I am running PostgreSQL 8.2.4 on Windows 2003 Server. I have a setup where WAL files are archived to another drive on the same machine. The command looks like this: archive_command = 'COPY "%p" "F:/recovery/%f"' This works well, but I often get the following error in the log: [2009-04-17 12:07:34 508] LOG: archive command "COPY "pg_xlog \000000010000035600000006" "F:/recovery/000000010000035600000006"" failed: return code 1 [2009-04-17 12:07:34 508] WARNING: transaction log file "000000010000035600000006" could not be archived: too many failures I checked and noticed that the handle to that file is being held by a postgres.exe process that corresponds to a particular user in the database who's been logged in since 2009-03-25 (a little less than month) and is in <IDLE> status. If I restart the database usually the connections disappear and the WAL gets archived correctly. However, I am not able to restart this database as it is a production database and I am wondering why these idle connections hold an open handle that seems to prevent archiving? Once this file is unable to be archived, no other files get archived (unless I remove its .ready file manually). I have another server with the same postgresql.conf parameters that does not exhibit this behavior. Any help is appreciated. Thanks. Nikola -- Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-general
