Hacker News new | ask | show | jobs
by davidgerard 3695 days ago
We got bitten by this back in the Solaris days in 2009, on a TV broadcasting production box with quite stringent uptime requirements: what happens is the defragmenter gets itself tied in knots and starts thrashing, and the symptom is 50% system CPU with no apparent cause. Got a Sun kernel engineer on call and all. SPOILER: it did in fact require a reboot to unfuck the system. Then we kept the disk in question being wasted.
1 comments
ryao 3695 days ago
A reboot implies a bug had to be fixed. That can be assumed to be fixed everywhere by now.
link
davidgerard 3693 days ago
... no, it doesn't imply anything of the sort. As far as I know, ZFS still has this issue - it can get itself tied in knots, and only a reboot will stop this from happening.
link