[FWIW Looks like mailman is *not* stamping a list reply-to as I set in the configs....]
On Tuesday, April 20, 2004, 5:51:18 PM, William Stearns wrote:
- Over half the size of those zones is in the TXT records. Just changing 'Message body contains domain in sa-blacklist. See: http://www.stearns.org/sa-blacklist/' to 'Blocked, See: http://www.stearns.org/sa-blacklist/' reduced the 'ws' zone size by 33%
Works for me! Jeff, feel free to make that change anytime.
Done. Hope the change doesn't break anyone... Hopefully they're using the A record first, and the TXT for a comment, if at all. :-)
(The default/sample SA 2.63 SpamCopURI and 3.0 urirhsbl rules seem to write their own text descriptions based on the A record, so I think we're ok. Other folks could be using the TXT record however. I'll announce the change and hope we catch them.
I may shorten the sc.surbl.org TXT message also....)
- It's possible to combine the zones but keep the data logically seperate so people can differentiate and adjust scores/policys accordingly. Check out how MAPS does RBL+, the A record returns an "IP address" that is effectivly a bit-mask flag to indicate which MAPS zone the original hit was from (DUL, RSS, RBL, OPS, etc). Look at how the 'check_rbl' and 'check_rbl_sub' routines are used inside SA to pull apart a single DNS query against RBL+ (at least in SA 2.6*, havn't looked at 3.0 yet ;)
No experience with this, so no opinion. Thanks for the ideas, Dave. Jeff, enough people have asked forthe combined list that I'm game to set up an "all.surbl.org" combined list if you are. It really sounds like the technical concerns are all handleable. We can still keep the sc and ws subdomains for those that think my taste in domains is questionable... :-)
We could combine into larger lists with:
1. Different TTLs per record (essentially as now, through the default zone file $TTL) 2. Different A records, i.e. 127.0.0.2 for sc, `127.0.0.3 for ws, etc. 3. Different TXT messsages (as now)
Sounds like keeping all the A records at 127.0.0.2, making the different TTLs and keeping the current TXTs indicating the data sources, while combining all the records into larger lists could work. (I.e. #1, #3 but not #2.)
Comments?
Jeff C.