Greetings,
This follows a post I just made with the subject line: [Backend core dump,
Please help, Urgent!]
We just had the same thing occur on a completely different server! It
happened on our development server which is different hardware and
different version of FreeBSD (3.2-Release) than on our production server.
However both are running pg-6.5.1. Both of these machines have been
running for over 6 months without incident.
One of the programmers was in psql, he created this table:
CREATE TABLE "instacom" (
"submit_date" date,
"instacom_id" int4,
"message" text);
CREATE INDEX "instacom_submit_date" on "instacom" using btree
"submit_date" "date_ops" );
I'm not sure how much data he had in the table, but I'm sure it was not
more than a few records, but when he submitted a query on this table
*only*, other tables queried just fine, the backend crashed and this
message displayed on the console:
Dec 14 16:00:56 bluebox /kernel: pid 79923 (postgres), uid 1001: exited on
signal 11
Dec 14 16:01:14 bluebox /kernel: pid 79925 (postgres), uid 1001: exited on
signal 11
Dec 14 16:03:06 bluebox /kernel: pid 79940 (postgres), uid 1001: exited on
signal 10
The signal 10 kind of caught my eye since we had only seen signal 11s so
far. He restarted the server, delete the table, and recreated it like this:
CREATE TABLE "instacom" (
"submit_date" date,
"instacom_id" int4);
CREATE INDEX "instacom_submit_date" on "instacom" using btree
"submit_date" "date_ops" );
The backend has not yet crashed. We are in the process of recreating the
old structure to see if we can reproduce the error. There are no error
messages in our postgres.log. I will gladly provide any log files, core
dumps, make config changes, etc.
Any insight would be greatly appreciated.
Thank you,
Matthew Hagerty