On 10.12.2010, at 0.54, Stan Hoeppner wrote:
However, this still doesn't seem to explain Ralf's issue, where the kernel stays the same, but the Dovecot version changes, with 2.0.x causing the high load and 1.2.x being normal. Maybe 2.0.x simply causes this bug to manifest itself more loudly?
Cor's debugging has so far shown that a single epoll_wait() call can sometimes generate a few thousand voluntary context switches. I can't really understand how that's possible. Those epoll_wait() calls about half of the total voluntary context switches generated by imap processes. We'll see tomorrow if poll() works better or if a small patch I made makes it better.
This Linux kernel bug doesn't explain the high load reported with 2.0.x on FreeBSD either.
Who has reported high load on FreeBSD? So far I know of only Ralf and Cor and both are using Linux.