April 3, 2011 § Leave a Comment
As you might have noticed, we are experiencing some downtime on mathblogging.org — once again we’ve hit the limits of our bandwidth.
We’re looking into the causes and we’ll be back with an update as soon as possible.
Sorry for the inconvenience.
Update: We’re still not absolutely sure what happened.
The cause of our outage is simple: we had depleted our daily quota for both CPU hours and incoming bandwidth. That’s different from our previous downtime where the sudden popularity via yCombinator led to the depletion of our outgoing bandwidth.
From what we can see right now, two reasons come to mind. The plausible one is that our update last week came with a bug; we probably did not estimate the effect of the new functionality correctly. This thought is supported by the steady increase in bandwidth use since our update. In fact, we only now noticed two shorter outages that we missed both yesterday night and the night before that. Ironically, it might be that the bug came with our efforts to reduce bandwidth (via more efficient caching etc). But such is life.
The much less plausible scenario would involve blaming somebody else, i.e., somebody else’s misdirected script activated our (unprotected) update mechanism for the feeds; given what we can see so far, this scenario does not seem likely but who knows.
Right now we’ve just spend some time to write fixes for both scenarios. Local testing looked good, but we’ll have to wait until the daily quota resets in a couple of hours to see what the reality of things looks like.
We’re sorry if this caused anybody any inconvenience. We’ll keep you updated here and on twitter.