what about the deadlocks?
Posted Jul 3, 2012 23:59 UTC (Tue) by mcisely
In reply to: what about the deadlocks?
Parent article: Leaping seconds and looping servers
I personally experienced the deadlock issue.
I spent 3+ unplanned hours at work last Saturday afternoon trying to figure out why our main server kept wedging itself every time it transitioned from single user to multi-user mode. I had no idea about the pending leap-second at the time and had been suspecting storm damage from the previous night. I finally clued-in when I found that the kernel hang would happen every time about 5-8 seconds after starting ntpd.
It was running Debian Lenny (yeah, old) with a Debian stock 2.6.32-5-amd64 kernel. Server architecture is a multicore Athlon setup.
The hangs started happening at about 7:43AM CDT. Once I figured out that trigger was ntp, I disabled that, then searched the web and found this:
Monday morning I enabled ntp again and started it - and as expected, no lockup...
to post comments)