@@ -1,6 +1,8 @@ here are a list of upcoming changes or ones I'm thinking about FUTURE CHANGES +[ ] when hardening server, change the SSH port + [ ] do not change the Capify file. This is an issue if you want to use ec2 for staging but something else (engineyard) for production. The issue is that require 'ec2onrails/recipes' cannot be called until after the deploy file has been run because of some loading dependencies. SO, is it possible to not have this behavior, or is this making a mnt out of a mole-hill? [ ] be able to save an ami file after cap ec2onrails:setup is run... this way you can just launch more instances of that specific ami file (can we automate this? ) @@ -33,17 +35,9 @@ FUTURE CHANGES http://pauldowman.com/2008/02/17/smtp-mail-from-ec2-web-server-setup/ also see this file for updates: http://www.babbleon.co.uk/2008/05/email-with-ec2/ - -[ ] preload mysql timezone information (UPDATE: not sure if this is needed...) - -[ ] mysql optimizations are not calculating the num of cores avail correctly - -[ ] right now cron has a task called 'app' in cron.daily, cron.hourly, cron.monthly, and cron.weekly. BUT, these are run on every server. we should probably provide some sort of mechanism so a user can specify particular jobs for particular roles [ ] consolidate all logs to /mnt/log, including /mnt/app/current... it makes it easier to rotate and shuffle them off to s3 -[ ] get all user-data and meta-data variables from the amazon image in a similar way to rightscale? Makes it dead simple to use/manipulate that information. UPDATE: rightscale gave us permission to use their open-sourced files, as long as we credit them - [ ] hook in encryption of backup archives [ ] hook in backup of server... or should we wait for persistent storage snapshots? http://www.webmonkey.com/tutorial/Back_Up_a_Web_Server @@ -72,6 +66,14 @@ QUESTIONS COMPLETED +[X] preload mysql timezone information (UPDATE: not sure if this is needed...NOTE: it is not) + +[X] mysql optimizations are not calculating the num of cores avail correctly + +[X] right now cron has a task called 'app' in cron.daily, cron.hourly, cron.monthly, and cron.weekly. BUT, these are run on every server. we should probably provide some sort of mechanism so a user can specify particular jobs for particular roles NOTE: paul already thought of this with the exec_runner script that we can run from within cron + +[X] get all user-data and meta-data variables from the amazon image in a similar way to rightscale? Makes it dead simple to use/manipulate that information. UPDATE: rightscale gave us permission to use their open-sourced files, as long as we credit them + [X] Move to Echoe? It just seems a LOT easier to setup and handle than hoe. It can remove a lot of cruft, probably including ./config, ./script, ./tasks (or most of them), and ./website (move that to the README file?... looks like would need to be moved to ./docs ). A few folks have branched ec2onrails into github before and they've all done this... [X] set -y flag when installing/updating gems UPDATE: not needed... does it automatically for the other non-interactive flags we are using TODO 093a05d9fe3ccdbbca344291ebec54fbe72af10a Adam Greene adam@6bar8.com http://github.com/skippy/ec2onrails/commit/7a3ff10cd8163c11195549223cb7ee8838cc94db 7a3ff10cd8163c11195549223cb7ee8838cc94db 2009-02-19T12:10:33-08:00 2009-02-19T12:10:33-08:00 adding a few items to the TODO list, and marking a few as either done or no longer needed 6f0967d5efb874ad05e9c3495873bb9e7f7e8683 Adam Greene adam@6bar8.com