(On the subject of LDAP queries failing because delivery processes
unbind their parent's LDAP connection.)
* Philip Hazel
> The fix for 4.02 was to insert a call to search_tidyup() before the
> fork, so that each subprocess starts off without copying the cache state
> of the parent.
I just had this bite me with 3.35. Is a solution to this going to be
backported to Exim 3? If it's not, I'll tell My Favourite Vendor to
patch the Exim 3 they ship.
(As an aside, I patched Exim to include the pid of the "owning"
process in the cache state, which seems to work for me.)
Arne.
unbind their parent's LDAP connection.)
* Philip Hazel
> The fix for 4.02 was to insert a call to search_tidyup() before the
> fork, so that each subprocess starts off without copying the cache state
> of the parent.
I just had this bite me with 3.35. Is a solution to this going to be
backported to Exim 3? If it's not, I'll tell My Favourite Vendor to
patch the Exim 3 they ship.
(As an aside, I patched Exim to include the pid of the "owning"
process in the cache state, which seems to work for me.)
Arne.