Back after dithering with DNS, chasing Mongrels and Googling the email!
Posted by nito, Thu Mar 15 23:42:00 UTC 2007
I won’t claim to be a sysadmin, in fact, I’ll willingly announce that ‘I am NOT a System Administrator’, but I will happily sign up to the belief that a first class one is worth at least 5 years worth of expensive rackspace with 24/7 support and maximised bandwidth for a server-farm of at least 10 top-end Apple Xserves. I’m nearly serious.
Right now though, I wish I was a sysadmin in the same way I wish I knew how all the bits in my car worked (and finding -that- out would probably lead me to repairing the inside light without having to pay! ). But, getting back to the point… …It probably wasn’t that noticeable, but we were gone for a bit. Seriously absent. No longer here. You may notice the gap in dates. On the other hand, you may not have noticed or even cared. if I’d been a good sysadmin, or had one handy, no doubt these problems would never have occurred. especially not “the case of the missing dot!”
It all started a few days ago. ‘Time to update the blog’ I thought. ‘Time to add some links, perhaps a post, add that Paul Ingles flickr plugin, but I couldn’t post into the blog. Any attempt to get to neverintheoffice.net/admin returned an error and my logs were peppered with the likes of:
[error] proxy: BALANCER: (balancer://mongrel_cluster). All workers are in error state
It was eventually fixed by a visit to mongrel/apache.html, modifying some code, making sure everything was shut down and then issuing the following commands:
$ mongrel_rails start -d -p 8000 \
-e production -P /full/path/to/log/mongrel-1.pid
$ mongrel_rails start -d -p 8001 \
-e production -P /full/path/to/log/mongrel-2.pid
Everything seems to be working now. I know there’s a lot of oddness and inconsistency as to how I’ve set this all up and sometime over the next week I will revisit this (and perhaps write about it).
But that wasn’t what brought us down. it was the previously mentioned missing dot. After playing around with postfix which brought about faint memories of Qmail I decided ‘hang this’ and, after discovering this: google apps with your vps, decided to take a look at the google way. I was encouraged by the fact that Larry Myers (another new Slicehost user) with the same sort of mail requirement and reasoning as me:
“My biggest motivation for doing this was not wanting to deal with the learning curve that comes with setting up Postfix and Spam Assassin.”
So, I travelled over to the googlesphere and joined up. Right now it’s a very satisfactory solution to a small need. I’m wondering how it will scale, and when it will show itself to be insufficient. Watch this space.
“But what about the missing dot?” I hear you cry out. Oh yes. The missing dot. Well, it’s simple really: when you go and sign up for your google apps account, do what Larry Myers tells you to do:
“When Google asks to verify your domain I’d suggest just going with the HTML check, as it’s much easier than dealing with changing DNS records.”
He’s right. Because if you mess up, like I did, you can end up with neverintheoffice.net.neverintheoffice.net and sink your domain below the watermark, all for want of a missing dot. Of course, the missing dot is just one potential problem amongst many.
Oh, and of course you can now email me here: nito@thisdomain.net (change thisdomain to neverintheoffice).
