I recently upgraded to Postgres 6.3.2 from 6.1 in a test environment. All tables
were accessible until this morning, when I discovered that one frequently read
tables was corrupted.
Performing a select on the table in psql resulted in the following error:
FATAL: unrecognized data from the backend. It probably dumped core.
Since the backend failed, no other commands would process from that point on.
To see what would happen I tried a pg_dump on the database and the following
message was displayed:
SQL query to dump the contents of Table ilsl_color_content did not execute
correctly. After we read all the table contents from the backend, PQendcopy()
failed. Explanation from backend: 'Error return detected from backend, but
attempt to read the message failed.'.
The query was: 'COPY ilsl_color_content TO stdout;
I then decided to \copy the table to a file and look at it. The copy did not
fail, but only part of the table was copied.
I then vacuumed the database, which was no done since the upgrade was preformed,
and the following message was displayed:
Rel ilsl_color_content: Uninitialized page 6 - fixing
After the vacuum, the corrupted table was accessible.
I would like to know what happened or where to look for answers. Is this type
of behavior typical with 6.3.2? I don't remember any problems like this with
6.1.
Lately, I have noticed several messages regarding the backend failing, and was
wondering if anyone else has had similar experiences?
Thanks in advance for your assistance.
Chris Bower