2009/5/15 Scott Haneda <talkli...@newgeo.com>: > Any suggestions on the simplest way to approach this. I was thinking > postfix with MySql backed data store. Today I read that RHEL is behind on > postfix, and I think does not have MySql support in their rpm's. I have > zero access to a staging server.
Correct, you'd have to roll your own or use the Centos-plus channel RPM. If you choose the latter you can have a very high degree of confidence that it will just work as you expect, but you need to maintain it yourself. Postfix 2.3 is otherwise fine for functionality in RHEL5 though. > If this turns into a high volume site, would file based aliases fall apart > after a certain amount? I also see maintaining a alias mapping via a file > managed by a web service to be prone to error. If anything I wold store the > mappings in a database, and write them out clean on schedule. What are the > upper limits of how many forwards I should feel comfortable maintaing as a > local file? With enough sanity checks you can manage an alias file with scripts (run via web frontend), but it's not much fun. I believe (redhat) default hash-maps perform and scale quite nicely. CDB maps are said to scale even better, and I think numbers quoted on this list previously say... 1 million is no problem for CDB? If you go this route you probably want a couple of sanity checks to make sure that the new map file isn't drastically different (smaller) to the current running one. I can just imagine a situation with some sort of temporary DB failure producing zero lines of output, which is then promptly used to create a new map...
