Date: Wed, 28 Aug 2013 15:06:59 -0700 (PDT) From: Robert_Burmeister <robert.burmeister@utoledo.edu> To: freebsd-stable@freebsd.org Subject: Re: Suggest changing dirhash defaults for FreeBSD 9.2. Message-ID: <1377727619133-5839769.post@n5.nabble.com> In-Reply-To: <kvkh1j$7fj$1@ger.gmane.org> References: <521C9E85.4060801@UToledo.edu> <CAE-mSOLCYRM0LRLRgmaEZN1u5ozttJZC3kWtw3Zarqik1N29zw@mail.gmail.com> <521D7552.5080008@UToledo.edu> <kvkh1j$7fj$1@ger.gmane.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Ivan Voras-7 wrote > On 28/08/2013 05:58, Robert Burmeister wrote: >> >> On 8/27/2013 9:40 AM, Sergey Kandaurov wrote: >>> On 27 August 2013 16:41, Robert Burmeister >>> < > Robert.Burmeister@ > > wrote: >>>> I believe that increasing the following values by 10 would benefit >>>> most FreeBSD users without disadvantage. >>>> vfs.ufs.dirhash_maxmem: 2097152 to 20971520 >>>> vfs.ufs.dirhash_reclaimage: 5 to 50 or 60 > > I'll bump vfs.ufs.dirhash_reclaimage to 60, it's worth it. Note that vfs.ufs.dirhash_reclaimage is not used to shield ufs_dirhashmem entries, rather all entries older than reclaimage are evicted from the cache on vm_lowmem before evicting a percentage of the oldest remaining entries. If it has been hours since a kernel vm_lowmem event, most of the entries in the cache can be lost due to a momentary memory spike. Also note, under kernel memory pressure continuous vm_lowmem events will effectively extinguish the ufs_dirhash cache, returning its memory to the kernel. I believe this prevents the ufsdirhash cache from in any way contributing to a kernel panic, so bumping vfs.ufs.dirhash_reclaimage to 60 should be riskless. -- View this message in context: http://freebsd.1045724.n5.nabble.com/Suggest-changing-dirhash-defaults-for-FreeBSD-9-2-tp5839351p5839769.html Sent from the freebsd-stable mailing list archive at Nabble.com.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?1377727619133-5839769.post>