I experienced this irregularity on a i686/SMP host using Linux
184.108.40.206-vs2.0.1-rc1 (a rediff, quite cleanly):
Maybe this counts:
It happened just a few seconds after booting including a few legacy
vservers and stopped after about a minute.
After converting the vservers' config to vs2.0, this never happened
again, neither it ever did on a test server (vs2.0 vservers only) using
the same kernel image.
Nevertheless the system runs just fine and survived all stress tests.
Vserver mailing list