Wed Feb 7 16:55:31 PST 2001 — Some hours later, we’ve got the errant mail server back online, the others carried the load in the meantime. Early on, we disabled the host that was causing the problem so that it would not affect customers, and we’ve been working since then to isolate the trouble with that host. Kernel instability was initially blamed for the intermittent crashes, but after some hardware shuffling the host appears to be stable and is back online. Our cluster of four redundant mail servers minimizes customer impact, and today at 13:44hrs we simply had the bad luck of losing two at once, which caused temporary problems. No mail was lost during this period. -Eli, Kelsey