2015年05月26日 エンジニアリング

Elasticsearch and the 2015 Leap Second

By Antonio Bonuccelli

We're approaching that time of the year when a day will actually be longer than a normal day, by just one second.

That is what the scientific community refers to as the “Leap Second.”

A Leap Second is a one-second adjustment that is occasionally applied to Coordinated Universal Time (UTC) in order to keep its time of day close to the mean solar time, or UT1.

The next scheduled occurrence of leap second will be on June 30, 2015 23:59:60 UTC.

While the leap second is certainly an interesting subject for discussions among scientists, it has proven to be a big headache for an IT infrastructure.

The most recent occurrence of the leap second in 2012 disrupted operations for many major companies, bringing down services by causing high cpu usage on production servers.

How does that affect my Elasticsearch deployment?

The main problem actually sits in Linux kernel layer. Most Linux vendors have published guides explaining what you need to do (if anything) to test and prepare for the upcoming leap second:

Please refer to your distribution's support channels if it is not listed above and for additional details. Elastic recommends you follow your distribution's recommended guidelines for handling this issue in order to avoid any further complications that may arise from unofficial or ad-hoc fixes.

Additionally, some of the major cloud providers have also published their own information which you should read if you are running on one of their platforms: