Movatterモバイル変換

[0]ホーム
URL:

Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly

 Sign up
Appearance settings

Commitac1d794

Browse files
committed
Make idle backends exit if the postmaster dies.
Letting backends continue to run if the postmaster has exited preventsPostgreSQL from being restarted, which in many environments iscatastrophic. Worse, if some other backend crashes, we no longer haveany protection against shared memory corruption. So, arrange for themto exit instead. We don't want to expend many cycles on this, butincluding postmaster death in the set of things that we wait for whena backend is idle seems cheap enough.Rajeev Rastogi and Robert Haas
1 parenta05dc4d commitac1d794

File tree

1 file changed

+31
-2
lines changed

1 file changed

+31
-2
lines changed

‎src/backend/libpq/be-secure.c

Lines changed: 31 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -35,6 +35,7 @@
3535
#include"miscadmin.h"
3636
#include"tcop/tcopprot.h"
3737
#include"utils/memutils.h"
38+
#include"storage/ipc.h"
3839
#include"storage/proc.h"
3940

4041

@@ -144,9 +145,31 @@ secure_read(Port *port, void *ptr, size_t len)
144145
Assert(waitfor);
145146

146147
w=WaitLatchOrSocket(MyLatch,
147-
WL_LATCH_SET |waitfor,
148+
WL_LATCH_SET |WL_POSTMASTER_DEATH |waitfor,
148149
port->sock,0);
149150

151+
/*
152+
* If the postmaster has died, it's not safe to continue running,
153+
* because it is the postmaster's job to kill us if some other backend
154+
* exists uncleanly. Moreover, we won't run very well in this state;
155+
* helper processes like walwriter and the bgwriter will exit, so
156+
* performance may be poor. Finally, if we don't exit, pg_ctl will
157+
* be unable to restart the postmaster without manual intervention,
158+
* so no new connections can be accepted. Exiting clears the deck
159+
* for a postmaster restart.
160+
*
161+
* (Note that we only make this check when we would otherwise sleep
162+
* on our latch. We might still continue running for a while if the
163+
* postmaster is killed in mid-query, or even through multiple queries
164+
* if we never have to wait for read. We don't want to burn too many
165+
* cycles checking for this very rare condition, and this should cause
166+
* us to exit quickly in most cases.)
167+
*/
168+
if (w&WL_POSTMASTER_DEATH)
169+
ereport(FATAL,
170+
(errcode(ERRCODE_ADMIN_SHUTDOWN),
171+
errmsg("terminating connection due to unexpected postmaster exit")));
172+
150173
/* Handle interrupt. */
151174
if (w&WL_LATCH_SET)
152175
{
@@ -223,9 +246,15 @@ secure_write(Port *port, void *ptr, size_t len)
223246
Assert(waitfor);
224247

225248
w=WaitLatchOrSocket(MyLatch,
226-
WL_LATCH_SET |waitfor,
249+
WL_LATCH_SET |WL_POSTMASTER_DEATH |waitfor,
227250
port->sock,0);
228251

252+
/* See comments in secure_read. */
253+
if (w&WL_POSTMASTER_DEATH)
254+
ereport(FATAL,
255+
(errcode(ERRCODE_ADMIN_SHUTDOWN),
256+
errmsg("terminating connection due to unexpected postmaster exit")));
257+
229258
/* Handle interrupt. */
230259
if (w&WL_LATCH_SET)
231260
{

0 commit comments

Comments
 (0)
[8]ページ先頭
©2009-2025 Movatter.jp