I don’t think it would be an exaggeration to say that we’ve had availability problems lately. In fact, we’ve had four problems, and three of them may be fully addressed.
First we had memory issues, which I didn’t understand at first. Y’all would have seen them as memory errors or lengthy timeouts when submitting comments. The solution turned out to be pretty simple, and that issue is by now long since dead.
But: That solution would have been masked, to the untrained eye, by problem number two. The account all of the Splendorquest.com domains live on had been set to 25GB, max, back when we lived on semi-dedicated server. This wasn’t changed when we moved, with the result that we’ve been thrashing for disk space for a couple of months. Again, an easy solution once the problem was discovered.
I said nothing about these two because I still haven’t solved problem number four — which used to be problem number three — a significant overcommitment of our MySQL server.
But, in the meantime, we got hit with problem number three, a three-day denial-of-service-like attack. The villain was probably an itinerant spammer, but the effect, from your point of view, was just like a DOS action: No action on your end.
Meanwhile, problem number four persists, but in a seemingly calmer state of exigency. We’re serving a lot of folks when the sun is up over North America, and we’re shipping 200GB of data every month. Put this all under the category of growing pains, but it remains that our growth has put us in this kind of trouble four times a year, at least, for three years running.
And even with all of that, comes today a note from Mark Madsen congratulating BloodhoundBlog for making it back up to a PR6. We’ve been there before, so this may just be temporary, but it’s doubly amazing given our late semi-compromised state.
Anyway, thanks to everyone for the thought and effort — and the links — you bring to BloodhoundBlog.Related posts: