Date: Sat, 29 Sep 2007 22:49:14 +0200 From: "Merot Kassad" <merot.kassad@gmail.com> To: freebsd-questions@freebsd.org Subject: Is gmirror safer with load algorithm than round-robin? Message-ID: <df502c590709291349j5cf10ffas56e28ce5f468e720@mail.gmail.com>
next in thread | raw e-mail | index | archive | help
List, I've been running into problems with gmirror and weak SATA controllers that appear to not handle the rapid simultaneous traffic to different disks.(Causes crashes.) I've seen other people have had these problems as well, and was wondering if anyone here knows if I could expect a difference between the "load" and "round-robin" algorithms with regards to this problem. Looking at gstat I see that with "round-robin" the reads to the mirror are distributed half and half to each provider, while with "load" I'm able to track the switching back and forth between the providers with the naked eye so to speak. My reasoning is that this should stress the controller less, could this be the case? Also, I'm wondering if anyone here have tested the recent patch by Mykola Zubach to the gmirror "load" algorith, and can relate their experiences. Sincerely, Merot
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?df502c590709291349j5cf10ffas56e28ce5f468e720>