BUG #1809: invalid pointer on pg_ctl reload
The following bug has been logged online:
Bug reference: 1809
Logged by: Chris Hoover
Email address: chris.hoover@companiongroup.com
PostgreSQL version: 8.0.3
Operating system: RHEL 4 AS Update 1
Description: invalid pointer on pg_ctl reload
Details:
I was working with my pg_hba.conf file and did a pg_ctl -D <path> reload. I
received the following message, and postgresql shutdown.
[postgres@edipgtst1 eedi]$ pg_ctl -D /database/data/eedi/ reload
postmaster signaled
[postgres@edipgtst1 eedi]$ *** glibc detected *** free(): invalid pointer:
0x082ced00 ***
To make changes to my configuration in 8.0.3, I am having to always to
restarts instead of reloads.
PostgreSQL was installed from the rhel 4 rpms provided on the PostgreSQL
mirrors.
"Chris Hoover" <chris.hoover@companiongroup.com> writes:
I was working with my pg_hba.conf file and did a pg_ctl -D <path> reload. I
received the following message, and postgresql shutdown.
[postgres@edipgtst1 eedi]$ pg_ctl -D /database/data/eedi/ reload
postmaster signaled
[postgres@edipgtst1 eedi]$ *** glibc detected *** free(): invalid pointer:
0x082ced00 ***
What non-comment entries do you have in postgresql.conf?
It's also possible that the problem is being triggered by an unusual
entry in one of the other config files, such as pg_hba.conf. Please see
if you can narrow down what's causing this.
regards, tom lane
[Message body could not be decoded - encoding: 7bit, error: This message cannot be decoded as _entire_ message, try calling #decoded on the various fields and body or parts if it is a multipart message.]
Import Notes
Resolved by subject fallback
"CHRIS HOOVER" <CHRIS.HOOVER@companiongroup.com> writes:
I was working with my pg_hba.conf file and did a pg_ctl -D <path> reload. I
received the following message, and postgresql shutdown.
[postgres@edipgtst1 eedi]$ pg_ctl -D /database/data/eedi/ reload
postmaster signaled
[postgres@edipgtst1 eedi]$ *** glibc detected *** free(): invalid pointer:
0x082ced00 ***
I was playing around and had most entries uncommented, even if I did not
change the values. I am attaching the cleaned up postgresql.conf and my
pg_hba.conf files for you.
Got it --- we had broken the handling of an explicit specification
timezone = unknown
in the config file. Thanks for sending along the test case.
regards, tom lane