On 31.07.21 16:10, Florian Hars wrote: > I am currently slightly confused by the interaction between > recipient_delimiter and virtual_alias_maps.
That interaction may actually not have been the true cause of my observations. What I now think happened is that the effects I observed were caused by the fact that postfix caches the result of address verifications even over changes in the configuration and the virtual alias maps so that after my first failed test with a virtual alias map that did not yet contain that alias, none of the later experiments with a fixed configuration ever produced a correct result because postfix still remembered that the first test with that address and a broken configuration failed. There are only two hard problems in IT... Regards, Florian.
