On Sun, 9 Nov 1997, Evan Leibovitch wrote:
> Test the functions
> > open stuff
> opened DB file /tmp/db/stuff: flags=42 fd = 3
> LOG: 0 MAIN
> Failed to get write lock for /tmp/db/stuff: Device or resource busy
This looks very much as though the DB_LOCKS_ITSELF didn't "take".
Advance Warning:
There continue to be a trickle of problems of various sorts with DBM
locking. I am "strongly minded" (as they say), in fact, I think I
probably will, make a change to the way Exim locks DBM files.
The change is to abandon any attempt at locking on the DBM files
themselves, but instead to create an entirely separate file, and do
the locking on that. This means that Exim will have exclusive access
to DBM files before even trying to open them. This will remove any
remaining worries that some DBM libraries "do things" to files at open
time, before the caller can lock on the open DBM file. Also, it will
avoid problems with GDBM and any other library that does its own
locking.
If anybody has any objections or other comments on this proposal, now is
the time to speak up, please!
--
Philip Hazel University Computing Service,
ph10@??? New Museums Site, Cambridge CB2 3QG,
P.Hazel@??? England. Phone: +44 1223 334714
--
*** Exim information can be found at
http://www.exim.org/ ***
