On Thu, Jun 25, 2009 at 04:55:14PM -0400, David Collier-Brown wrote: > johan...@sun.com replied > > I disagree with the premise that the free memory calculation isn't > > useful because it doesn't count ZFS caches as free space. > [big snip] > > The space that's consumed by ZFS is in use and can't simply be peeled > > off a free-list and used immediately. Adding a variable that counts > > space used by ZFS as free is actually going to be more confusing. > > I worry that we're making a silent change to "all memory is > considered used" again. If we are, we will have to kludge > up a metric, like scan rate that spiked when we run out of memory.
That's not what I said. There isn't a deterministic way to ascertain how much of the memory that ZFS is consuming may be free'd when we perform an ARC reap operation. Therefore, trying to create a beancounter that represents free memory in use by ZFS isn't productive since there's no way to guarantee that the number that's reported free could actually be free'd. If you don't want ZFS to consume all of your memory, and want to keep memory out of the ARC so it can always be allocated, there are a well documented set of procedures for configuring limits on the size of the ARC. -j _______________________________________________ perf-discuss mailing list perf-discuss@opensolaris.org
