On Mon, Sep 22, 2003 at 11:33:30AM -0400, Tom Lane wrote: > Patrick Welche <[EMAIL PROTECTED]> writes: > > I hope I guessed the right syntax... > > % pg_filedump -R 71716 data/base/17148/283342 > > Yes, but this doesn't give all the available info. Add -i and -f > options. A plain -d dump might be interesting too.
Indeed, the plain -d dump says that I have a chunk of /var/mail/prlw1 in 1000-13ff. No wonder postgres complained! Highlight: 0fe0: 06000000 00000000 00000000 00000000 ................ 0ff0: 01000000 3e000000 00000000 00000000 ....>........... 1000: 52657475 726e2d70 6174683a 203c7072 Return-path: <pr 1010: 6c773140 6e65776e 2e63616d 2e61632e [EMAIL PROTECTED] ... 13e0: 38323020 20202020 37313139 38202020 820 71198 13f0: 20323425 20202020 32303637 20202032 24% 2067 2 1400: 3e000000 00000000 03000000 b6090000 >...........¶... 1410: 00000000 00000000 01002418 0f001a00 ..........$..... Would you be interested in the full dump anyway? It seems this is trashed and I need to bring out the backups, right? Next is speculation as to how? I read a very large mail file with mutt which I think uses mmap. It still begs the question how did that end up in the database.. Worth reloading into same database server, or upgrade to current cvs? NetBSD-1.6ZC/i386 with 2Gb memory. Thanks for the help! Cheers, Patrick ---------------------------(end of broadcast)--------------------------- TIP 4: Don't 'kill -9' the postmaster