From: Chao Li <li.evan.chao@gmail.com>
Message-Id: <93526C6D-DE0A-4B7D-B908-366735FC211D@gmail.com>
Content-Type: multipart/mixed;
	boundary="Apple-Mail=_52E463A0-C8F1-40F8-A35D-38AA926DA8D3"
Mime-Version: 1.0 (Mac OS X Mail 16.0 \(3864.600.51.1.1\))
Subject: Re: Fix pg_stat_wal_receiver to show CONNECTING status
Date: Fri, 22 May 2026 10:06:33 +0800
In-Reply-To: <1B9D2BAF-C73A-4A79-852C-17FB2A474AC0@gmail.com>
Cc: PostgreSQL-development <pgsql-hackers@postgresql.org>,
 Michael Paquier <michael.paquier@gmail.com>,
 Xuneng Zhou <xunengzhou@gmail.com>
To: Michael Paquier <michael@paquier.xyz>
References: <EF91FF76-1E2B-4F3B-9162-290B4DC517FF@gmail.com>
 <agxr29Hsz7FjxzlN@paquier.xyz>
 <1F153E64-B791-42FA-A60A-64813B20B81E@gmail.com>
 <ag00OeH1sbt5ie_6@paquier.xyz>
 <75CDE990-29D5-4D5C-BFE1-3840F19C0163@gmail.com>
 <ag4dCGAPBc5VFhCi@paquier.xyz>
 <1B695040-F544-447C-A6A8-C8BFF7F799D1@gmail.com>
 <E2A0E5FB-E841-4D4B-96F8-B4016359CDF0@gmail.com>
 <ag711gjcxuaBqpp8@paquier.xyz>
 <1B9D2BAF-C73A-4A79-852C-17FB2A474AC0@gmail.com>
Archived-At: 
 <https://www.postgresql.org/message-id/93526C6D-DE0A-4B7D-B908-366735FC211D%40gmail.com>
Precedence: bulk


--Apple-Mail=_52E463A0-C8F1-40F8-A35D-38AA926DA8D3
Content-Transfer-Encoding: quoted-printable
Content-Type: text/plain;
	charset=utf-8


> On May 21, 2026, at 20:29, Chao Li <li.evan.chao@gmail.com> wrote:
>=20
>=20
>=20
>> On May 21, 2026, at 20:08, Michael Paquier <michael@paquier.xyz> =
wrote:
>>=20
>> On Thu, May 21, 2026 at 03:20:13PM +0800, Chao Li wrote:
>>> I spent more time here, and found that it is still possible to leak
>>> conninfo in the WAL receiver reuse path:
>>>=20
>>> * WalRcvWaitForStartPosition() sets the state to WALRCV_WAITING.
>>> * Then RequestXLogStreaming() copies raw conninfo into
>>> * walrcv->conninfo and sets the state to WALRCV_RESTARTING.
>>> * WalRcvWaitForStartPosition() then moves the state to
>>> * WALRCV_CONNECTING, but this path does not clear walrcv->conninfo
>>> * again.
>>>=20
>>> The attached nocfbot_test.diff demonstrates the leak.
>>=20
>> File is missing, but I get it.  This is a legit bug from what I can
>> see, that also affects all the stable branches, not only HEAD.
>>=20
>>> Initially I thought we could also set ready_to_display to false when
>>> setting the state to WALRCV_WAITING in WalRcvWaitForStartPosition(),
>>> and set it back to true when switching back to
>>> WALRCV_CONNECTING. However, that would make the WALRCV_WAITING and
>>> WALRCV_RESTARTING states invisible in pg_stat_wal_receiver.
>>=20
>> Nah, we should not do that.  We want to track the waiting and
>> restarting states in the view.
>>=20
>>> I ended up with a solution that copies the primary connection info
>>> to walrcv->conninfo only when RequestXLogStreaming() is switching to
>>> WALRCV_STARTING. In the WALRCV_WAITING reuse path, the WAL receiver
>>> keeps using the existing wrconn, so it does not need raw conninfo to
>>> be copied into shared memory again. See the attached
>>> nocfbot_walreceiverfuncs.c.diff.
>>=20
>> Ah, yeah.  This solution to this problem makes sense.  We should not
>> clobber conninfo either in this case, or we'd lose the
>> user-displayable string returned by walrcv_get_conninfo() (conninfo
>> cannot be NULL based on the in-core callers of RequestXLogStreaming()
>> AFAIK, but who knows for things out there).  As mentioned above, this
>> is a different issue than the visibility of the connection =
information
>> while we are connecting, and it should be backpatched.  Would you =
like
>> to send a patch?
>> --
>> Michael
>=20
> Sorry for missing the attachments. Please take a look first. It=E2=80=99=
s late here, I can spend more time tomorrow.
>=20
> Best regards,
> --
> Chao Li (Evan)
> HighGo Software Co., Ltd.
> https://www.highgo.com/
>=20
>=20
>=20
>=20
> <nocfbot_test.diff><nocfbot_walreceiverfuncs.c.diff>

Here comes the patch set:

* v3-0001 is the exactly same as v2-0001
* In v3-0002, the change in walreceiverfuncs.c is the same as the =
previous diff, and I tuned the test change a little bit.

Best regards,
--
Chao Li (Evan)
HighGo Software Co., Ltd.
https://www.highgo.com/


--Apple-Mail=_52E463A0-C8F1-40F8-A35D-38AA926DA8D3
Content-Disposition: attachment;
	filename=v3-0001-Improve-pg_stat_wal_receiver-for-CONNECTING-statu.patch
Content-Type: application/octet-stream;
	x-unix-mode=0644;
	name="v3-0001-Improve-pg_stat_wal_receiver-for-CONNECTING-statu.patch"
Content-Transfer-Encoding: quoted-printable

=46rom=20f0091c46ddfba03cfc7f7a1e4154041694edcda6=20Mon=20Sep=2017=20=
00:00:00=202001=0AFrom:=20Michael=20Paquier=20<michael@paquier.xyz>=0A=
Date:=20Tue,=2019=20May=202026=2022:52:38=20+0900=0ASubject:=20[PATCH=20=
v3=201/2]=20Improve=20pg_stat_wal_receiver=20for=20CONNECTING=20status=0A=
=0ACommit=20a36164e7465=20added=20a=20CONNECTING=20status=20for=20the=20=
WAL=20receiver,=20but=0Apg_stat_wal_receiver=20returned=20no=20=
information=20while=20the=20connection=20to=20the=0Aprimary=20was=20=
attempted,=20limiting=20the=20usability=20of=20the=20feature=20in=0A=
high-latency=20environments=20where=20the=20connection=20attempt=20to=20=
the=20primary=0Acould=20take=20time.=0A=0AThis=20commit=20improves=20the=20=
report=20of=20the=20status=20by=20splitting=20the=20way=20the=0Ashared=20=
memory=20state=20of=20the=20WAL=20receiver=20is=20filled=20before=20and=20=
after=20the=0Aconnection=20to=20the=20primary=20is=20attempted:=0A-=20=
Before=20the=20attempt,=20reset=20all=20the=20connection=20fields,=20=
switch=0Aready_to_display=20to=20true.=0A-=20After=20the=20attempt,=20=
fill=20in=20the=20connection=20fields.=0A=0AThis=20change=20means=20two=20=
spinlock=20acquisitions=20instead=20of=20one,=20but=20at=20least=0A=
monitoring=20tools=20can=20know=20about=20the=20connection=20attempt=20=
before=20its=0Acompletion,=20enlarging=20the=20usability=20of=20the=20=
feature.=0A=0AReported-by:=20Chao=20Li=20<li.evan.chao@gmail.com>=0A=
Author:=20Michael=20Paquier=20<michael@paquier.xyz>=0ADiscussion:=20=
https://postgr.es/m/XXX=0A---=0A=20src/backend/replication/walreceiver.c=20=
|=2024=20++++++++++++++++--------=0A=201=20file=20changed,=2016=20=
insertions(+),=208=20deletions(-)=0A=0Adiff=20--git=20=
a/src/backend/replication/walreceiver.c=20=
b/src/backend/replication/walreceiver.c=0Aindex=20=
07eac07b9ce..d19317703c1=20100644=0A---=20=
a/src/backend/replication/walreceiver.c=0A+++=20=
b/src/backend/replication/walreceiver.c=0A@@=20-267,6=20+267,20=20@@=20=
WalReceiverMain(const=20void=20*startup_data,=20size_t=20=
startup_data_len)=0A=20=09/*=20Unblock=20signals=20(they=20were=20=
blocked=20when=20the=20postmaster=20forked=20us)=20*/=0A=20=09=
sigprocmask(SIG_SETMASK,=20&UnBlockSig,=20NULL);=0A=20=0A+=09/*=0A+=09=20=
*=20Switch=20the=20WAL=20receiver=20state=20as=20ready=20for=20display=20=
before=20doing=20a=0A+=09=20*=20connection=20attempt,=20so=20as=20its=20=
connecting=20state=20is=20visible=20before=0A+=09=20*=20attempting=20to=20=
contact=20the=20primary=20server.=20=20Note=20that=20this=20resets=20the=0A=
+=09=20*=20original=20conninfo,=20sender_port=20and=20sender_host,=20for=20=
security.=20=20These=0A+=09=20*=20fields=20are=20filled=20once=20the=20=
connection=20is=20fully=20established.=0A+=09=20*/=0A+=09=
SpinLockAcquire(&walrcv->mutex);=0A+=09memset(walrcv->conninfo,=200,=20=
MAXCONNINFO);=0A+=09memset(walrcv->sender_host,=200,=20NI_MAXHOST);=0A+=09=
walrcv->sender_port=20=3D=200;=0A+=09walrcv->ready_to_display=20=3D=20=
true;=0A+=09SpinLockRelease(&walrcv->mutex);=0A+=0A=20=09/*=20Establish=20=
the=20connection=20to=20the=20primary=20for=20XLOG=20streaming=20*/=0A=20=
=09appname=20=3D=20cluster_name[0]=20?=20cluster_name=20:=20=
"walreceiver";=0A=20=09wrconn=20=3D=20walrcv_connect(conninfo,=20true,=20=
false,=20false,=20appname,=20&err);=0A@@=20-277,23=20+291,17=20@@=20=
WalReceiverMain(const=20void=20*startup_data,=20size_t=20=
startup_data_len)=0A=20=09=09=09=09=09=09appname,=20err)));=0A=20=0A=20=09=
/*=0A-=09=20*=20Save=20user-visible=20connection=20string.=20=20This=20=
clobbers=20the=20original=0A-=09=20*=20conninfo,=20for=20security.=20=
Also=20save=20host=20and=20port=20of=20the=20sender=20server=0A-=09=20*=20=
this=20walreceiver=20is=20connected=20to.=0A+=09=20*=20Save=20=
user-visible=20connection=20string,=20now=20that=20the=20connection=20=
has=20been=0A+=09=20*=20achieved.=0A=20=09=20*/=0A=20=09tmp_conninfo=20=3D=
=20walrcv_get_conninfo(wrconn);=0A=20=09walrcv_get_senderinfo(wrconn,=20=
&sender_host,=20&sender_port);=0A=20=09SpinLockAcquire(&walrcv->mutex);=0A=
-=09memset(walrcv->conninfo,=200,=20MAXCONNINFO);=0A=20=09if=20=
(tmp_conninfo)=0A=20=09=09strlcpy(walrcv->conninfo,=20tmp_conninfo,=20=
MAXCONNINFO);=0A-=0A-=09memset(walrcv->sender_host,=200,=20NI_MAXHOST);=0A=
=20=09if=20(sender_host)=0A=20=09=09strlcpy(walrcv->sender_host,=20=
sender_host,=20NI_MAXHOST);=0A-=0A=20=09walrcv->sender_port=20=3D=20=
sender_port;=0A-=09walrcv->ready_to_display=20=3D=20true;=0A=20=09=
SpinLockRelease(&walrcv->mutex);=0A=20=0A=20=09if=20(tmp_conninfo)=0A--=20=
=0A2.50.1=20(Apple=20Git-155)=0A=0A=

--Apple-Mail=_52E463A0-C8F1-40F8-A35D-38AA926DA8D3
Content-Disposition: attachment;
	filename=v3-0002-Avoid-exposing-raw-WAL-receiver-conninfo-during-t.patch
Content-Type: application/octet-stream;
	x-unix-mode=0644;
	name="v3-0002-Avoid-exposing-raw-WAL-receiver-conninfo-during-t.patch"
Content-Transfer-Encoding: quoted-printable

=46rom=204dfab1d904ac6b86a7452d011e86379fba86623b=20Mon=20Sep=2017=20=
00:00:00=202001=0AFrom:=20"Chao=20Li=20(Evan)"=20<lic@highgo.com>=0A=
Date:=20Fri,=2022=20May=202026=2009:46:16=20+0800=0ASubject:=20[PATCH=20=
v3=202/2]=20Avoid=20exposing=20raw=20WAL=20receiver=20conninfo=20during=0A=
=20timeline=20jumps=0A=0AWhen=20reusing=20an=20existing=20WAL=20receiver=20=
after=20it=20has=20reached=0AWALRCV_WAITING,=20RequestXLogStreaming()=20=
copied=20PrimaryConnInfo=20into=0AWalRcv->conninfo=20before=20switching=20=
the=20state=20to=20WALRCV_RESTARTING.=20=20At=0Athat=20point=20=
ready_to_display=20could=20still=20be=20true,=20so=20=
pg_stat_wal_receiver=0Acould=20expose=20the=20raw=20connection=20string,=20=
including=20sensitive=20fields.=0A=0AWALRCV_RESTARTING=20does=20not=20=
establish=20a=20new=20connection.=20=20The=20waiting=20WAL=0Areceiver=20=
reuses=20its=20existing=20connection=20and=20only=20needs=20a=20new=20=
startpoint=0Aand=20timeline,=20so=20there=20is=20no=20need=20to=20copy=20=
the=20raw=20connection=20string=20into=0Ashared=20memory=20again.=20=20=
Only=20copy=20conninfo=20when=20launching=20a=20new=20WAL=20receiver=0A=
from=20WALRCV_STOPPED.=0A=0AAdd=20coverage=20to=20the=20timeline-switch=20=
test=20to=20verify=20that=20the=20WAL=20receiver=0Aprocess=20remains=20=
visible=20in=20pg_stat_wal_receiver=20across=20the=20jump,=20while=20a=0A=
raw=20password=20in=20primary_conninfo=20is=20not=20exposed.=0A=0A=
Author:=20Chao=20Li=20<lic@highgo.com>=0AReviewed-by:=0ADiscussion:=20=
https://postgr.es/m/EF91FF76-1E2B-4F3B-9162-290B4DC517FF@gmail.com=0A---=0A=
=20src/backend/replication/walreceiverfuncs.c=20|=20=209=20++++-----=0A=20=
src/test/recovery/t/004_timeline_switch.pl=20|=2016=20++++++++++++++--=0A=
=202=20files=20changed,=2018=20insertions(+),=207=20deletions(-)=0A=0A=
diff=20--git=20a/src/backend/replication/walreceiverfuncs.c=20=
b/src/backend/replication/walreceiverfuncs.c=0Aindex=20=
a0ed853e2f6..279c6c8a7e1=20100644=0A---=20=
a/src/backend/replication/walreceiverfuncs.c=0A+++=20=
b/src/backend/replication/walreceiverfuncs.c=0A@@=20-281,11=20+281,6=20=
@@=20RequestXLogStreaming(TimeLineID=20tli,=20XLogRecPtr=20recptr,=20=
const=20char=20*conninfo,=0A=20=09Assert(walrcv->walRcvState=20=3D=3D=20=
WALRCV_STOPPED=20||=0A=20=09=09=20=20=20walrcv->walRcvState=20=3D=3D=20=
WALRCV_WAITING);=0A=20=0A-=09if=20(conninfo=20!=3D=20NULL)=0A-=09=09=
strlcpy(walrcv->conninfo,=20conninfo,=20MAXCONNINFO);=0A-=09else=0A-=09=09=
walrcv->conninfo[0]=20=3D=20'\0';=0A-=0A=20=09/*=0A=20=09=20*=20Use=20=
configured=20replication=20slot=20if=20present,=20and=20ignore=20the=20=
value=20of=0A=20=09=20*=20create_temp_slot=20as=20the=20slot=20name=20=
should=20be=20persistent.=20=20Otherwise,=20use=0A@@=20-307,6=20+302,10=20=
@@=20RequestXLogStreaming(TimeLineID=20tli,=20XLogRecPtr=20recptr,=20=
const=20char=20*conninfo,=0A=20=09{=0A=20=09=09launch=20=3D=20true;=0A=20=
=09=09walrcv->walRcvState=20=3D=20WALRCV_STARTING;=0A+=09=09if=20=
(conninfo=20!=3D=20NULL)=0A+=09=09=09strlcpy(walrcv->conninfo,=20=
conninfo,=20MAXCONNINFO);=0A+=09=09else=0A+=09=09=09walrcv->conninfo[0]=20=
=3D=20'\0';=0A=20=09}=0A=20=09else=0A=20=09=09walrcv->walRcvState=20=3D=20=
WALRCV_RESTARTING;=0Adiff=20--git=20=
a/src/test/recovery/t/004_timeline_switch.pl=20=
b/src/test/recovery/t/004_timeline_switch.pl=0Aindex=20=
5afd2f44466..eb7cfa9f8e0=20100644=0A---=20=
a/src/test/recovery/t/004_timeline_switch.pl=0A+++=20=
b/src/test/recovery/t/004_timeline_switch.pl=0A@@=20-47,11=20+47,15=20@@=20=
$node_standby_1->psql(=0A=20=09stdout=20=3D>=20\$psql_out);=0A=20=
is($psql_out,=20't',=20"promotion=20of=20standby=20with=20pg_promote");=0A=
=20=0A-#=20Switch=20standby=202=20to=20replay=20from=20standby=201=0A+#=20=
Switch=20standby=202=20to=20replay=20from=20standby=201.=20=20During=20=
the=20timeline=20switch,=0A+#=20the=20WAL=20receiver=20process=20on=20=
standby=202=20should=20not=20be=20stopped,=20and=20the=0A+#=20new=20=
primary=20connection=20string=20should=20not=20be=20visible=0A+#=20in=20=
pg_stat_wal_receiver.=0A+my=20$secret=20=3D=20'dont_show_me';=0A=20my=20=
$connstr_1=20=3D=20$node_standby_1->connstr;=0A=20=
$node_standby_2->append_conf(=0A=20=09'postgresql.conf',=20qq(=0A=
-primary_conninfo=3D'$connstr_1'=0A+primary_conninfo=3D'$connstr_1=20=
password=3D$secret'=0A=20));=0A=20=0A=20#=20Rotate=20logfile=20before=20=
restarting,=20for=20the=20log=20checks=20done=20below.=0A@@=20-93,6=20=
+97,14=20@@=20my=20$wr_pid_after_switch=20=3D=20=
$node_standby_2->safe_psql('postgres',=0A=20is($wr_pid_before_switch,=20=
$wr_pid_after_switch,=0A=20=09'WAL=20receiver=20PID=20matches=20across=20=
timeline=20jumps');=0A=20=0A+my=20$raw_conninfo_count=20=3D=20=
$node_standby_2->safe_psql(=0A+=09'postgres',=0A+=09"SELECT=20count(*)=20=
FROM=20pg_stat_wal_receiver=20WHERE=20conninfo=20LIKE=20'%$secret%'");=0A=
+=0A+is(=0A+=09$raw_conninfo_count,=20'0',=0A+=09'raw=20primary_conninfo=20=
password=20is=20not=20visible=20after=20timeline=20jumps');=0A+=0A=20#=20=
Ensure=20that=20a=20standby=20is=20able=20to=20follow=20a=20primary=20on=20=
a=20newer=20timeline=0A=20#=20when=20WAL=20archiving=20is=20enabled.=0A=20=
=0A--=20=0A2.50.1=20(Apple=20Git-155)=0A=0A=

--Apple-Mail=_52E463A0-C8F1-40F8-A35D-38AA926DA8D3--