I have a bit of an odd problem with an update statement. I have a table
logs2 at 35 columns wide and a miserly 193 rows. I also have a table
testtime at 4 columns wide and 2.2 million rows. When I run the following
sql I get a success on testtime and a failure of "pqReadData() -- backend
closed the channel unexpectedly on logs2".
update <either tablename> set time2 = to_timestamp(time, 'HH24:MI:SS');
This is quite confusing to me, I have a success on updating a 2.2 million
row table and a failure on a 193 row table (the smaller data set is a subset
of the 2.2 million row selection). Both tables are fresh, built with a
select into statement, a variety of other queries work fine with these
tables. The machine has a gig of ram and a 800mb shared buffer. The sort
buffer set to 16mb.
What is causing this problem? How can I figure this out?
I have some theories on what might be going wrong. The sort buffers are
probably too small (I am working on getting my hands on the root password so
I can play with this). There is some problem with version 7.0.3 that is
causing the back end to drop out, but I have seen no postings on that
possibility.
So any advice on how to trace this problem down would be great.
-Jeff Barrett