Hi,
I have postgres 9.4 standby with archive-based replication (via restore_command).
PostgreSQL 9.4.12 on x86_64-unknown-linux-gnu, compiled by gcc (Debian 4.9.2-10) 4.9.2, 64-bit
One day I noticed strange behavior in postgres logs: the same WAL has been restored many times:
"restored log file ""0000000200064AA80000005D"" from archive",,,,,,,,,""
"unexpected pageaddr 64A81/F9000000 in log segment 0000000200064AA80000005E, offset 0",,,,,,,,,""
"restored log file ""0000000200064AA80000005D"" from archive",,,,,,,,,""
"restored log file ""0000000200064AA80000005D"" from archive",,,,,,,,,""
"restored log file ""0000000200064AA80000005D"" from archive",,,,,,,,,""
"restored log file ""0000000200064AA80000005D"" from archive",,,,,,,,,""
"restored log file ""0000000200064AA80000005D"" from archive",,,,,,,,,""
"restored log file ""0000000200064AA80000005D"" from archive",,,,,,,,,""
"restored log file ""0000000200064AA80000005D"" from archive",,,,,,,,,""
"restored log file ""0000000200064AA80000005D"" from archive",,,,,,,,,""
"restored log file ""0000000200064AA80000005E"" from archive",,,,,,,,,""
"restored log file ""0000000200064AA80000005F"" from archive",,,,,,,,,""
"restored log file ""0000000200064AA800000060"" from archive",,,,,,,,,""
I enabled logging of successful execution in 'restore_command' script and realized that it's really happens: some WALs are recovering again and again. And this is always connected with 'unexpected pageaddr' log message.
Some statistics of how many 'same WALs' have been restored in 3 minutes:
[vyagofarov@myserver:~] $ cut -f 2 -d '-' wals_restored.log | uniq -c -d | sort -r
9 0000000200064AA9000000A4
9 0000000200064AA80000005D
8 0000000200064AAA00000044
7 0000000200064AAA000000AB
6 0000000200064AAA000000ED
3 0000000200064AA7000000E8
2 0000000200064AA80000003D
2 0000000200064AA800000033
2 0000000200064AA800000028
2 0000000200064AA800000018
2 0000000200064AA800000004
2 0000000200064AA7000000FD
Is this a bug?
--
Best regards,
Victor Yagofarov
