On Fri, Mar 16, 2018 at 11:38 AM, Matthew Ahrens <[email protected]> wrote:
> On Thu, Mar 15, 2018 at 9:48 PM, Mike Gerdts <[email protected]> > wrote: > >> Frankly, I have some concerns about these numbers from before the >> snapshot (same output as first set above, just trimmed) >> >> # zfs get -p space,refreservation,volsize zones/t/100m >> NAME PROPERTY VALUE SOURCE >> zones/t/100m used 110362624 - >> zones/t/100m usedbydataset 105127936 - >> zones/t/100m usedbyrefreservation 5234688 - >> zones/t/100m refreservation 110362624 local >> zones/t/100m volsize 104857600 local >> >> In that output, used and refreservation are equal, but >> usebyreferservation is 5112 KiB. I would have expected that all of the >> 5376 KiB of metadata (used - volsize) would have been counted against the >> refreservation, bringing usedbyrefreservation to 0. >> >> > That output looks right to me. The "used - volsize" is not all metadata, > it also includes the unconsumed refreservation (which is the > usedbyrefreservation property). The "metadata overhead" calculation used > to set the refreservation is a conservative estimate, so normally it you > wouldn't go over the refreservation. The metadata is usedbydataset - > volsize == 270,336 bytes. > > usedbydataset is less than refreservation, so used == refreservation. > > usedbyrefreservation == refreservation - usedbydataset. > > used == usedbydataset + usedbyrefreservation > Thanks for clarifying. Mike ------------------------------------------ openzfs: openzfs-developer Permalink: https://openzfs.topicbox.com/groups/developer/discussions/Te3d593ba00521b6d-Ma80425738c196e6f69c9a358 Delivery options: https://openzfs.topicbox.com/groups
