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

Commitcaba8f0

Browse files
committed
Fix corner case failure of new standby to follow new primary.
This only happens if (1) the new standby has no WAL available locally,(2) the new standby is starting from the old timeline, (3) the promotionhappened in the WAL segment from which the new standby is starting,(4) the timeline history file for the new timeline is available fromthe archive but the WAL files for are not (i.e. this is a race),(5) the WAL files for the new timeline are available via streaming,and (6) recovery_target_timeline='latest'.Commitee99427 introduced thislogic and was an improvement over the previous code, but it mishandledthis case. If recovery_target_timeline='latest' and restore_command isset, validateRecoveryParameters() can change recoveryTargetTLI to bedifferent from receiveTLI. If streaming is then tried afterward,expectedTLEs gets initialized with the history of the wrong timeline.It's supposed to be a list of entries explaining how to get to thetarget timeline, but in this case it ends up with a list of entriesexplaining how to get to the new standby's original timeline, whichisn't right.Dilip Kumar and Robert Haas, reviewed by Kyotaro Horiguchi.Discussion:http://postgr.es/m/CAFiTN-sE-jr=LB8jQuxeqikd-Ux+jHiXyh4YDiZMPedgQKup0g@mail.gmail.com
1 parent845cad4 commitcaba8f0

File tree

3 files changed

+115
-1
lines changed

3 files changed

+115
-1
lines changed

‎src/backend/access/transam/xlog.c

Lines changed: 9 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -12658,11 +12658,19 @@ WaitForWALToBecomeAvailable(XLogRecPtr RecPtr, bool randAccess,
1265812658
* pg_wal by now. Use XLOG_FROM_STREAM so that source
1265912659
* info is set correctly and XLogReceiptTime isn't
1266012660
* changed.
12661+
*
12662+
* NB: We must set readTimeLineHistory based on
12663+
* recoveryTargetTLI, not receiveTLI. Normally they'll
12664+
* be the same, but if recovery_target_timeline is
12665+
* 'latest' and archiving is configured, then it's
12666+
* possible that we managed to retrieve one or more
12667+
* new timeline history files from the archive,
12668+
* updating recoveryTargetTLI.
1266112669
*/
1266212670
if (readFile<0)
1266312671
{
1266412672
if (!expectedTLEs)
12665-
expectedTLEs=readTimeLineHistory(receiveTLI);
12673+
expectedTLEs=readTimeLineHistory(recoveryTargetTLI);
1266612674
readFile=XLogFileRead(readSegNo,PANIC,
1266712675
receiveTLI,
1266812676
XLOG_FROM_STREAM, false);
Lines changed: 96 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,96 @@
1+
2+
# Copyright (c) 2021, PostgreSQL Global Development Group
3+
4+
# Testing streaming replication where standby is promoted and a new cascading
5+
# standby (without WAL) is connected to the promoted standby. Both archiving
6+
# and streaming are enabled, but only the history file is available from the
7+
# archive, so the WAL files all have to be streamed. Test that the cascading
8+
# standby can follow the new primary (promoted standby).
9+
use strict;
10+
use warnings;
11+
use PostgresNode;
12+
use TestLib;
13+
use FindBin;
14+
use Test::Moretests=> 1;
15+
16+
# Initialize primary node
17+
my$node_primary = get_new_node('primary');
18+
19+
# Set up an archive command that will copy the history file but not the WAL
20+
# files. No real archive command should behave this way; the point is to
21+
# simulate a race condition where the new cascading standby starts up after
22+
# the timeline history file reaches the archive but before any of the WAL files
23+
# get there.
24+
$node_primary->init(allows_streaming=> 1,has_archiving=> 1);
25+
my$perlbin = $^X;
26+
$perlbin =~s{\\}{\\\\}gif ($TestLib::windows_os);
27+
my$archivedir_primary =$node_primary->archive_dir;
28+
$node_primary->append_conf('postgresql.conf',qq(
29+
archive_command = '$perlbin "$FindBin::RealBin/cp_history_files" "%p" "$archivedir_primary/%f"'
30+
));
31+
$node_primary->start;
32+
33+
# Take backup from primary
34+
my$backup_name ='my_backup';
35+
$node_primary->backup($backup_name);
36+
37+
# Create streaming standby linking to primary
38+
my$node_standby = get_new_node('standby');
39+
$node_standby->init_from_backup($node_primary,$backup_name,
40+
allows_streaming=> 1,has_streaming=> 1,has_archiving=> 1);
41+
$node_standby->start;
42+
43+
# Take backup of standby, use -Xnone so that pg_wal is empty.
44+
$node_standby->backup($backup_name,backup_options=> ['-Xnone']);
45+
46+
# Create cascading standby but don't start it yet.
47+
# Must set up both streaming and archiving.
48+
my$node_cascade = get_new_node('cascade');
49+
$node_cascade->init_from_backup($node_standby,$backup_name,
50+
has_streaming=> 1);
51+
$node_cascade->enable_restoring($node_primary);
52+
$node_cascade->append_conf('postgresql.conf',qq(
53+
recovery_target_timeline='latest'
54+
));
55+
56+
# Promote the standby.
57+
$node_standby->promote;
58+
59+
# Wait for promotion to complete
60+
$node_standby->poll_query_until('postgres',
61+
"SELECT NOT pg_is_in_recovery();")
62+
ordie"Timed out while waiting for promotion";
63+
64+
# Find next WAL segment to be archived
65+
my$walfile_to_be_archived =$node_standby->safe_psql('postgres',
66+
"SELECT pg_walfile_name(pg_current_wal_lsn());");
67+
68+
# Make WAL segment eligible for archival
69+
$node_standby->safe_psql('postgres','SELECT pg_switch_wal()');
70+
71+
# Wait until the WAL segment has been archived.
72+
# Since the history file gets created on promotion and is archived before any
73+
# WAL segment, this is enough to guarantee that the history file was
74+
# archived.
75+
my$archive_wait_query =
76+
"SELECT '$walfile_to_be_archived' <= last_archived_wal FROM pg_stat_archiver;";
77+
$node_standby->poll_query_until('postgres',$archive_wait_query)
78+
ordie"Timed out while waiting for WAL segment to be archived";
79+
my$last_archived_wal_file =$walfile_to_be_archived;
80+
81+
# Start cascade node
82+
$node_cascade->start;
83+
84+
# Create some content on promoted standby and check its presence on the
85+
# cascading standby.
86+
$node_standby->safe_psql('postgres',"CREATE TABLE tab_int AS SELECT 1 AS a");
87+
88+
# Wait for the replication to catch up
89+
$node_standby->wait_for_catchup($node_cascade,'replay',
90+
$node_standby->lsn('insert'));
91+
92+
# Check that cascading standby has the new content
93+
my$result =
94+
$node_cascade->safe_psql('postgres',"SELECT count(*) FROM tab_int");
95+
print"cascade:$result\n";
96+
is($result, 1,'check streamed content on cascade standby');

‎src/test/recovery/t/cp_history_files

Lines changed: 10 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,10 @@
1+
#!/usr/bin/perl
2+
3+
use File::Copy;
4+
use strict;
5+
use warnings;
6+
7+
die"wrong number of arguments"if@ARGV != 2;
8+
my ($source,$target) =@ARGV;
9+
exitif$source !~/history/;
10+
copy($source,$target)ordie"couldn't copy$source to$target:$!";

0 commit comments

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