Thread: Weird WAL numbering

Weird WAL numbering

From
"Christopher Kings-Lynne"
Date:
In my logs I have this:

2002-03-05 23:59:26 DEBUG:  MoveOfflineLogs: remove 0000000200000054
2002-03-05 23:59:26 DEBUG:  MoveOfflineLogs: remove 0000000200000055
2002-03-05 23:59:26 DEBUG:  MoveOfflineLogs: remove 0000000200000053
2002-03-05 23:59:38 DEBUG:  XLogWrite: new log file created - consider
increasing WAL_FILES
2002-03-06 00:00:10 DEBUG:  XLogWrite: new log file created - consider
increasing WAL_FILES
2002-03-06 00:00:23 DEBUG:  XLogWrite: new log file created - consider
increasing WAL_FILES
2002-03-06 00:00:25 DEBUG:  MoveOfflineLogs: remove 0000000200000057
2002-03-06 00:00:25 DEBUG:  MoveOfflineLogs: remove 0000000200000058
2002-03-06 00:00:25 DEBUG:  MoveOfflineLogs: remove 0000000200000056
2002-03-06 00:01:38 DEBUG:  XLogWrite: new log file created - consider
increasing WAL_FILES

Notice the '2' in the middle of the WAL file name?  It was '1' for quite a
while.  I get the feeling it increments when an overflow occurs, just as
going from 'FF' to '100' sort of thing.

ie. Is there an endian problem here?  Is it purely cosmetic?  Is it
deliberate?

usa=# select version();                          version
--------------------------------------------------------------PostgreSQL 7.1.3 on i386--freebsd4.3, compiled by GCC
2.95.2
(1 row)

uname -a
FreeBSD xxx 4.4-RELEASE FreeBSD 4.4-RELEASE #4: Mon Jan 21 0
7:14:44 GMT 2002     xxx:/usr/obj/usr/src/sys/CANAVERAL  i386

Chris



Re: Weird WAL numbering

From
Tom Lane
Date:
"Christopher Kings-Lynne" <chriskl@familyhealth.com.au> writes:
> Notice the '2' in the middle of the WAL file name?

Yeah, that's just how the WAL file-naming code works.  The upper and
lower halves of the 64-bit WAL address are handled independently.
        regards, tom lane