On Mon, 11 Jun 2001, Paxton, Jon wrote:
> The message id at the top *is* the message id of the filename
> (155m7y-0004yd-00-D). This message id doesn't appear to be in any of the 3
> exim logfiles.
That supports the following hypothesis: Exim creates the file and writes
in the message id (which is the first thing it does). Before it can
write anything else, it crashes. Somehow the file pointer gets
scrambled, so that junk in the I/O buffer following the message id gets
written to the file.
> I can send you the full-file offlist if you want. All the files created by
> this bug are different but exactly the same filesize (that filesize seemed
> to be different on the different servers we experienced this problem on).
No, I don't think I need to see the file. The fact that the file sizes
are the same also supports my hypothesis.
> exim was compiled using gcc without any special gcc flags other than those
> exim's makefile includes itself. it was compiled on the same machine as the
> 3.16 build we are now using that has no problems.
I'll do a comparison in that area of the code, though I don't think
anything was changed (but I may be wrong).
Here's another possible test: re-build 3.16 and try that. This will test
whether something else on your system has changed.
Oh, and are you using embedded Perl?
I fear this is going to be hard to track down.
--
Philip Hazel University of Cambridge Computing Service,
ph10@??? Cambridge, England. Phone: +44 1223 334714.
