panic? return of the btree-bug

Hi,

2008-02-25 17:07:28 CET PANIC: failed to add item to the right
sibling for "ttrss_user_entries_feed_id"
2008-02-25 17:07:28 CET STATEMENT: INSERT INTO ttrss_user_entries

(ref_id, owner_uid, feed_id, unread, last_read, marked, published)
VALUES
('7135', '2', '18', true,

NULL, false, false)
2008-02-25 17:07:28 CET LOG: server process (PID 29709) was
terminated by signal 6
2008-02-25 17:07:28 CET LOG: terminating any other active server processes
2008-02-25 17:07:28 CET LOG: all server processes terminated; reinitializing
2008-02-25 17:07:28 CET LOG: database system was interrupted at
2008-02-25 17:05:30 CET
2008-02-25 17:07:28 CET LOG: checkpoint record is at 1/3C5A4FD8
2008-02-25 17:07:28 CET LOG: redo record is at 1/3C5A4FD8; undo
record is at 0/0; shutdown FALSE
2008-02-25 17:07:28 CET LOG: next transaction ID: 0/6130996; next OID: 42761
2008-02-25 17:07:28 CET LOG: next MultiXactId: 1; next MultiXactOffset: 0
2008-02-25 17:07:28 CET LOG: database system was not properly shut
down; automatic recovery in progress
2008-02-25 17:07:28 CET LOG: redo starts at 1/3C5A5028
2008-02-25 17:07:28 CET LOG: record with zero length at 1/3C641230
2008-02-25 17:07:28 CET LOG: redo done at 1/3C641200
2008-02-25 17:07:28 CET LOG: database system is ready

There is 1 process updating the ttrss_user_entries table rather often
(every 10 seconds +/- 100 queries).

There are several client processes (perl cgi) with their own dbi
connections to the database. I am unaware how many parallel users I
have.

Version: PostgreSQL 8.2.5 on x86_64-pc-linux-gnu, compiled by GCC cc
(GCC) 4.1.3 20070831 (prerelease) (Ubuntu 4.1.2-16ubuntu1)

Vacuum cron entry:
2 0,5,10,15,20 * * 1-6 root if [ -x /usr/sbin/pg_maintenance ]; then
/usr/sbin/pg_maintenance --analyze >/dev/null; fi

What information is needed to further investigate this?

Kind regards,
Pieter Jansen
ICTree