gah - mail problems
Dec. 1st, 2004 08:47 pmSometimes knowing a little is a dangerous thing. I just spent an hour trying to figure out why my mail to the Debian package maintainers of SCIM was getting bounced by SPF. (They need to nudge scim and scim-chinese into testing together, because each is blocking waiting for the other.)
Turns out it's mostly because -- I think -- the sysadmins at my lab haven't put a record in for the SMTP server at the main university, so if I send mail "from:" my lab account, but using the main campus SMTP server, I get an error from this page. grr.
It was made worse by forwarding at debian, which leaves the file marked as "from:" my lab address, but now the MX machine is actually master.debian.org, which is definitely not in the SPF record for my lab. (there's a long discussion about SPF on the debian-devel list.)
So I read a lot of documentation and played with various SMTP servers to try to work out what was really going wrong. I'm pretty convinced now that (1) the record at the lab should include mail sent from the main campus computers and (2) the whole SPF idea is a bit cockamamie. Whose idea was this SPF stuff? What's the real solution it's trying to make here? seems to me that crypto is the only way out of this, and this solution is worse than a band-aid.
The worst bit is, if I was a little more ignorant, I would have just given up after a few minutes -- at most, I would have filed a bug with my sysadmin. But instead I hacked on it for 90 minutes and learned a lot of useless information. gaaah.
Turns out it's mostly because -- I think -- the sysadmins at my lab haven't put a record in for the SMTP server at the main university, so if I send mail "from:" my lab account, but using the main campus SMTP server, I get an error from this page. grr.
It was made worse by forwarding at debian, which leaves the file marked as "from:" my lab address, but now the MX machine is actually master.debian.org, which is definitely not in the SPF record for my lab. (there's a long discussion about SPF on the debian-devel list.)
So I read a lot of documentation and played with various SMTP servers to try to work out what was really going wrong. I'm pretty convinced now that (1) the record at the lab should include mail sent from the main campus computers and (2) the whole SPF idea is a bit cockamamie. Whose idea was this SPF stuff? What's the real solution it's trying to make here? seems to me that crypto is the only way out of this, and this solution is worse than a band-aid.
The worst bit is, if I was a little more ignorant, I would have just given up after a few minutes -- at most, I would have filed a bug with my sysadmin. But instead I hacked on it for 90 minutes and learned a lot of useless information. gaaah.