On Jun 30, 2012, at 17:58, Ask Bjørn Hansen wrote: > Looks like about 250 servers "missed" the leap second. > > This query counts how many servers had an offset 100ms around "one > second off". It was basically none yesterday, and 253 in the first > hour after the leap.
Compared to the last leap second it looks like more servers are sorting it out in a hurry. In the first hour about 8% had it wrong, but the number is halved each hour. The updated numbers are below. At my work some Java processes (for Elastic Search) crashed (hung/used all the CPU/some variation of that) just after the leap second. Reading serverfault.com it seems like some Linux kernels have been crashing, too. mysql> select date(ts),hour(ts),count(distinct(server_id)) from log_scores where offset > 0.9 and offset < 1.1 and ts > '2012-06-30 21:00:00' and monitor_id is null group by 1,2; +------------+----------+----------------------------+ | date(ts) | hour(ts) | count(distinct(server_id)) | +------------+----------+----------------------------+ | 2012-06-30 | 23 | 1 | | 2012-07-01 | 0 | 253 | | 2012-07-01 | 1 | 104 | | 2012-07-01 | 2 | 46 | | 2012-07-01 | 3 | 27 | +------------+----------+----------------------------+ 5 rows in set (0.60 sec) Ask _______________________________________________ pool mailing list pool@lists.ntp.org http://lists.ntp.org/listinfo/pool
