Leaping seconds and looping servers
Posted Jul 4, 2012 1:48 UTC (Wed) by leromarinvit
Parent article: Leaping seconds and looping servers
Now I finally know what happened to my file server/router a few days ago! A few minutes after the leap second, Nagios sent me a warning about the load being somewhere around 15. I didn't think much of it, since the nightly backup sometimes does that. The next day I couldn't log in over SSH any more, and some time later dnsmasq stopped serving DHCP requests. Killing everything via sysrq finally brought it back to its senses, but at the expense of all the services normally started by init, so I just rebooted in the end.
There was a Java program running that updates a timestamp in its GUI every second, so that probably caused it. I initially thought said Java app had just gone mad and used up all memory, causing the box to swap itself to death, because I saw the HDD light flickering like mad. Of course, this being Java, it quite possibly did allocate some memory every time the counter fired, saving the GC for later...
For reference, this is a Ubuntu 10.04 box running 3.4.3-ck.
to post comments)