Definitely broken -- I've opened a new tracker ticket to get it fixed [1]. [1] http://tracker.ceph.com/issues/16386 On Sat, Jun 18, 2016 at 5:29 AM, Alexandre DERUMIER <aderumier@xxxxxxxxx> wrote: >>>Thanks Jason . I had tried it some week ago, and my fstrim was hanging. >>>It'll do more test next week. >>> >>>It must be defined on client right ? >>> >>>[client] >>>rbd_skip_partial_discard = true > > > Just tested again, fstrim is totally hanging > > I have attached the client.client.log in debug verbose, the log start when fstrim start > > ----- Mail original ----- > De: "aderumier" <aderumier@xxxxxxxxx> > À: dillaman@xxxxxxxxxx > Cc: "ceph-devel" <ceph-devel@xxxxxxxxxxxxxxx> > Envoyé: Samedi 18 Juin 2016 09:40:40 > Objet: Re: using discard from qemu guest when snapshot exist, fill the rbd layer with zeroes and use more space > >>>There is a "rbd_skip_partial_discard" configuration option which can >>>be set to true. When enabled, discard will avoid zeroing holes if the >>>discard extent is in the middle of an object. > > Thanks Jason . I had tried it some week ago, and my fstrim was hanging. > It'll do more test next week. > > It must be defined on client right ? > > [client] > rbd_skip_partial_discard = true > > > > ----- Mail original ----- > De: "Jason Dillaman" <jdillama@xxxxxxxxxx> > À: "aderumier" <aderumier@xxxxxxxxx> > Cc: "ceph-devel" <ceph-devel@xxxxxxxxxxxxxxx> > Envoyé: Vendredi 17 Juin 2016 15:21:53 > Objet: Re: using discard from qemu guest when snapshot exist, fill the rbd layer with zeroes and use more space > > There is a "rbd_skip_partial_discard" configuration option which can > be set to true. When enabled, discard will avoid zeroing holes if the > discard extent is in the middle of an object. > > On Fri, Jun 17, 2016 at 5:10 AM, Alexandre DERUMIER <aderumier@xxxxxxxxx> wrote: >> Hi, >> >> I'm using fstrim inside my qemu guests to remove unused space from the rbd storage. >> >> This is working fine, but I just notice that if a snapshot exist, >> >> It just do the reverse, it's filling the snasphot with zeroes, and the total cluster space is growing. >> >> It's is possible to avoid that ? >> >> >> Regards, >> >> Alexandre >> -- >> To unsubscribe from this list: send the line "unsubscribe ceph-devel" in >> the body of a message to majordomo@xxxxxxxxxxxxxxx >> More majordomo info at http://vger.kernel.org/majordomo-info.html > > > > -- > Jason > -- > To unsubscribe from this list: send the line "unsubscribe ceph-devel" in > the body of a message to majordomo@xxxxxxxxxxxxxxx > More majordomo info at http://vger.kernel.org/majordomo-info.html -- Jason -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
