I'm running exim with gdbm without a problem, and I didn't have to do
any of these tweaks. However, I did have to recompile gdbm and set the
flock define in autoconf.h to 0. Redhat linux ships gdbm with flock()
locking whereas Debian appearantly uses fcntl().
On Mar 22, Ian Pallfreeman <ip@???> wrote:
> I'm running exim 1.61 with gdbm as the database manager. I've patched db.c to
> allow gdbm to do its own locking (by removing the exim locks), tweaked it to
> use O_RDWR|O_CREAT rather than O_RDONLY when a dbm file isn't found, and made
> the retry algorithm smarter by using a random delay rather than each potential
> writer using the same sleep times. This has cured most of my "can't open DB
> file wait-smtp" problems, but I'm still seeing them when attempting to relay
> messages to a large number of addresses.
>
> The dbm code in transport.c seems OK (except for a minor error-handling fix):
> there don't seem to be any paths through the code which might leave the dbm
> file open at exit.
>
> Anyone got any further suggestions?
>
> Ian.
--
Artificial Intelligence stands no chance against Natural Stupidity.
GAT d- -p+(--) c++++ l++ u++ t- m--- W--- !v
b+++ e* s-/+ n-(?) h++ f+g+ w+++ y*
