So last night I stayed up till something like 0145 this morning working on Mike.Geiger.ca. I like that I’ve got something up there finally, I don’t intend for it to be nearly as dynamic as the ‘blog, but at least it’s pretty now. What I’m really surprised about is the reactions to the photos; some people love it (I do! It’s my new favourite photo of me), and others think it’s horrible… and Stef says it looks like a photo for a German airline – WTF? At this point I’m still very open to suggestions/comments/criticism.
Not 20 minutes after I’d fallen asleep, Thomas phones me. I crawl out of bed just to miss the call, and in my sleep-stupor, I decide he hasn’t left me a voice mail so I crawl back under the covers. Then the voicemail alert goes off: “Power’s out.” Doh. I call him back and neither of us are very coherent… I think we talked about trying to make his desktop UPS be quiet but it never occurred to me to tell him to just turn the unit off. SSH’d in and set mWare to do auto-shutdown. At 0255 (according to my Mike) I called Ottawa Hydro, where the automated message rattled off almost every street downdown from Elgin to Kent. This was clearly not a minor failure… got me worried that they were doing maintenance that might revolve around to the south-end. What do I care? Sleep. Then just before 0400 the power came back on, but I decided not to do anything till later. Once I crawled out of bed at 0900 (keep in mind I’ve been getting up around 0715 lately) everything was running reasonably well. Collocution had to be told that it’s not part of an /8, eclipse needed a rule for 10.100.36.0/24, and I needed to restart xchat for the TimeFysh, dircproxy and apache w/SSL on deuten. The big hiccough was that chisisi doesn’t auto-power on, so I took care of that before heading to work.
With fresh pizza in hand (free drinks because I fixed their Interac terminal), I was off to work for the afternoon/evening. More MySQL head-smashing, turns out you have to warm up the engine before you can assault it will 100s of connections. Also seems that the InnoDB engine can only handle so much concurrency, then it just hangs the connection threads. At least we know about these problems now. Hopefully by the time the production system gets to that kind of load, the developers will have finished changing the code to use Hibernate and we can switch to Postgres or Oracle for a backend. You know, _real_ RDBMes. I did manage to write some scripts that you can, by issuing a single command, can have a MySQL instance configure and sync itself up as a replication slave. Felt pretty accomplished with that one.
Since I’m so easily amused… and the KVM switch on my rack has a really bright blue LED on it…
