_errcause = SLRU_OPEN_FAILED;
slru_errno = errno;
return false;
}
ereport(LOG,
(errmsg("file \"%s\" doesn't exist, reading as zeroes",
path)));
MemSet(shared->page_buffer[slotno], 0, BLCKSZ);
return true;
}
if (lseek(fd, (off_t) offset, SEEK_SET) < 0)
{
slru_errcause = SLRU_SEEK_FAILED;
slru_errno = errno;
CloseTransientFile(fd);
return false;
}
errno = 0;
pgstat_report_wait_start(WAIT_EVENT_SLRU_READ);
if (read(fd, shared->page_buffer[slotno], BLCKSZ) != BLCKSZ)
{
pgstat_report_wait_end();
slru_errcause = SLRU_READ_FAILED;
slru_errno = errno;
CloseTransientFile(fd);
return false;
}
pgstat_report_wait_end();
if (CloseTransientFile(fd))
{
slru_errcause = SLRU_CLOSE_FAILED;
slru_errno = errno;
return false;
}
return true;
}
/*
* 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;
}
}
/*
* copy one file
*/
void
copy_file(char *fromfile, char *tofile)
{
char *buffer;
int srcfd;
int dstfd;
int nbytes;
off_t offset;
/* Use palloc to ensure we get a maxaligned buffer */
#define COPY_BUF_SIZE (8 * BLCKSZ)
buffer = palloc(COPY_BUF_SIZE);
/*
* Open the files
*/
srcfd = OpenTransientFile(fromfile, O_RDONLY | PG_BINARY);
if (srcfd < 0)
ereport(ERROR,
(errcode_for_file_access(),
errmsg("could not open file \"%s\": %m", fromfile)));
dstfd = OpenTransientFile(tofile, O_RDWR | O_CREAT | O_EXCL | PG_BINARY);
if (dstfd < 0)
ereport(ERROR,
(errcode_for_file_access(),
errmsg("could not create file \"%s\": %m", tofile)));
/*
* Do the data copying.
*/
for (offset = 0;; offset += nbytes)
{
/* If we got a cancel signal during the copy of the file, quit */
CHECK_FOR_INTERRUPTS();
pgstat_report_wait_start(WAIT_EVENT_COPY_FILE_READ);
nbytes = read(srcfd, buffer, COPY_BUF_SIZE);
pgstat_report_wait_end();
if (nbytes < 0)
ereport(ERROR,
(errcode_for_file_access(),
errmsg("could not read file \"%s\": %m", fromfile)));
if (nbytes == 0)
break;
errno = 0;
pgstat_report_wait_start(WAIT_EVENT_COPY_FILE_WRITE);
if ((int) write(dstfd, buffer, nbytes) != nbytes)
{
pgstat_report_wait_end();
/* if write didn't set errno, assume problem is no disk space */
if (errno == 0)
errno = ENOSPC;
ereport(ERROR,
(errcode_for_file_access(),
errmsg("could not write to file \"%s\": %m", tofile)));
}
pgstat_report_wait_end();
/*
* We fsync the files later but first flush them to avoid spamming the
* cache and hopefully get the kernel to start writing them out before
* the fsync comes.
*/
pg_flush_data(dstfd, offset, nbytes);
}
if (CloseTransientFile(dstfd))
ereport(ERROR,
(errcode_for_file_access(),
errmsg("could not close file \"%s\": %m", tofile)));
CloseTransientFile(srcfd);
pfree(buffer);
}
/*
* Write out a new shared or local map file with the given contents.
*
* The magic number and CRC are automatically updated in *newmap. On
* success, we copy the data to the appropriate permanent static variable.
*
* If write_wal is TRUE then an appropriate WAL message is emitted.
* (It will be false for bootstrap and WAL replay cases.)
*
* If send_sinval is TRUE then a SI invalidation message is sent.
* (This should be true except in bootstrap case.)
*
* If preserve_files is TRUE then the storage manager is warned not to
* delete the files listed in the map.
*
* Because this may be called during WAL replay when MyDatabaseId,
* DatabasePath, etc aren't valid, we require the caller to pass in suitable
* values. The caller is also responsible for being sure no concurrent
* map update could be happening.
*/
static void
write_relmap_file(bool shared, RelMapFile *newmap,
bool write_wal, bool send_sinval, bool preserve_files,
Oid dbid, Oid tsid, const char *dbpath)
{
int fd;
RelMapFile *realmap;
char mapfilename[MAXPGPATH];
/*
* Fill in the overhead fields and update CRC.
*/
newmap->magic = RELMAPPER_FILEMAGIC;
if (newmap->num_mappings < 0 || newmap->num_mappings > MAX_MAPPINGS)
elog(ERROR, "attempt to write bogus relation mapping");
INIT_CRC32C(newmap->crc);
COMP_CRC32C(newmap->crc, (char *) newmap, offsetof(RelMapFile, crc));
FIN_CRC32C(newmap->crc);
/*
* Open the target file. We prefer to do this before entering the
* critical section, so that an open() failure need not force PANIC.
*/
if (shared)
{
snprintf(mapfilename, sizeof(mapfilename), "global/%s",
RELMAPPER_FILENAME);
realmap = &shared_map;
}
else
{
snprintf(mapfilename, sizeof(mapfilename), "%s/%s",
dbpath, RELMAPPER_FILENAME);
realmap = &local_map;
}
fd = OpenTransientFile(mapfilename,
O_WRONLY | O_CREAT | PG_BINARY,
S_IRUSR | S_IWUSR);
if (fd < 0)
ereport(ERROR,
(errcode_for_file_access(),
errmsg("could not open relation mapping file \"%s\": %m",
mapfilename)));
if (write_wal)
{
xl_relmap_update xlrec;
XLogRecPtr lsn;
/* now errors are fatal ... */
START_CRIT_SECTION();
xlrec.dbid = dbid;
xlrec.tsid = tsid;
xlrec.nbytes = sizeof(RelMapFile);
XLogBeginInsert();
XLogRegisterData((char *) (&xlrec), MinSizeOfRelmapUpdate);
XLogRegisterData((char *) newmap, sizeof(RelMapFile));
lsn = XLogInsert(RM_RELMAP_ID, XLOG_RELMAP_UPDATE);
/* As always, WAL must hit the disk before the data update does */
XLogFlush(lsn);
}
errno = 0;
pgstat_report_wait_start(WAIT_EVENT_RELATION_MAP_WRITE);
if (write(fd, newmap, sizeof(RelMapFile)) != sizeof(RelMapFile))
{
/* if write didn't set errno, assume problem is no disk space */
if (errno == 0)
errno = ENOSPC;
ereport(ERROR,
(errcode_for_file_access(),
errmsg("could not write to relation mapping file \"%s\": %m",
mapfilename)));
}
pgstat_report_wait_end();
/*
* We choose to fsync the data to disk before considering the task done.
* It would be possible to relax this if it turns out to be a performance
* issue, but it would complicate checkpointing --- see notes for
* CheckPointRelationMap.
*/
pgstat_report_wait_start(WAIT_EVENT_RELATION_MAP_SYNC);
if (pg_fsync(fd) != 0)
ereport(ERROR,
(errcode_for_file_access(),
errmsg("could not fsync relation mapping file \"%s\": %m",
mapfilename)));
pgstat_report_wait_end();
if (CloseTransientFile(fd))
ereport(ERROR,
(errcode_for_file_access(),
errmsg("could not close relation mapping file \"%s\": %m",
mapfilename)));
/*
* Now that the file is safely on disk, send sinval message to let other
* backends know to re-read it. We must do this inside the critical
* section: if for some reason we fail to send the message, we have to
* force a database-wide PANIC. Otherwise other backends might continue
* execution with stale mapping information, which would be catastrophic
* as soon as others began to use the now-committed data.
*/
if (send_sinval)
CacheInvalidateRelmap(dbid);
/*
* Make sure that the files listed in the map are not deleted if the outer
* transaction aborts. This had better be within the critical section
* too: it's not likely to fail, but if it did, we'd arrive at transaction
* abort with the files still vulnerable. PANICing will leave things in a
* good state on-disk.
*
* Note: we're cheating a little bit here by assuming that mapped files
* are either in pg_global or the database's default tablespace.
*/
if (preserve_files)
{
int32 i;
for (i = 0; i < newmap->num_mappings; i++)
{
RelFileNode rnode;
rnode.spcNode = tsid;
rnode.dbNode = dbid;
rnode.relNode = newmap->mappings[i].mapfilenode;
RelationPreserveStorage(rnode, false);
}
}
/* Success, update permanent copy */
memcpy(realmap, newmap, sizeof(RelMapFile));
/* Critical section done */
if (write_wal)
END_CRIT_SECTION();
}
/*
* load_relmap_file -- load data from the shared or local map file
*
* Because the map file is essential for access to core system catalogs,
* failure to read it is a fatal error.
*
* Note that the local case requires DatabasePath to be set up.
*/
static void
load_relmap_file(bool shared)
{
RelMapFile *map;
char mapfilename[MAXPGPATH];
pg_crc32c crc;
int fd;
if (shared)
{
snprintf(mapfilename, sizeof(mapfilename), "global/%s",
RELMAPPER_FILENAME);
map = &shared_map;
}
else
{
snprintf(mapfilename, sizeof(mapfilename), "%s/%s",
DatabasePath, RELMAPPER_FILENAME);
map = &local_map;
}
/* Read data ... */
fd = OpenTransientFile(mapfilename,
O_RDONLY | PG_BINARY, S_IRUSR | S_IWUSR);
if (fd < 0)
ereport(FATAL,
(errcode_for_file_access(),
errmsg("could not open relation mapping file \"%s\": %m",
mapfilename)));
/*
* Note: we could take RelationMappingLock in shared mode here, but it
* seems unnecessary since our read() should be atomic against any
* concurrent updater's write(). If the file is updated shortly after we
* look, the sinval signaling mechanism will make us re-read it before we
* are able to access any relation that's affected by the change.
*/
pgstat_report_wait_start(WAIT_EVENT_RELATION_MAP_READ);
if (read(fd, map, sizeof(RelMapFile)) != sizeof(RelMapFile))
ereport(FATAL,
(errcode_for_file_access(),
errmsg("could not read relation mapping file \"%s\": %m",
mapfilename)));
pgstat_report_wait_end();
CloseTransientFile(fd);
/* check for correct magic number, etc */
if (map->magic != RELMAPPER_FILEMAGIC ||
map->num_mappings < 0 ||
map->num_mappings > MAX_MAPPINGS)
ereport(FATAL,
(errmsg("relation mapping file \"%s\" contains invalid data",
mapfilename)));
/* verify the CRC */
INIT_CRC32C(crc);
COMP_CRC32C(crc, (char *) map, offsetof(RelMapFile, crc));
FIN_CRC32C(crc);
if (!EQ_CRC32C(crc, map->crc))
ereport(FATAL,
(errmsg("relation mapping file \"%s\" contains incorrect checksum",
mapfilename)));
}
/*
* 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);
}
}
/*
* Load a single slot from disk into memory.
*/
static void
RestoreSlotFromDisk(const char *name)
{
ReplicationSlotOnDisk cp;
int i;
char slotdir[MAXPGPATH + 12];
char path[MAXPGPATH + 22];
int fd;
bool restored = false;
int readBytes;
pg_crc32c checksum;
/* no need to lock here, no concurrent access allowed yet */
/* delete temp file if it exists */
sprintf(slotdir, "pg_replslot/%s", name);
sprintf(path, "%s/state.tmp", slotdir);
if (unlink(path) < 0 && errno != ENOENT)
ereport(PANIC,
(errcode_for_file_access(),
errmsg("could not remove file \"%s\": %m", path)));
sprintf(path, "%s/state", slotdir);
elog(DEBUG1, "restoring replication slot from \"%s\"", path);
fd = OpenTransientFile(path, O_RDONLY | PG_BINARY);
/*
* We do not need to handle this as we are rename()ing the directory into
* place only after we fsync()ed the state file.
*/
if (fd < 0)
ereport(PANIC,
(errcode_for_file_access(),
errmsg("could not open file \"%s\": %m", path)));
/*
* Sync state file before we're reading from it. We might have crashed
* while it wasn't synced yet and we shouldn't continue on that basis.
*/
pgstat_report_wait_start(WAIT_EVENT_REPLICATION_SLOT_RESTORE_SYNC);
if (pg_fsync(fd) != 0)
ereport(PANIC,
(errcode_for_file_access(),
errmsg("could not fsync file \"%s\": %m",
path)));
pgstat_report_wait_end();
/* Also sync the parent directory */
START_CRIT_SECTION();
fsync_fname(slotdir, true);
END_CRIT_SECTION();
/* read part of statefile that's guaranteed to be version independent */
pgstat_report_wait_start(WAIT_EVENT_REPLICATION_SLOT_READ);
readBytes = read(fd, &cp, ReplicationSlotOnDiskConstantSize);
pgstat_report_wait_end();
if (readBytes != ReplicationSlotOnDiskConstantSize)
{
if (readBytes < 0)
ereport(PANIC,
(errcode_for_file_access(),
errmsg("could not read file \"%s\": %m", path)));
else
ereport(PANIC,
(errcode(ERRCODE_DATA_CORRUPTED),
errmsg("could not read file \"%s\": read %d of %zu",
path, readBytes,
(Size) ReplicationSlotOnDiskConstantSize)));
}
/* verify magic */
if (cp.magic != SLOT_MAGIC)
ereport(PANIC,
(errcode(ERRCODE_DATA_CORRUPTED),
errmsg("replication slot file \"%s\" has wrong magic number: %u instead of %u",
path, cp.magic, SLOT_MAGIC)));
/* verify version */
if (cp.version != SLOT_VERSION)
ereport(PANIC,
(errcode(ERRCODE_DATA_CORRUPTED),
errmsg("replication slot file \"%s\" has unsupported version %u",
path, cp.version)));
/* boundary check on length */
if (cp.length != ReplicationSlotOnDiskV2Size)
ereport(PANIC,
(errcode(ERRCODE_DATA_CORRUPTED),
errmsg("replication slot file \"%s\" has corrupted length %u",
path, cp.length)));
/* Now that we know the size, read the entire file */
pgstat_report_wait_start(WAIT_EVENT_REPLICATION_SLOT_READ);
readBytes = read(fd,
(char *) &cp + ReplicationSlotOnDiskConstantSize,
cp.length);
pgstat_report_wait_end();
if (readBytes != cp.length)
{
if (readBytes < 0)
ereport(PANIC,
(errcode_for_file_access(),
errmsg("could not read file \"%s\": %m", path)));
else
ereport(PANIC,
(errcode(ERRCODE_DATA_CORRUPTED),
errmsg("could not read file \"%s\": read %d of %zu",
path, readBytes, (Size) cp.length)));
}
if (CloseTransientFile(fd))
ereport(PANIC,
(errcode_for_file_access(),
errmsg("could not close file \"%s\": %m", path)));
/* now verify the CRC */
INIT_CRC32C(checksum);
COMP_CRC32C(checksum,
(char *) &cp + SnapBuildOnDiskNotChecksummedSize,
SnapBuildOnDiskChecksummedSize);
FIN_CRC32C(checksum);
if (!EQ_CRC32C(checksum, cp.checksum))
ereport(PANIC,
(errmsg("checksum mismatch for replication slot file \"%s\": is %u, should be %u",
path, checksum, cp.checksum)));
/*
* If we crashed with an ephemeral slot active, don't restore but delete
* it.
*/
if (cp.slotdata.persistency != RS_PERSISTENT)
{
if (!rmtree(slotdir, true))
{
ereport(WARNING,
(errmsg("could not remove directory \"%s\"",
slotdir)));
}
fsync_fname("pg_replslot", true);
return;
}
/*
* Verify that requirements for the specific slot type are met. That's
* important because if these aren't met we're not guaranteed to retain
* all the necessary resources for the slot.
*
* NB: We have to do so *after* the above checks for ephemeral slots,
* because otherwise a slot that shouldn't exist anymore could prevent
* restarts.
*
* NB: Changing the requirements here also requires adapting
* CheckSlotRequirements() and CheckLogicalDecodingRequirements().
*/
if (cp.slotdata.database != InvalidOid && wal_level < WAL_LEVEL_LOGICAL)
ereport(FATAL,
(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
errmsg("logical replication slot \"%s\" exists, but wal_level < logical",
NameStr(cp.slotdata.name)),
errhint("Change wal_level to be logical or higher.")));
else if (wal_level < WAL_LEVEL_REPLICA)
ereport(FATAL,
(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
errmsg("physical replication slot \"%s\" exists, but wal_level < replica",
NameStr(cp.slotdata.name)),
errhint("Change wal_level to be replica or higher.")));
/* nothing can be active yet, don't lock anything */
for (i = 0; i < max_replication_slots; i++)
{
ReplicationSlot *slot;
slot = &ReplicationSlotCtl->replication_slots[i];
if (slot->in_use)
continue;
/* restore the entire set of persistent data */
memcpy(&slot->data, &cp.slotdata,
sizeof(ReplicationSlotPersistentData));
/* initialize in memory state */
slot->effective_xmin = cp.slotdata.xmin;
slot->effective_catalog_xmin = cp.slotdata.catalog_xmin;
slot->candidate_catalog_xmin = InvalidTransactionId;
slot->candidate_xmin_lsn = InvalidXLogRecPtr;
slot->candidate_restart_lsn = InvalidXLogRecPtr;
slot->candidate_restart_valid = InvalidXLogRecPtr;
slot->in_use = true;
slot->active_pid = 0;
restored = true;
break;
}
if (!restored)
ereport(FATAL,
(errmsg("too many replication slots active before shutdown"),
errhint("Increase max_replication_slots and try again.")));
}
/*
* copy one file
*/
void
copy_file(char *fromfile, char *tofile)
{
char *buffer;
int srcfd;
int dstfd;
int nbytes;
off_t offset;
off_t flush_offset;
/* Size of copy buffer (read and write requests) */
#define COPY_BUF_SIZE (8 * BLCKSZ)
/*
* Size of data flush requests. It seems beneficial on most platforms to
* do this every 1MB or so. But macOS, at least with early releases of
* APFS, is really unfriendly to small mmap/msync requests, so there do it
* only every 32MB.
*/
#if defined(__darwin__)
#define FLUSH_DISTANCE (32 * 1024 * 1024)
#else
#define FLUSH_DISTANCE (1024 * 1024)
#endif
/* Use palloc to ensure we get a maxaligned buffer */
buffer = palloc(COPY_BUF_SIZE);
/*
* Open the files
*/
srcfd = OpenTransientFile(fromfile, O_RDONLY | PG_BINARY);
if (srcfd < 0)
ereport(ERROR,
(errcode_for_file_access(),
errmsg("could not open file \"%s\": %m", fromfile)));
dstfd = OpenTransientFile(tofile, O_RDWR | O_CREAT | O_EXCL | PG_BINARY);
if (dstfd < 0)
ereport(ERROR,
(errcode_for_file_access(),
errmsg("could not create file \"%s\": %m", tofile)));
/*
* Do the data copying.
*/
flush_offset = 0;
for (offset = 0;; offset += nbytes)
{
/* If we got a cancel signal during the copy of the file, quit */
CHECK_FOR_INTERRUPTS();
/*
* We fsync the files later, but during the copy, flush them every so
* often to avoid spamming the cache and hopefully get the kernel to
* start writing them out before the fsync comes.
*/
if (offset - flush_offset >= FLUSH_DISTANCE)
{
pg_flush_data(dstfd, flush_offset, offset - flush_offset);
flush_offset = offset;
}
pgstat_report_wait_start(WAIT_EVENT_COPY_FILE_READ);
nbytes = read(srcfd, buffer, COPY_BUF_SIZE);
pgstat_report_wait_end();
if (nbytes < 0)
ereport(ERROR,
(errcode_for_file_access(),
errmsg("could not read file \"%s\": %m", fromfile)));
if (nbytes == 0)
break;
errno = 0;
pgstat_report_wait_start(WAIT_EVENT_COPY_FILE_WRITE);
if ((int) write(dstfd, buffer, nbytes) != nbytes)
{
/* if write didn't set errno, assume problem is no disk space */
if (errno == 0)
errno = ENOSPC;
ereport(ERROR,
(errcode_for_file_access(),
errmsg("could not write to file \"%s\": %m", tofile)));
}
pgstat_report_wait_end();
}
if (offset > flush_offset)
pg_flush_data(dstfd, flush_offset, offset - flush_offset);
if (CloseTransientFile(dstfd))
ereport(ERROR,
(errcode_for_file_access(),
errmsg("could not close file \"%s\": %m", tofile)));
if (CloseTransientFile(srcfd))
ereport(ERROR,
(errcode_for_file_access(),
errmsg("could not close file \"%s\": %m", fromfile)));
pfree(buffer);
}