On Sat, Jul 13, 2013 at 06:39:28PM +0200, Ian FREISLICH wrote: > Konstantin Belousov wrote: > > > Yes. This state of affairs doesn't happen on r251445 and further > > > testing on my side shows it doesn't hapen on all my amd64 servers. > > > It appears that this particular server type (Dell R200) running > > > amd64 with geom_mirror is affected. I will have to test further > > > by destroying the mirror and removing it from the kernel and see > > > if I can still reproduce the issue. Perhaps r251446 exposes > > > insufficient locking on opperations affecting these variables. > > No. The lorunningspace is constant for the system lifetime. > > It can only be changed by the sysctl vfs.lorunningspace. > > Look into /etc/sysctl.conf or scripts which apply sysctl settings. > > Wipes egg from face. > > /etc/sysctl.conf: > > vfs.hirunningspace=4194304 > > So, then did r251446 actually start using this value or did other > values get significantly tuned up? I recall now setting this nearly > a year ago when we did our ZFS tuning and it was a four fold increase > on the defaults. r251446 optimized the wakeups by only doing wakeups when runningbufspace actually crossed the lorunningspace. Before, wakeups were performed always on the runningbufspace changes. For ZFS, these settings are completely irrelevant, ZFS does not use buffer cache.
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:40:39 UTC