Hello
> So, I'd like to propose to move the stuff to the second switch().
> (See the attached incomplete patch.) This is rather similar to
> Sergei's previous proposal, but the structure of the state
> machine is kept.
Very similar to my v4 proposal (also move RequestXLogStreaming call), but
closer to currentSource reading. No objections from me, attached patch is
changed this way.
I renamed start_wal_receiver to startWalReceiver - this style looks more
consistent to near code.
> + /*
> + * Else, check if WAL receiver is still
> active.
> + */
> + else if (!WalRcvStreaming())
I think we still need wait WalRcvStreaming after RequestXLogStreaming call. So
I remove else branch and leave separate condition.
> In ProcessStartupSigHup, conninfo and slotname don't need to be
> retained until the end of the function.
Agreed, I move pfree
> The log message in the function seems to be too detailed. On the
> other hand, if we changed primary_conninfo to '' (stop) or vise
> versa (start), the message (restart) looks strange.
I have no strong opinion here. These messages was changed many times during
this thread lifetime, can be changed anytime. I think this is not issue since
we have no consensus about overall design.
Write detailed messages was proposed here:
https://www.postgresql.org/message-id/20190216151025.GJ2240%40paquier.xyz
> or vise versa (start)
I explicitly check currentSource and WalRcvRunning, so we have no such messages
if user had no walreceiver before.
regards, Sergei
diff --git a/doc/src/sgml/config.sgml b/doc/src/sgml/config.sgml
index 0191ec84b1..587031dea8 100644
--- a/doc/src/sgml/config.sgml
+++ b/doc/src/sgml/config.sgml
@@ -3983,9 +3983,15 @@ ANY <replaceable class="parameter">num_sync</replaceable> ( <replaceable class="
<varname>primary_conninfo</varname> string.
</para>
<para>
- This parameter can only be set at server start.
+ This parameter can only be set in the <filename>postgresql.conf</filename>
+ file or on the server command line.
This setting has no effect if the server is not in standby mode.
</para>
+ <para>
+ If <varname>primary_conninfo</varname> is changed while WAL receiver is running,
+ the WAL receiver shuts down and then restarts with new setting,
+ except when primary_conninfo is an empty string.
+ </para>
</listitem>
</varlistentry>
<varlistentry id="guc-primary-slot-name" xreflabel="primary_slot_name">
@@ -4000,9 +4006,13 @@ ANY <replaceable class="parameter">num_sync</replaceable> ( <replaceable class="
connecting to the sending server via streaming replication to control
resource removal on the upstream node
(see <xref linkend="streaming-replication-slots"/>).
- This parameter can only be set at server start.
+ This parameter can only be set in the <filename>postgresql.conf</filename>
+ file or on the server command line.
This setting has no effect if <varname>primary_conninfo</varname> is not
- set.
+ set or the server is not in standby mode.
+ </para>
+ <para>
+ The WAL receiver is restarted after an update of <varname>primary_slot_name</varname>.
</para>
</listitem>
</varlistentry>
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 2e3cc51006..8eee079eb2 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -803,6 +803,12 @@ static XLogSource readSource = 0; /* XLOG_FROM_* code */
static XLogSource currentSource = 0; /* XLOG_FROM_* code */
static bool lastSourceFailed = false;
+/*
+ * Need for restart running WalReceiver due the configuration change.
+ * Suitable only for XLOG_FROM_STREAM source
+ */
+static bool pendingWalRcvRestart = false;
+
typedef struct XLogPageReadPrivate
{
int emode;
@@ -11773,6 +11779,7 @@ WaitForWALToBecomeAvailable(XLogRecPtr RecPtr, bool randAccess,
for (;;)
{
int oldSource = currentSource;
+ bool startWalReceiver = false;
/*
* First check if we failed to read from the current source, and
@@ -11806,54 +11813,13 @@ WaitForWALToBecomeAvailable(XLogRecPtr RecPtr, bool randAccess,
if (!StandbyMode)
return false;
- /*
- * If primary_conninfo is set, launch walreceiver to try
- * to stream the missing WAL.
- *
- * If fetching_ckpt is true, RecPtr points to the initial
- * checkpoint location. In that case, we use RedoStartLSN
- * as the streaming start position instead of RecPtr, so
- * that when we later jump backwards to start redo at
- * RedoStartLSN, we will have the logs streamed already.
- */
- if (PrimaryConnInfo && strcmp(PrimaryConnInfo, "") != 0)
- {
- XLogRecPtr ptr;
- TimeLineID tli;
-
- if (fetching_ckpt)
- {
- ptr = RedoStartLSN;
- tli = ControlFile->checkPointCopy.ThisTimeLineID;
- }
- else
- {
- ptr = RecPtr;
-
- /*
- * Use the record begin position to determine the
- * TLI, rather than the position we're reading.
- */
- tli = tliOfPointInHistory(tliRecPtr, expectedTLEs);
-
- if (curFileTLI > 0 && tli < curFileTLI)
- elog(ERROR, "according to history file, WAL location %X/%X belongs to timeline %u, but previous recovered WAL file came from timeline %u",
- (uint32) (tliRecPtr >> 32),
- (uint32) tliRecPtr,
- tli, curFileTLI);
- }
- curFileTLI = tli;
- RequestXLogStreaming(tli, ptr, PrimaryConnInfo,
- PrimarySlotName);
- receivedUpto = 0;
- }
-
/*
* Move to XLOG_FROM_STREAM state in either case. We'll
* get immediate failure if we didn't launch walreceiver,
* and move on to the next state.
*/
currentSource = XLOG_FROM_STREAM;
+ startWalReceiver = true;
break;
case XLOG_FROM_STREAM:
@@ -11999,7 +11965,69 @@ WaitForWALToBecomeAvailable(XLogRecPtr RecPtr, bool randAccess,
Assert(StandbyMode);
/*
- * Check if WAL receiver is still active.
+ * shutdown WAL receiver if restart is requested.
+ */
+ if (!startWalReceiver && pendingWalRcvRestart)
+ {
+ if (WalRcvRunning())
+ ShutdownWalRcv();
+
+ /*
+ * Re-scan for possible new timelines if we were
+ * requested to recover to the latest timeline.
+ */
+ if (recoveryTargetTimeLineGoal ==
+ RECOVERY_TARGET_TIMELINE_LATEST)
+ rescanLatestTimeLine();
+
+ startWalReceiver = true;
+ }
+ pendingWalRcvRestart = false;
+
+ /*
+ * Launch walreceiver if needed.
+ *
+ * If fetching_ckpt is true, RecPtr points to the initial
+ * checkpoint location. In that case, we use RedoStartLSN
+ * as the streaming start position instead of RecPtr, so
+ * that when we later jump backwards to start redo at
+ * RedoStartLSN, we will have the logs streamed already.
+ */
+ if (startWalReceiver &&
+ PrimaryConnInfo && strcmp(PrimaryConnInfo, "") != 0)
+ {
+ XLogRecPtr ptr;
+ TimeLineID tli;
+
+ if (fetching_ckpt)
+ {
+ ptr = RedoStartLSN;
+ tli = ControlFile->checkPointCopy.ThisTimeLineID;
+ }
+ else
+ {
+ ptr = RecPtr;
+
+ /*
+ * Use the record begin position to determine the
+ * TLI, rather than the position we're reading.
+ */
+ tli = tliOfPointInHistory(tliRecPtr, expectedTLEs);
+
+ if (curFileTLI > 0 && tli < curFileTLI)
+ elog(ERROR, "according to history file, WAL location %X/%X belongs to timeline %u, but previous recovered WAL file came from timeline %u",
+ (uint32) (tliRecPtr >> 32),
+ (uint32) tliRecPtr,
+ tli, curFileTLI);
+ }
+ curFileTLI = tli;
+ RequestXLogStreaming(tli, ptr, PrimaryConnInfo,
+ PrimarySlotName);
+ receivedUpto = 0;
+ }
+
+ /*
+ * Check if WAL receiver is active or wait to start up.
*/
if (!WalRcvStreaming())
{
@@ -12127,6 +12155,45 @@ WaitForWALToBecomeAvailable(XLogRecPtr RecPtr, bool randAccess,
return false; /* not reached */
}
+/*
+ * Re-read config file and plan to restart running walreceiver if
+ * connection settings was changed.
+ */
+void
+ProcessStartupSigHup(void)
+{
+ char *conninfo = pstrdup(PrimaryConnInfo);
+ char *slotname = pstrdup(PrimarySlotName);
+ bool conninfoChanged;
+ bool slotnameChanged;
+
+ ProcessConfigFile(PGC_SIGHUP);
+
+ /*
+ * We need restart walreceiver if replication settings was changed.
+ */
+ conninfoChanged = (strcmp(conninfo, PrimaryConnInfo) != 0);
+ slotnameChanged = (strcmp(slotname, PrimarySlotName) != 0);
+
+ pfree(conninfo);
+ pfree(slotname);
+
+ if ((conninfoChanged || slotnameChanged) &&
+ currentSource == XLOG_FROM_STREAM
+ && WalRcvRunning())
+ {
+ if (conninfoChanged && slotnameChanged)
+ ereport(LOG,
+ (errmsg("The WAL receiver is going to be restarted due to change of primary_conninfo and primary_slot_name")));
+ else
+ ereport(LOG,
+ (errmsg("The WAL receiver is going to be restarted due to change of %s",
+ conninfoChanged ? "primary_conninfo" : "primary_slot_name")));
+
+ pendingWalRcvRestart = true;
+ }
+}
+
/*
* Determine what log level should be used to report a corrupt WAL record
* in the current WAL page, previously read by XLogPageRead().
diff --git a/src/backend/postmaster/startup.c b/src/backend/postmaster/startup.c
index 5048a2c2aa..62492daf6b 100644
--- a/src/backend/postmaster/startup.c
+++ b/src/backend/postmaster/startup.c
@@ -142,12 +142,12 @@ void
HandleStartupProcInterrupts(void)
{
/*
- * Check if we were requested to re-read config file.
+ * Process any requests or signals received recently.
*/
if (got_SIGHUP)
{
got_SIGHUP = false;
- ProcessConfigFile(PGC_SIGHUP);
+ ProcessStartupSigHup();
}
/*
diff --git a/src/backend/replication/walreceiver.c b/src/backend/replication/walreceiver.c
index f54ae7690d..8698e07f8a 100644
--- a/src/backend/replication/walreceiver.c
+++ b/src/backend/replication/walreceiver.c
@@ -644,7 +644,11 @@ WalRcvWaitForStartPosition(XLogRecPtr *startpoint, TimeLineID *startpointTLI)
walrcv->walRcvState == WALRCV_STOPPING);
if (walrcv->walRcvState == WALRCV_RESTARTING)
{
- /* we don't expect primary_conninfo to change */
+ /*
+ * We don't need handle changes of primary_conninfo or
+ * primary_slotname here. Startup process will shutdown running
+ * walreceiver in this case.
+ */
*startpoint = walrcv->receiveStart;
*startpointTLI = walrcv->receiveStartTLI;
walrcv->walRcvState = WALRCV_STREAMING;
diff --git a/src/backend/utils/misc/guc.c b/src/backend/utils/misc/guc.c
index 31a5ef0474..432296714f 100644
--- a/src/backend/utils/misc/guc.c
+++ b/src/backend/utils/misc/guc.c
@@ -3558,7 +3558,7 @@ static struct config_string ConfigureNamesString[] =
},
{
- {"primary_conninfo", PGC_POSTMASTER, REPLICATION_STANDBY,
+ {"primary_conninfo", PGC_SIGHUP, REPLICATION_STANDBY,
gettext_noop("Sets the connection string to be used to connect to the sending server."),
NULL,
GUC_SUPERUSER_ONLY
@@ -3569,7 +3569,7 @@ static struct config_string ConfigureNamesString[] =
},
{
- {"primary_slot_name", PGC_POSTMASTER, REPLICATION_STANDBY,
+ {"primary_slot_name", PGC_SIGHUP, REPLICATION_STANDBY,
gettext_noop("Sets the name of the replication slot to use on the sending server."),
NULL
},
diff --git a/src/backend/utils/misc/postgresql.conf.sample b/src/backend/utils/misc/postgresql.conf.sample
index 0fc23e3a61..c25e406b9e 100644
--- a/src/backend/utils/misc/postgresql.conf.sample
+++ b/src/backend/utils/misc/postgresql.conf.sample
@@ -308,9 +308,7 @@
# These settings are ignored on a master server.
#primary_conninfo = '' # connection string to sending server
- # (change requires restart)
#primary_slot_name = '' # replication slot on sending server
- # (change requires restart)
#promote_trigger_file = '' # file name whose presence ends recovery
#hot_standby = on # "off" disallows queries during recovery
# (change requires restart)
diff --git a/src/include/access/xlog.h b/src/include/access/xlog.h
index d519252aad..9e49020b19 100644
--- a/src/include/access/xlog.h
+++ b/src/include/access/xlog.h
@@ -320,6 +320,7 @@ extern void SetWalWriterSleeping(bool sleeping);
extern void XLogRequestWalReceiverReply(void);
+extern void ProcessStartupSigHup(void);
extern void assign_max_wal_size(int newval, void *extra);
extern void assign_checkpoint_completion_target(double newval, void *extra);
diff --git a/src/test/recovery/t/001_stream_rep.pl b/src/test/recovery/t/001_stream_rep.pl
index 3c743d7d7c..ae80f4df3a 100644
--- a/src/test/recovery/t/001_stream_rep.pl
+++ b/src/test/recovery/t/001_stream_rep.pl
@@ -3,7 +3,7 @@ use strict;
use warnings;
use PostgresNode;
use TestLib;
-use Test::More tests => 32;
+use Test::More tests => 33;
# Initialize master node
my $node_master = get_new_node('master');
@@ -208,7 +208,9 @@ $node_standby_2->append_conf('postgresql.conf',
"primary_slot_name = $slotname_2");
$node_standby_2->append_conf('postgresql.conf',
"wal_receiver_status_interval = 1");
-$node_standby_2->restart;
+# should be able change primary_slot_name without restart
+# will wait effect in get_slot_xmins above
+$node_standby_2->reload;
# Fetch xmin columns from slot's pg_replication_slots row, after waiting for
# given boolean condition to be true to ensure we've reached a quiescent state
@@ -344,3 +346,21 @@ is($catalog_xmin, '',
is($xmin, '', 'xmin of cascaded slot null with hs feedback reset');
is($catalog_xmin, '',
'catalog xmin of cascaded slot still null with hs_feedback reset');
+
+note "check change primary_conninfo without restart";
+$node_standby_2->append_conf('postgresql.conf',
+ "primary_slot_name = ''");
+$node_standby_2->enable_streaming($node_master);
+$node_standby_2->reload;
+
+# be sure do not streaming from cascade
+$node_standby_1->stop;
+
+my $newval = $node_master->safe_psql('postgres',
+'INSERT INTO replayed(val) SELECT coalesce(max(val),0) + 1 AS newval FROM replayed RETURNING val'
+);
+$node_master->wait_for_catchup($node_standby_2, 'replay',
+ $node_master->lsn('insert'));
+my $is_replayed = $node_standby_2->safe_psql('postgres',
+ qq[SELECT 1 FROM replayed WHERE val = $newval]);
+is($is_replayed, qq(1), "standby_2 didn't replay master value $newval");