https://bz.apache.org/SpamAssassin/show_bug.cgi?id=7500
Adrian Pepper <[email protected]> changed: What |Removed |Added ---------------------------------------------------------------------------- Version|3.4.1 |3.4.0 --- Comment #12 from Adrian Pepper <[email protected]> --- Thanks to everyone for all the help. I am chagrined and appreciate your patience. Everyone has pointed me to appropriate routes of experimentation. I actually do not have access to the actual remote spamd server which exhibits the problem. I believe I misinterpreted some of what those administrators said in the past, and thought they ran closer to stock than is the case. Or perhaps in some cases they have brought forward past defaults via upgrades. And, very sorry, along the way I mixed up the spamc version and the actual remote spamd version. The remote spamd server runs 3.4.0. Obviously the remote spamd uses trusted_networks. I have asked for the details of those settings but not received any replies. I also suspect they are using a statistically incorrectly high negative value for ALL_TRUSTED, but cannot fully verify that. I have experimented in lxc containers on my workstation (3.4.0 and 3.4.1) and, together with reading, it's not clear a private subnet is supposed to be automatically assumed to be trusted for this purpose. I get good results from adding something to match the [fe80::d4bf:9432:cac5:5f3b] to both trusted_networks and internal_networks. (Adding to internal_networks only generates an informational /var/log/syslog message suggesting you add it to trusted_networks too). I have not heard back from my contact with access to the spamd server. My experiments on my containers have the same difficulty triggering RDNS_NONE at all (even for blatant examples), suggesting you need to do something besides just having non-zero rules defined to trigger/enable the test. But at this point I can cause "ALL_TRUSTED" to appear or not, on those experimental containers, more-or-less at will. -- You are receiving this mail because: You are the assignee for the bug.
