Hi all,
The problem:
I do a large bulk copy once a day (100,000 records of Radius data),
tearing down indices, truncating a large table that contains summary
information, and rebuilding everything after the copy. Over the course
of this operation, I can generate up to 1.5 gigs of WAL data in
pg_xlog. Sometimes (like just now), I will run out of disk space and
the postmaster will crash. I try to restart it, and it errors out.
Then I delete all the WAL logs, try to restart, and (surprise) it errors
out again.
I tried to set some of the of the WAL parameters in postgres.conf like
so:
wal_buffers = 4 # min 4
wal_files = 8 # range 0-64
wal_sync_method = fdatasync # fsync or fdatasync or open_sync or
open_datasync
but I get 24+ separate files.
I would like to recover without an initdb, but if that isn't possible, I
would definitely like to avoid this problem in the future.
Thanks to all