/*
* Determine the cutoff time at which we want to start canceling conflicting
* transactions. Returns zero (a time safely in the past) if we are willing
* to wait forever.
*/
static TimestampTz
GetStandbyLimitTime(void)
{
TimestampTz rtime;
bool fromStream;
/*
* The cutoff time is the last WAL data receipt time plus the appropriate
* delay variable. Delay of -1 means wait forever.
*/
GetXLogReceiptTime(&rtime, &fromStream);
if (fromStream)
{
if (max_standby_streaming_delay < 0)
return 0; /* wait forever */
return TimestampTzPlusMilliseconds(rtime, max_standby_streaming_delay);
}
else
{
if (max_standby_archive_delay < 0)
return 0; /* wait forever */
return TimestampTzPlusMilliseconds(rtime, max_standby_archive_delay);
}
}
/*
* Enable the specified timeout to fire after the specified delay.
*
* Delay is given in milliseconds.
*/
void
enable_timeout_after(TimeoutId id, int delay_ms)
{
TimestampTz now;
TimestampTz fin_time;
/* Disable timeout interrupts for safety. */
disable_alarm();
/* Queue the timeout at the appropriate time. */
now = GetCurrentTimestamp();
fin_time = TimestampTzPlusMilliseconds(now, delay_ms);
enable_timeout(id, now, fin_time);
/* Set the timer interrupt. */
schedule_alarm(now);
}
/*
* Reset state (called by ReScan).
*/
Datum
tsm_system_time_reset(PG_FUNCTION_ARGS)
{
TableSampleDesc *tsdesc = (TableSampleDesc *) PG_GETARG_POINTER(0);
SystemSamplerData *sampler = (SystemSamplerData *) tsdesc->tsmdata;
sampler->lt = InvalidOffsetNumber;
sampler->start_time = GetCurrentTimestamp();
sampler->end_time = TimestampTzPlusMilliseconds(sampler->start_time,
sampler->time);
sampler->estblocks = 2;
sampler->doneblocks = 0;
sampler_random_init_state(sampler->seed, sampler->randstate);
sampler->step = random_relative_prime(sampler->nblocks, sampler->randstate);
sampler->lb = sampler_random_fract(sampler->randstate) * (sampler->nblocks / sampler->step);
PG_RETURN_VOID();
}
/*
* Initializes the state.
*/
Datum
tsm_system_time_init(PG_FUNCTION_ARGS)
{
TableSampleDesc *tsdesc = (TableSampleDesc *) PG_GETARG_POINTER(0);
uint32 seed = PG_GETARG_UINT32(1);
int32 time = PG_ARGISNULL(2) ? -1 : PG_GETARG_INT32(2);
HeapScanDesc scan = tsdesc->heapScan;
SystemSamplerData *sampler;
if (time < 1)
ereport(ERROR,
(errcode(ERRCODE_NUMERIC_VALUE_OUT_OF_RANGE),
errmsg("invalid time limit"),
errhint("Time limit must be positive integer value.")));
sampler = palloc0(sizeof(SystemSamplerData));
/* Remember initial values for reinit */
sampler->seed = seed;
sampler->nblocks = scan->rs_nblocks;
sampler->lt = InvalidOffsetNumber;
sampler->estblocks = 2;
sampler->doneblocks = 0;
sampler->time = time;
sampler->start_time = GetCurrentTimestamp();
sampler->end_time = TimestampTzPlusMilliseconds(sampler->start_time,
sampler->time);
sampler_random_init_state(sampler->seed, sampler->randstate);
/* Find relative prime as step size for linear probing. */
sampler->step = random_relative_prime(sampler->nblocks, sampler->randstate);
/*
* Randomize start position so that blocks close to step size don't have
* higher probability of being chosen on very short scan.
*/
sampler->lb = sampler_random_fract(sampler->randstate) * (sampler->nblocks / sampler->step);
tsdesc->tsmdata = (void *) sampler;
PG_RETURN_VOID();
}
/*
* Enable multiple timeouts at once.
*
* This works like calling enable_timeout_after() and/or enable_timeout_at()
* multiple times. Use this to reduce the number of GetCurrentTimestamp()
* and setitimer() calls needed to establish multiple timeouts.
*/
void
enable_timeouts(const EnableTimeoutParams *timeouts, int count)
{
TimestampTz now;
int i;
/* Disable timeout interrupts for safety. */
disable_alarm();
/* Queue the timeout(s) at the appropriate times. */
now = GetCurrentTimestamp();
for (i = 0; i < count; i++)
{
TimeoutId id = timeouts[i].id;
TimestampTz fin_time;
switch (timeouts[i].type)
{
case TMPARAM_AFTER:
fin_time = TimestampTzPlusMilliseconds(now,
timeouts[i].delay_ms);
enable_timeout(id, now, fin_time);
break;
case TMPARAM_AT:
enable_timeout(id, now, timeouts[i].fin_time);
break;
default:
elog(ERROR, "unrecognized timeout type %d",
(int) timeouts[i].type);
break;
}
}
/* Set the timer interrupt. */
schedule_alarm(now);
}
/* Main loop of walsender process */
static int
WalSndLoop(void)
{
char *output_message;
bool caughtup = false;
/*
* Allocate buffer that will be used for each output message. We do this
* just once to reduce palloc overhead. The buffer must be made large
* enough for maximum-sized messages.
*/
output_message = palloc(1 + sizeof(WalDataMessageHeader) + MAX_SEND_SIZE);
/*
* Allocate buffer that will be used for processing reply messages. As
* above, do this just once to reduce palloc overhead.
*/
initStringInfo(&reply_message);
/* Initialize the last reply timestamp */
last_reply_timestamp = GetCurrentTimestamp();
/* Loop forever, unless we get an error */
for (;;)
{
/* Clear any already-pending wakeups */
ResetLatch(&MyWalSnd->latch);
/*
* Emergency bailout if postmaster has died. This is to avoid the
* necessity for manual cleanup of all postmaster children.
*/
if (!PostmasterIsAlive())
exit(1);
/* Process any requests or signals received recently */
if (got_SIGHUP)
{
got_SIGHUP = false;
ProcessConfigFile(PGC_SIGHUP);
SyncRepInitConfig();
}
/* Normal exit from the walsender is here */
if (walsender_shutdown_requested)
{
/* Inform the standby that XLOG streaming is done */
pq_puttextmessage('C', "COPY 0");
pq_flush();
proc_exit(0);
}
/* Check for input from the client */
ProcessRepliesIfAny();
/*
* If we don't have any pending data in the output buffer, try to send
* some more. If there is some, we don't bother to call XLogSend
* again until we've flushed it ... but we'd better assume we are not
* caught up.
*/
if (!pq_is_send_pending())
XLogSend(output_message, &caughtup);
else
caughtup = false;
/* Try to flush pending output to the client */
if (pq_flush_if_writable() != 0)
break;
/* If nothing remains to be sent right now ... */
if (caughtup && !pq_is_send_pending())
{
/*
* If we're in catchup state, move to streaming. This is an
* important state change for users to know about, since before
* this point data loss might occur if the primary dies and we
* need to failover to the standby. The state change is also
* important for synchronous replication, since commits that
* started to wait at that point might wait for some time.
*/
if (MyWalSnd->state == WALSNDSTATE_CATCHUP)
{
ereport(DEBUG1,
(errmsg("standby \"%s\" has now caught up with primary",
application_name)));
WalSndSetState(WALSNDSTATE_STREAMING);
}
/*
* When SIGUSR2 arrives, we send any outstanding logs up to the
* shutdown checkpoint record (i.e., the latest record) and exit.
* This may be a normal termination at shutdown, or a promotion,
* the walsender is not sure which.
*/
if (walsender_ready_to_stop)
{
/* ... let's just be real sure we're caught up ... */
XLogSend(output_message, &caughtup);
if (caughtup && !pq_is_send_pending())
{
walsender_shutdown_requested = true;
continue; /* don't want to wait more */
}
}
}
/*
* We don't block if not caught up, unless there is unsent data
* pending in which case we'd better block until the socket is
* write-ready. This test is only needed for the case where XLogSend
* loaded a subset of the available data but then pq_flush_if_writable
* flushed it all --- we should immediately try to send more.
*/
if (caughtup || pq_is_send_pending())
{
TimestampTz finish_time = 0;
long sleeptime = -1;
int wakeEvents;
wakeEvents = WL_LATCH_SET | WL_POSTMASTER_DEATH |
WL_SOCKET_READABLE;
if (pq_is_send_pending())
wakeEvents |= WL_SOCKET_WRITEABLE;
/* Determine time until replication timeout */
if (replication_timeout > 0)
{
long secs;
int usecs;
finish_time = TimestampTzPlusMilliseconds(last_reply_timestamp,
replication_timeout);
TimestampDifference(GetCurrentTimestamp(),
finish_time, &secs, &usecs);
sleeptime = secs * 1000 + usecs / 1000;
/* Avoid Assert in WaitLatchOrSocket if timeout is past */
if (sleeptime < 0)
sleeptime = 0;
wakeEvents |= WL_TIMEOUT;
}
/* Sleep until something happens or replication timeout */
WaitLatchOrSocket(&MyWalSnd->latch, wakeEvents,
MyProcPort->sock, sleeptime);
/*
* Check for replication timeout. Note we ignore the corner case
* possibility that the client replied just as we reached the
* timeout ... he's supposed to reply *before* that.
*/
if (replication_timeout > 0 &&
GetCurrentTimestamp() >= finish_time)
{
/*
* Since typically expiration of replication timeout means
* communication problem, we don't send the error message to
* the standby.
*/
ereport(COMMERROR,
(errmsg("terminating walsender process due to replication timeout")));
break;
}
}
}
/*
* Get here on send failure. Clean up and exit.
*
* Reset whereToSendOutput to prevent ereport from attempting to send any
* more messages to the standby.
*/
if (whereToSendOutput == DestRemote)
whereToSendOutput = DestNone;
proc_exit(0);
return 1; /* keep the compiler quiet */
}
/* Main loop of walsender process */
static int
WalSndLoop(void)
{
char *output_message;
bool caughtup = false;
/*
* Allocate buffer that will be used for each output message. We do this
* just once to reduce palloc overhead. The buffer must be made large
* enough for maximum-sized messages.
*/
output_message = palloc(1 + sizeof(WalDataMessageHeader) + MAX_SEND_SIZE);
/*
* Allocate buffer that will be used for processing reply messages. As
* above, do this just once to reduce palloc overhead.
*/
initStringInfo(&reply_message);
/* Initialize the last reply timestamp */
last_reply_timestamp = GetCurrentTimestamp();
/* Loop forever, unless we get an error */
for (;;)
{
/*
* 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);
SyncRepInitConfig();
}
/* Normal exit from the walsender is here */
if (walsender_shutdown_requested)
{
/* Inform the standby that XLOG streaming was done */
pq_puttextmessage('C', "COPY 0");
pq_flush();
proc_exit(0);
}
/*
* If we don't have any pending data in the output buffer, try to send
* some more.
*/
if (!pq_is_send_pending())
{
XLogSend(output_message, &caughtup);
/*
* Even if we wrote all the WAL that was available when we started
* sending, more might have arrived while we were sending this
* batch. We had the latch set while sending, so we have not
* received any signals from that time. Let's arm the latch again,
* and after that check that we're still up-to-date.
*/
if (caughtup && !pq_is_send_pending())
{
ResetLatch(&MyWalSnd->latch);
XLogSend(output_message, &caughtup);
}
}
/* Flush pending output to the client */
if (pq_flush_if_writable() != 0)
break;
/*
* When SIGUSR2 arrives, we send any outstanding logs up to the
* shutdown checkpoint record (i.e., the latest record) and exit.
*/
if (walsender_ready_to_stop && !pq_is_send_pending())
{
XLogSend(output_message, &caughtup);
ProcessRepliesIfAny();
if (caughtup && !pq_is_send_pending())
walsender_shutdown_requested = true;
}
if ((caughtup || pq_is_send_pending()) &&
!got_SIGHUP &&
!walsender_shutdown_requested)
{
TimestampTz finish_time = 0;
long sleeptime;
/* Reschedule replication timeout */
if (replication_timeout > 0)
{
long secs;
int usecs;
finish_time = TimestampTzPlusMilliseconds(last_reply_timestamp,
replication_timeout);
TimestampDifference(GetCurrentTimestamp(),
finish_time, &secs, &usecs);
sleeptime = secs * 1000 + usecs / 1000;
if (WalSndDelay < sleeptime)
sleeptime = WalSndDelay;
}
else
{
/*
* XXX: Without timeout, we don't really need the periodic
* wakeups anymore, WaitLatchOrSocket should reliably wake up
* as soon as something interesting happens.
*/
sleeptime = WalSndDelay;
}
/* Sleep */
WaitLatchOrSocket(&MyWalSnd->latch, MyProcPort->sock,
true, pq_is_send_pending(),
sleeptime);
/* Check for replication timeout */
if (replication_timeout > 0 &&
GetCurrentTimestamp() >= finish_time)
{
/*
* Since typically expiration of replication timeout means
* communication problem, we don't send the error message to
* the standby.
*/
ereport(COMMERROR,
(errmsg("terminating walsender process due to replication timeout")));
break;
}
}
/*
* If we're in catchup state, see if its time to move to streaming.
* This is an important state change for users, since before this
* point data loss might occur if the primary dies and we need to
* failover to the standby. The state change is also important for
* synchronous replication, since commits that started to wait at that
* point might wait for some time.
*/
if (MyWalSnd->state == WALSNDSTATE_CATCHUP && caughtup)
{
ereport(DEBUG1,
(errmsg("standby \"%s\" has now caught up with primary",
application_name)));
WalSndSetState(WALSNDSTATE_STREAMING);
}
ProcessRepliesIfAny();
}
/*
* Get here on send failure. Clean up and exit.
*
* Reset whereToSendOutput to prevent ereport from attempting to send any
* more messages to the standby.
*/
if (whereToSendOutput == DestRemote)
whereToSendOutput = DestNone;
proc_exit(0);
return 1; /* keep the compiler quiet */
}
/*
* Receive a log stream starting at the specified position.
*
* If sysidentifier is specified, validate that both the system
* identifier and the timeline matches the specified ones
* (by sending an extra IDENTIFY_SYSTEM command)
*
* All received segments will be written to the directory
* specified by basedir.
*
* The stream_stop callback will be called every time data
* is received, and whenever a segment is completed. If it returns
* true, the streaming will stop and the function
* return. As long as it returns false, streaming will continue
* indefinitely.
*
* standby_message_timeout controls how often we send a message
* back to the master letting it know our progress, in seconds.
* This message will only contain the write location, and never
* flush or replay.
*
* Note: The log position *must* be at a log segment start!
*/
bool
ReceiveXlogStream(PGconn *conn, XLogRecPtr startpos, uint32 timeline,
char *sysidentifier, char *basedir,
stream_stop_callback stream_stop,
int standby_message_timeout, bool rename_partial)
{
char query[128];
char current_walfile_name[MAXPGPATH];
PGresult *res;
char *copybuf = NULL;
int64 last_status = -1;
XLogRecPtr blockpos = InvalidXLogRecPtr;
if (sysidentifier != NULL)
{
/* Validate system identifier and timeline hasn't changed */
res = PQexec(conn, "IDENTIFY_SYSTEM");
if (PQresultStatus(res) != PGRES_TUPLES_OK)
{
fprintf(stderr,
_("%s: could not send replication command \"%s\": %s"),
progname, "IDENTIFY_SYSTEM", PQerrorMessage(conn));
PQclear(res);
return false;
}
if (PQnfields(res) != 3 || PQntuples(res) != 1)
{
fprintf(stderr,
_("%s: could not identify system: got %d rows and %d fields, expected %d rows and %d fields\n"),
progname, PQntuples(res), PQnfields(res), 1, 3);
PQclear(res);
return false;
}
if (strcmp(sysidentifier, PQgetvalue(res, 0, 0)) != 0)
{
fprintf(stderr,
_("%s: system identifier does not match between base backup and streaming connection\n"),
progname);
PQclear(res);
return false;
}
if (timeline != atoi(PQgetvalue(res, 0, 1)))
{
fprintf(stderr,
_("%s: timeline does not match between base backup and streaming connection\n"),
progname);
PQclear(res);
return false;
}
PQclear(res);
}
/* Initiate the replication stream at specified location */
snprintf(query, sizeof(query), "START_REPLICATION %X/%X", startpos.xlogid, startpos.xrecoff);
res = PQexec(conn, query);
if (PQresultStatus(res) != PGRES_COPY_BOTH)
{
fprintf(stderr, _("%s: could not send replication command \"%s\": %s"),
progname, "START_REPLICATION", PQresultErrorMessage(res));
PQclear(res);
return false;
}
PQclear(res);
/*
* Receive the actual xlog data
*/
while (1)
{
int r;
int xlogoff;
int bytes_left;
int bytes_written;
int64 now;
if (copybuf != NULL)
{
PQfreemem(copybuf);
copybuf = NULL;
}
/*
* Check if we should continue streaming, or abort at this point.
*/
if (stream_stop && stream_stop(blockpos, timeline, false))
{
if (walfile != -1 && !close_walfile(basedir, current_walfile_name,
rename_partial))
/* Potential error message is written by close_walfile */
goto error;
return true;
}
/*
* Potentially send a status message to the master
*/
now = localGetCurrentTimestamp();
if (standby_message_timeout > 0 &&
localTimestampDifferenceExceeds(last_status, now,
standby_message_timeout))
{
/* Time to send feedback! */
char replybuf[sizeof(StandbyReplyMessage) + 1];
StandbyReplyMessage *replymsg;
replymsg = (StandbyReplyMessage *) (replybuf + 1);
replymsg->write = blockpos;
replymsg->flush = InvalidXLogRecPtr;
replymsg->apply = InvalidXLogRecPtr;
replymsg->sendTime = now;
replybuf[0] = 'r';
if (PQputCopyData(conn, replybuf, sizeof(replybuf)) <= 0 ||
PQflush(conn))
{
fprintf(stderr, _("%s: could not send feedback packet: %s"),
progname, PQerrorMessage(conn));
goto error;
}
last_status = now;
}
r = PQgetCopyData(conn, ©buf, 1);
if (r == 0)
{
/*
* In async mode, and no data available. We block on reading but
* not more than the specified timeout, so that we can send a
* response back to the client.
*/
fd_set input_mask;
struct timeval timeout;
struct timeval *timeoutptr;
FD_ZERO(&input_mask);
FD_SET(PQsocket(conn), &input_mask);
if (standby_message_timeout)
{
TimestampTz targettime;
long secs;
int usecs;
targettime = TimestampTzPlusMilliseconds(last_status,
standby_message_timeout - 1);
localTimestampDifference(now,
targettime,
&secs,
&usecs);
if (secs <= 0)
timeout.tv_sec = 1; /* Always sleep at least 1 sec */
else
timeout.tv_sec = secs;
timeout.tv_usec = usecs;
timeoutptr = &timeout;
}
else
timeoutptr = NULL;
r = select(PQsocket(conn) + 1, &input_mask, NULL, NULL, timeoutptr);
if (r == 0 || (r < 0 && errno == EINTR))
{
/*
* Got a timeout or signal. Continue the loop and either
* deliver a status packet to the server or just go back into
* blocking.
*/
continue;
}
else if (r < 0)
{
fprintf(stderr, _("%s: select() failed: %s\n"),
progname, strerror(errno));
goto error;
}
/* Else there is actually data on the socket */
if (PQconsumeInput(conn) == 0)
{
fprintf(stderr,
_("%s: could not receive data from WAL stream: %s"),
progname, PQerrorMessage(conn));
goto error;
}
continue;
}
if (r == -1)
/* End of copy stream */
break;
if (r == -2)
{
fprintf(stderr, _("%s: could not read COPY data: %s"),
progname, PQerrorMessage(conn));
goto error;
}
if (copybuf[0] == 'k')
{
/*
* keepalive message, sent in 9.2 and newer. We just ignore this
* message completely, but need to skip past it in the stream.
*/
if (r != STREAMING_KEEPALIVE_SIZE)
{
fprintf(stderr,
_("%s: keepalive message has incorrect size %d\n"),
progname, r);
goto error;
}
continue;
}
else if (copybuf[0] != 'w')
{
fprintf(stderr, _("%s: unrecognized streaming header: \"%c\"\n"),
progname, copybuf[0]);
goto error;
}
if (r < STREAMING_HEADER_SIZE + 1)
{
fprintf(stderr, _("%s: streaming header too small: %d\n"),
progname, r);
goto error;
}
/* Extract WAL location for this block */
memcpy(&blockpos, copybuf + 1, 8);
xlogoff = blockpos.xrecoff % XLOG_SEG_SIZE;
/*
* Verify that the initial location in the stream matches where we
* think we are.
*/
if (walfile == -1)
{
/* No file open yet */
if (xlogoff != 0)
{
fprintf(stderr,
_("%s: received transaction log record for offset %u with no file open\n"),
progname, xlogoff);
goto error;
}
}
else
{
/* More data in existing segment */
/* XXX: store seek value don't reseek all the time */
if (lseek(walfile, 0, SEEK_CUR) != xlogoff)
{
fprintf(stderr,
_("%s: got WAL data offset %08x, expected %08x\n"),
progname, xlogoff, (int) lseek(walfile, 0, SEEK_CUR));
goto error;
}
}
bytes_left = r - STREAMING_HEADER_SIZE;
bytes_written = 0;
while (bytes_left)
{
int bytes_to_write;
/*
* If crossing a WAL boundary, only write up until we reach
* XLOG_SEG_SIZE.
*/
if (xlogoff + bytes_left > XLOG_SEG_SIZE)
bytes_to_write = XLOG_SEG_SIZE - xlogoff;
else
bytes_to_write = bytes_left;
if (walfile == -1)
{
walfile = open_walfile(blockpos, timeline,
basedir, current_walfile_name);
if (walfile == -1)
/* Error logged by open_walfile */
goto error;
}
if (write(walfile,
copybuf + STREAMING_HEADER_SIZE + bytes_written,
bytes_to_write) != bytes_to_write)
{
fprintf(stderr,
_("%s: could not write %u bytes to WAL file \"%s\": %s\n"),
progname, bytes_to_write, current_walfile_name,
strerror(errno));
goto error;
}
/* Write was successful, advance our position */
bytes_written += bytes_to_write;
bytes_left -= bytes_to_write;
XLByteAdvance(blockpos, bytes_to_write);
xlogoff += bytes_to_write;
/* Did we reach the end of a WAL segment? */
if (blockpos.xrecoff % XLOG_SEG_SIZE == 0)
{
if (!close_walfile(basedir, current_walfile_name, false))
/* Error message written in close_walfile() */
goto error;
xlogoff = 0;
if (stream_stop != NULL)
{
/*
* Callback when the segment finished, and return if it
* told us to.
*/
if (stream_stop(blockpos, timeline, true))
return true;
}
}
}
/* No more data left to write, start receiving next copy packet */
}
/*
* The only way to get out of the loop is if the server shut down the
* replication stream. If it's a controlled shutdown, the server will send
* a shutdown message, and we'll return the latest xlog location that has
* been streamed.
*/
res = PQgetResult(conn);
if (PQresultStatus(res) != PGRES_COMMAND_OK)
{
fprintf(stderr,
_("%s: unexpected termination of replication stream: %s"),
progname, PQresultErrorMessage(res));
goto error;
}
PQclear(res);
/* Complain if we've not reached stop point yet */
if (stream_stop != NULL && !stream_stop(blockpos, timeline, false))
{
fprintf(stderr, _("%s: replication stream was terminated before stop point\n"),
progname);
goto error;
}
if (copybuf != NULL)
PQfreemem(copybuf);
if (walfile != -1 && close(walfile) != 0)
fprintf(stderr, _("%s: could not close file \"%s\": %s\n"),
progname, current_walfile_name, strerror(errno));
walfile = -1;
return true;
error:
if (copybuf != NULL)
PQfreemem(copybuf);
if (walfile != -1 && close(walfile) != 0)
fprintf(stderr, _("%s: could not close file \"%s\": %s\n"),
progname, current_walfile_name, strerror(errno));
walfile = -1;
return false;
}
/*
* 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;
}
}
/*
* ResolveRecoveryConflictWithBufferPin is called from LockBufferForCleanup()
* to resolve conflicts with other backends holding buffer pins.
*
* We either resolve conflicts immediately or set a SIGALRM to wake us at
* the limit of our patience. The sleep in LockBufferForCleanup() is
* performed here, for code clarity.
*
* Resolve conflict by sending a SIGUSR1 reason to all backends to check if
* they hold one of the buffer pins that is blocking Startup process. If so,
* backends will take an appropriate error action, ERROR or FATAL.
*
* We also check for deadlocks before we wait, though applications that cause
* these will be extremely rare. Deadlocks occur because if queries
* wait on a lock, that must be behind an AccessExclusiveLock, which can only
* be cleared if the Startup process replays a transaction completion record.
* If Startup process is also waiting then that is a deadlock. The deadlock
* can occur if the query is waiting and then the Startup sleeps, or if
* Startup is sleeping and the query waits on a lock. We protect against
* only the former sequence here, the latter sequence is checked prior to
* the query sleeping, in CheckRecoveryConflictDeadlock().
*/
void
ResolveRecoveryConflictWithBufferPin(void)
{
bool sig_alarm_enabled = false;
Assert(InHotStandby);
if (MaxStandbyDelay == 0)
{
/*
* We don't want to wait, so just tell everybody holding the pin to
* get out of town.
*/
SendRecoveryConflictWithBufferPin(PROCSIG_RECOVERY_CONFLICT_BUFFERPIN);
}
else if (MaxStandbyDelay < 0)
{
/*
* Send out a request to check for buffer pin deadlocks before we
* wait. This is fairly cheap, so no need to wait for deadlock timeout
* before trying to send it out.
*/
SendRecoveryConflictWithBufferPin(PROCSIG_RECOVERY_CONFLICT_STARTUP_DEADLOCK);
}
else
{
TimestampTz then = GetLatestXLogTime();
TimestampTz now = GetCurrentTimestamp();
/* Are we past max_standby_delay? */
if (TimestampDifferenceExceeds(then, now, MaxStandbyDelay))
{
/*
* We're already behind, so clear a path as quickly as possible.
*/
SendRecoveryConflictWithBufferPin(PROCSIG_RECOVERY_CONFLICT_BUFFERPIN);
}
else
{
TimestampTz fin_time; /* Expected wake-up time by timer */
long timer_delay_secs; /* Amount of time we set timer
* for */
int timer_delay_usecs;
/*
* Send out a request to check for buffer pin deadlocks before we
* wait. This is fairly cheap, so no need to wait for deadlock
* timeout before trying to send it out.
*/
SendRecoveryConflictWithBufferPin(PROCSIG_RECOVERY_CONFLICT_STARTUP_DEADLOCK);
/*
* How much longer we should wait?
*/
fin_time = TimestampTzPlusMilliseconds(then, MaxStandbyDelay);
TimestampDifference(now, fin_time,
&timer_delay_secs, &timer_delay_usecs);
/*
* It's possible that the difference is less than a microsecond;
* ensure we don't cancel, rather than set, the interrupt.
*/
if (timer_delay_secs == 0 && timer_delay_usecs == 0)
timer_delay_usecs = 1;
if (enable_standby_sig_alarm(timer_delay_secs, timer_delay_usecs, fin_time))
sig_alarm_enabled = true;
else
elog(FATAL, "could not set timer for process wakeup");
}
}
/* Wait to be signaled by UnpinBuffer() */
ProcWaitForSignal();
if (sig_alarm_enabled)
{
if (!disable_standby_sig_alarm())
elog(FATAL, "could not disable timer for process wakeup");
}
}
static void
grab_ExecutorEnd(QueryDesc * queryDesc)
{
Datum values[10];
bool nulls[10] = {false, false, false, false, false, false, false, false, false, false};
Relation dump_heap;
RangeVar *dump_table_rv;
HeapTuple tuple;
Oid namespaceId;
/* lookup schema */
namespaceId = GetSysCacheOid1(NAMESPACENAME, CStringGetDatum(EXTENSION_SCHEMA));
if (OidIsValid(namespaceId)) {
/* lookup table */
if (OidIsValid(get_relname_relid(EXTENSION_LOG_TABLE, namespaceId))) {
/* get table heap */
dump_table_rv = makeRangeVar(EXTENSION_SCHEMA, EXTENSION_LOG_TABLE, -1);
dump_heap = heap_openrv(dump_table_rv, RowExclusiveLock);
/* transaction info */
values[0] = Int32GetDatum(GetCurrentTransactionId());
values[1] = Int32GetDatum(GetCurrentCommandId(false));
values[2] = Int32GetDatum(MyProcPid);
values[3] = Int32GetDatum(GetUserId());
/* query timing */
if (queryDesc->totaltime != NULL) {
InstrEndLoop(queryDesc->totaltime);
values[4] = TimestampGetDatum(
TimestampTzPlusMilliseconds(GetCurrentTimestamp(),
(queryDesc->totaltime->total * -1000.0)));
values[5] = Float8GetDatum(queryDesc->totaltime->total);
} else {
nulls[4] = true;
nulls[5] = true;
}
/* query command type */
values[6] = Int32GetDatum(queryDesc->operation);
/* query text */
values[7] = CStringGetDatum(
cstring_to_text(queryDesc->sourceText));
/* query params */
if (queryDesc->params != NULL) {
int numParams = queryDesc->params->numParams;
Oid out_func_oid, ptype;
Datum pvalue;
bool isvarlena;
FmgrInfo *out_functions;
bool arr_nulls[numParams];
size_t arr_nelems = (size_t) numParams;
Datum *arr_val_elems = palloc(sizeof(Datum) * arr_nelems);
Datum *arr_typ_elems = palloc(sizeof(Datum) * arr_nelems);
char elem_val_byval, elem_val_align, elem_typ_byval,
elem_typ_align;
int16 elem_val_len, elem_typ_len;
int elem_dims[1], elem_lbs[1];
int paramno;
/* init */
out_functions = (FmgrInfo *) palloc(
(numParams) * sizeof(FmgrInfo));
get_typlenbyvalalign(TEXTOID, &elem_val_len, &elem_val_byval, &elem_val_align);
get_typlenbyvalalign(REGTYPEOID, &elem_typ_len, &elem_typ_byval, &elem_typ_align);
elem_dims[0] = arr_nelems;
elem_lbs[0] = 1;
for (paramno = 0; paramno < numParams; paramno++) {
pvalue = queryDesc->params->params[paramno].value;
ptype = queryDesc->params->params[paramno].ptype;
getTypeOutputInfo(ptype, &out_func_oid, &isvarlena);
fmgr_info(out_func_oid, &out_functions[paramno]);
arr_typ_elems[paramno] = ptype;
arr_nulls[paramno] = true;
if (!queryDesc->params->params[paramno].isnull) {
arr_nulls[paramno] = false;
arr_val_elems[paramno] = PointerGetDatum(
cstring_to_text(
OutputFunctionCall(&out_functions[paramno], pvalue)));
}
}
values[8] = PointerGetDatum(
construct_md_array(
arr_val_elems,
arr_nulls,
1,
elem_dims,
elem_lbs,
TEXTOID,
elem_val_len,
elem_val_byval,
elem_val_align));
values[9] = PointerGetDatum(
construct_array(
arr_typ_elems,
arr_nelems,
REGTYPEOID,
elem_typ_len,
elem_typ_byval,
elem_typ_align));
pfree(out_functions);
pfree(arr_val_elems);
} else {
nulls[8] = true;
nulls[9] = true;
}
/* insert */
tuple = heap_form_tuple(dump_heap->rd_att, values, nulls);
simple_heap_insert(dump_heap, tuple);
heap_close(dump_heap, RowExclusiveLock);
}
}
if (prev_ExecutorEnd)
prev_ExecutorEnd(queryDesc);
else
standard_ExecutorEnd(queryDesc);
}
/*
* Enable the SIGALRM interrupt to fire after the specified delay
*
* Delay is given in milliseconds. Caller should be sure a SIGALRM
* signal handler is installed before this is called.
*
* This code properly handles nesting of deadlock timeout alarms within
* statement timeout alarms.
*
* Returns TRUE if okay, FALSE on failure.
*/
bool
enable_sig_alarm(int delayms, bool is_statement_timeout)
{
TimestampTz fin_time;
struct itimerval timeval;
if (is_statement_timeout)
{
/*
* Begin statement-level timeout
*
* Note that we compute statement_fin_time with reference to the
* statement_timestamp, but apply the specified delay without any
* correction; that is, we ignore whatever time has elapsed since
* statement_timestamp was set. In the normal case only a small
* interval will have elapsed and so this doesn't matter, but there
* are corner cases (involving multi-statement query strings with
* embedded COMMIT or ROLLBACK) where we might re-initialize the
* statement timeout long after initial receipt of the message. In
* such cases the enforcement of the statement timeout will be a bit
* inconsistent. This annoyance is judged not worth the cost of
* performing an additional gettimeofday() here.
*/
Assert(!deadlock_timeout_active);
fin_time = GetCurrentStatementStartTimestamp();
fin_time = TimestampTzPlusMilliseconds(fin_time, delayms);
statement_fin_time = fin_time;
cancel_from_timeout = false;
statement_timeout_active = true;
}
else if (statement_timeout_active)
{
/*
* Begin deadlock timeout with statement-level timeout active
*
* Here, we want to interrupt at the closer of the two timeout times.
* If fin_time >= statement_fin_time then we need not touch the
* existing timer setting; else set up to interrupt at the deadlock
* timeout time.
*
* NOTE: in this case it is possible that this routine will be
* interrupted by the previously-set timer alarm. This is okay
* because the signal handler will do only what it should do according
* to the state variables. The deadlock checker may get run earlier
* than normal, but that does no harm.
*/
fin_time = GetCurrentTimestamp();
fin_time = TimestampTzPlusMilliseconds(fin_time, delayms);
deadlock_timeout_active = true;
if (fin_time >= statement_fin_time)
return true;
}
else
{
/* Begin deadlock timeout with no statement-level timeout */
deadlock_timeout_active = true;
}
/* If we reach here, okay to set the timer interrupt */
MemSet(&timeval, 0, sizeof(struct itimerval));
timeval.it_value.tv_sec = delayms / 1000;
timeval.it_value.tv_usec = (delayms % 1000) * 1000;
if (setitimer(ITIMER_REAL, &timeval, NULL))
return false;
return true;
}
