On Jun 08, 2011, at 04:20 PM, Odhiambo Washington wrote:
Is a valid MX record a MUST for this project?
My test box is inside a 2Mbps symmetric connection from anywhere (USofA, etc) but it does not have a valid MX record. Will that do? If it must have, then I will need to register a domain name and talk to the ISP to do the mappings.
However, if this test is non-intrusive to Exim, then I can use one of the production servers I use.
PS: My hope is that we'll be using Exim as the MTA:)
As long as you can send email to foo-list@example.com, I think that's all that matters for right now.
My idea is that we'd bring up a test domain with a couple of test lists, and let people subscribe and unsubscribe via email (e.g. foo-join@example.com and foo-leave@example.com). Then people could post test messages to the list and we'd all see how MM3 performed under real-world conditions.
A few other things:
My main goal right now is to get the core to releasable state by 11.11.11. This probably means it's got to go to beta fairly soon, probably no later than the end of the northern hemisphere summer (probably best right after GSoC ends).
We don't need to wait until the web ui is ready, so we'll start with just the email interface.
Probably best to run the system off of bzr trunk. That way, you can much more quickly deploy bug fixes during this test phase.
I actually don't need shell access if you're willing and able to maintain the system, provide logs if necessary, update the bzr branch, etc.
I think the risk is small, but you should be prepared to deal with any runaway mail storms. Since most of the queuing architecture is a fairly straight port from MM2, and we have at least a *decade* of experience with that subsystem, we should be in good shape. But the MM3 code has been refactored and updated to more modern Pythonic idioms, so it's entirely possible that some catastrophic bug was introduced. Don't be afraid, but be prepared. :)
-Barry