Hi, while running a nohup <job> & my disk filled up due to an unexpected amount of output in nohup.out. The job included lots of simple SELECT queries, but no changes in the data. Now the pg_log file seems to be damaged. If I try to vacuum the database I get the error: ERROR: cannot read block 1630 of pg_log. When I try to dump the database with pg_dump I get: BEGIN command failed. When I try to create a table I get: AbortTransaction and not in in-progress state. I can still SELECT data under psql, but only for one SELECT command. If I do a second SELECT I get again: ERROR: cannot read block 1630 of pg_log. Every time I get an error I have to kill the postmaster process or stop & start postgres. Can I - somehow repair pg_log? - dump the database without access to pg_log? - delete pg_log? - restore pg_log from my last backup? Any suggestions? Thanks -- Thomas Metz GSF - National Research Center for Environment and Health
