On Tue, 26 Feb 2013, Johannes Kroll wrote:
While trying to load http://toolserver.org/~render/stools/tlg, we got 500 errors first and then "connection reset". SSH to nightshade took 2 minutes or so to connect. Now web & ssh seems to be working again.
At which time did you try about?
Yesterday evening up till early in the morning today, SQL queries were very slow. I did't take measurements but simple page queries that would normally execute instantly would take minutes.
Did you try the whole night? Or which time? And which databases seemed to answer slower? The problem is that the head nodes are doing SQL forwarding too. So if the active one is fishy you might not even have SQL connections. But the phenomenon should have occured between about 0:30 and 1:30 am UTC (1:30 and 2:30 CET). If you tried outside of this timeframe it would be good to know if you had any other errors and what they looked like.
Cheers nosy