This looks to me like malfeasance of a VACUUM FULL: the reason there are multiple copies is that VACUUM FULL was trying to move the row around, and somehow you ended up with all three copies marked "good", when there should have been only one "good" copy at any instant.
So: (1) have you had any system crashes recently? (2) what sort of disk hardware is this running on? I'm wondering about IDE drives with write caching enabled :-(
For the record there is only supposed to be one row, and we do run a vacuum full each night, and it happened *after* this record was inserted.
1) No. Up 377 days.
2) SCSI - configured as a hardware RAID 5 thru a LSI Logic / Symbios Logic 53c1010 Ultra3 SCSI adapter
Can I now remove these two errant records or do you think there is anything to gain from continuing with this investigation ?
Kind regards
Rob
---------------------------(end of broadcast)--------------------------- TIP 5: Have you checked our extensive FAQ?
http://www.postgresql.org/docs/faq
