Not logged in
Log in now
Create an account
Subscribe to LWN
An "enum" for Python 3
An unexpected perf feature
LWN.net Weekly Edition for May 16, 2013
A look at the PyPy 2.0 release
PostgreSQL 9.3 beta: Federated databases and more
Leaping seconds and looping servers
Posted Jul 3, 2012 15:29 UTC (Tue) by ejr (subscriber, #51652)
Posted Jul 3, 2012 17:16 UTC (Tue) by Ringding (subscriber, #34316)
Posted Jul 7, 2012 1:36 UTC (Sat) by BenHutchings (subscriber, #37955)
The known bugs involving kernel hangs are:
1. Deadlock in printk at midnight (introduced in ???, fixed by commits b845b51, fa33507 in 2.6.29, 184.108.40.206, Debian package version 2.6.26-20)
2. Possible live-lock when NTP sets up the leap second (introduced in ???, fixed by 6b43ae8 in 3.4)
When I saw the fix for bug 2 it appeared that the live-lock being described was introduced by bd33126 etc. (post-3.3) and therefore not possible in earlier versions, but I'm no longer certain of this.
So I'm still hoping to find out what went wrong in 2.6.32 and in 3.2 so these can be fixed in stable updates.
Posted Jul 3, 2012 18:32 UTC (Tue) by smoogen (subscriber, #97)
My Fedora laptop running Chrome went 100% load during the leap second. My Fedora and RHEL server just running apache didn't.
However other RHEL and Fedora boxes reported issues depending on the software that was being run at that time.
Posted Jul 4, 2012 11:05 UTC (Wed) by man_ls (subscriber, #15091)
The funny thing was that after stopping both browsers everything went back to normal, but restarting any of them CPU load went up again. I had never seen something like this, so I took it as a divine sign telling me to go to bed and didn't look back... until the first LWN report.
My Debian stable SheevaPlug server displayed the deadlock, or something like it. That is a lot of bugs for a tiny second!
Posted Jul 4, 2012 9:09 UTC (Wed) by ftc (subscriber, #2378)
The installation on CentOS was affected by the problem (i.e. java started using an insane amount of CPU due to timers expiring immediately), while the installations on Debian appeared to work fine.
Posted Jul 4, 2012 17:38 UTC (Wed) by Lennie (subscriber, #49641)
So maybe you have 32-bit Java on Debian and 64-bit Java on CentOS ?
Copyright © 2013, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds