Transaction problem? - Mailing list pgsql-hackers

From Kurt Roeckx
Subject Transaction problem?
Date
Msg-id 20030415223620.GA26386@ping.be
Whole thread Raw
Responses Re: Transaction problem?  (Tom Lane <tgl@sss.pgh.pa.us>)
Re: Transaction problem?  (Hannu Krosing <hannu@tm.ee>)
List pgsql-hackers
I'm not exactly sure what the problem is, or how it happened.

I have a database where I deleted all records from all tables.  I
run a vacuum full and am left with about 1 GB of data.

What I think is happening is that it somehow thinks that data is
still in use by an other transanction, but that is not the case.
I even restated to make sure.

Anybody got any idea how this could have happened?

I once saw this in the log, and think it might be related:

2003-04-12 21:30:55 LOG:  database system was interrupted at
2003-04-12 21:30:35 CEST
2003-04-12 21:30:55 LOG:  checkpoint record is at 1/1DC89028
2003-04-12 21:30:55 LOG:  redo record is at 1/1C128B54; undo
record is at 0/0; shutdown FALSE
2003-04-12 21:30:55 LOG:  next transaction id: 741; next oid:
9753304
2003-04-12 21:30:55 LOG:  database system was not properly shut
down; automatic
recovery in progress
2003-04-12 21:30:55 LOG:  redo starts at 1/1C128B54
2003-04-12 21:32:23 LOG:  ReadRecord: unexpected pageaddr
1/15AE2000 in log file 1, segment 36, offset 11411456
2003-04-12 21:32:23 LOG:  redo done at 1/24AE0B60

Should I worry about that "unexpected pageaddr"?  It ussually
says "ReadRecord: record with zero length" after an unclean shut
down.


Kurt



pgsql-hackers by date:

Previous
From: "Dann Corbit"
Date:
Subject: Re: Are we losing momentum?
Next
From: "Robert E. Bruccoleri"
Date:
Subject: Re: Are we losing momentum?