On Wed, Jan 18, 2012 at 05:13, Jens Rehpöhler <[email protected]> wrote: > Hi all, > > todo i've done few performance tests with my two clusters. > > I've used : rados -p data load-gen --run-length 10 --target-throughput 50 > > to generate some load. Now i have a question regarding "ceph -w" output. > > The output before the test was: > > 2012-01-18 11:55:28.621107 pg v113011: 594 pgs: 594 peering; 61022 MB > data, 120 GB used, 244 GB / 372 GB avail > > here is the output after the test: > > 2012-01-18 13:53:30.190602 pg v113119: 594 pgs: 594 active+clean; > 1058 GB data, 120 GB used, 244 GB / 372 GB avail > > How can i have more Data then the available discspace ? Is there any > possiblity to identify the rados objects of this test and delete them > manualy ?
A quick read of the source seems to say "rados load-gen" writes at random offsets.. size_t off = get_random(0, info.len); which means you have a lot of sparse objects. And that makes the "1058 GB data, 120 GB used" make sense. -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html
