Remember when Linden Lab disabled assorted services on Sunday with the intention of stabilizing the Second Life grid through its weekly peak period? Well the big switch (actually, that is something of a misnomer, as the shutdown and startup processes for the subsystems seem to be quite complex) was thrown again for Friday.
This time, an extra subsystem was thrown into the mix. Logins. That's right, as the grid came up to the busiest time of the day, Linden Lab made the call to prevent any more users from logging in.
As user concurrency hit about 56,208 at 12:45PM SLT (US Pacific time), all the group and profile functions from Sunday's partial close-down were switched off, along with the Lindex, transactions, account pages and the Support Portal.
Within an hour, user concurrency had plummeted to 31,996 1t 1:40PM. Linden Lab began re-enabling some subsystems at 1:20PM - but it wasn't until just after 1:40 that logins came back online. And it was close to two hours later that things had returned to relative normal.
Not that things were much smoother after that. The grid suffered at least two bursts of login failures after that, along with transaction failures and other symptoms.
Turning off much of the user experience (and logins) to improve the user experience. That's novel.
