Probably should have posted about this a couple of days ago, but after spending 4 days of 20 hours days chasing down one little issue after another with the migration, I took the weekend off :)
So, yep – we migrated to the new hardware finally!
Now Seesmic is running behind some new load-balanced file servers and the recording cluster has been increased (and also now load-balanced) but the big thing behind the migration was that we are now running everything thru the Ruby (Nginx + Merb + Memcached) API/Proxy. This means that all client requests (ours and any 3rd party) are handled and routed by the file/proxy cluster.
Sure does make it easier to log and monitor when there is a single entry point now.
I need to create an entry or two on some of the curious things we had to do with Nginx and the Gluster DFS to get all of our various API calls to work and to also allow the older URL routes to static resources to not break. Woo legacy URLs!
We had some glitches with the code but by far the biggest glitch was not realizing that our DNS provider would not allow us to change the CNAME for www to point to any server except those it managed! Waaa! yea, we had to scramble and come up with a www2 -> www proxy on the old server until we got the Nameserver changed for seesmic.com and wow was that a long 6 hours waiting for all the entries to propagate.
but yea, migration over – now on to adding more features :)
Technorati Tags: news, seesmic
