Hi!
>Those ten seconds are coming from RemoveXlogFile(), where pgrename()
loops 100 times for 100ms before giving up. So something holding up
the file's handle prevents the removal to happen.
I tried sysinternals
handle pg_wal
It shows that only postgres processes have opened files in pg_wal directory:
postgres.exe pid: 11800 type: File CC0: C:\Program
Files\PostgreSQL\13\data\pg_wal\0000000100000012000000B7
postgres.exe pid: 11800 type: File 12B8: C:\Program
Files\PostgreSQL\13\data\pg_wal
postgres.exe pid: 23904 type: File 1134: C:\Program
Files\PostgreSQL\13\data\pg_wal\0000000100000012000000B5
postgres.exe pid: 20908 type: File 17C: C:\Program
Files\PostgreSQL\13\data\pg_wal\0000000100000012000000B7
postgres.exe pid: 29892 type: File C08: C:\Program
Files\PostgreSQL\13\data\pg_wal\0000000100000011000000F4
...
postgres.exe pid: 34260 type: File C18: C:\Program
Files\PostgreSQL\13\data\pg_wal\0000000100000011000000FC
postgres.exe pid: 34484 type: File C48: C:\Program
Files\PostgreSQL\13\data\pg_wal\0000000100000011000000FC
postgres.exe pid: 38740 type: File 180: C:\Program
Files\PostgreSQL\13\data\pg_wal\0000000100000012000000B7
Attached is the
patch that should be tested, based on the suspected commit. There are
actually two scenarios to worry about:
- Check that the code of 13.2 compiled manually is enough to see the
failure.
- Check that once the patch attached is applied makes the failure go
away.
Intel server has
PostgreSQL 13.1, compiled by Visual C++ build 1914, 64-bit
and AMD server
PostgreSQL 13.2, compiled by Visual C++ build 1914, 64-bit
Should I try install Visual C++ , compile and replace postgres.exe file
in AMD server.
I am trying on my side to reproduce the problem in a more reliable
way. One thing I saw breaking in my setup is archive_command, where
it was not able to archive a segment with a simple copy, failing with
the same error as yours.
wal archiving is not used on those servers.
In one of those servers, do you have in pg_wal/ some files named
xlogtemp.N? N is an integer that would be the PID of the process that
generated it.
No. Intel server has 4 files with .deleted extension. AMD server has no
.deleted files. It has probably lower traffic.
Andrus.
