People,
Here's another:
Summary: TOAST error on frequently truncated tables
Version: 7.4.2
Severity: Abort
Platform: Red Hat 9.0
When running a script against a reporting database, I'm getting the following:
DBD::Pg::st execute failed: ERROR: expected both swapped tables to have TOAST
tables at sql_data_xfer.pl line 120, <STDIN> line 2.
Process Failed: from step SELECT count(*) FROM elbs_client;Process Failed:
from step SELECT count(*) FROM elbs_client;
1 at sql_data_xfer.pl line 120, <STDIN> line 2.
What's happening in the scipt is that it's checking for the presence of data
in load tables and truncating them if data is found. I'm not absolutely
certain whether it is the COUNT(*) query, or the TRUNCATE step which is
producing the above error; TRUNCATE would seem more reasonable given the
nature of the error.
I've searched the source code, and the only reference to the error I can find
is for CLUSTER, which is confusing because none of the tables involved are
clustered.
Is this related to the index problem fixed in 7.4.x?
--
Josh Berkus
Aglio Database Solutions
San Francisco