src/storage/storage_backend_rbd.c | 7 +------ 1 file changed, 1 insertion(+), 6 deletions(-)
When cloning an RBD volume we try to find a snapshot which is
not different to the image we're trying to clone. This boils down
to calling rbd_diff_iterate() or rbd_diff_iterate2() on systems
with newer ceph. These two are passed a callback -
virStorageBackendRBDIterateCb() which mistakenly states that if
a negative value is returned from the callback, the iteration is
stopped but no error is reported. Well, that is not the case.
I've looked into librbd sources and found that a negative value
does mean stop of iteration but it also means an error. That's
why any attempt to clone an RBD image results in the following
error:
virsh # vol-clone --pool rbd_image_root coreos_2023 coreos00.disk
error: Failed to clone vol from coreos_2023
error: failed to iterate RBD snapshot coreos_2023@base: Operation not
permitted
Reported on libvirt-users list:
https://www.redhat.com/archives/libvirt-users/2019-April/msg00060.html
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
---
src/storage/storage_backend_rbd.c | 7 +------
1 file changed, 1 insertion(+), 6 deletions(-)
diff --git a/src/storage/storage_backend_rbd.c b/src/storage/storage_backend_rbd.c
index f8c968e682..3fe188551c 100644
--- a/src/storage/storage_backend_rbd.c
+++ b/src/storage/storage_backend_rbd.c
@@ -1031,14 +1031,9 @@ virStorageBackendRBDIterateCb(uint64_t offset ATTRIBUTE_UNUSED,
{
/*
* Just set that there is a diff for this snapshot, we do not care where
- *
- * When it returns a negative number the rbd_diff_iterate() function will stop
- *
- * That's why we return -1, meaning that there is a difference and we can stop
- * searching any further.
*/
*(int*) arg = 1;
- return -1;
+ return 0;
}
static int
--
2.21.0
--
libvir-list mailing list
libvir-list@redhat.com
https://www.redhat.com/mailman/listinfo/libvir-list
On Mon, Apr 29, 2019 at 8:57 AM Michal Privoznik <mprivozn@redhat.com> wrote: > > When cloning an RBD volume we try to find a snapshot which is > not different to the image we're trying to clone. This boils down > to calling rbd_diff_iterate() or rbd_diff_iterate2() on systems > with newer ceph. These two are passed a callback - > virStorageBackendRBDIterateCb() which mistakenly states that if > a negative value is returned from the callback, the iteration is > stopped but no error is reported. Well, that is not the case. > I've looked into librbd sources and found that a negative value > does mean stop of iteration but it also means an error. That's > why any attempt to clone an RBD image results in the following > error: > > virsh # vol-clone --pool rbd_image_root coreos_2023 coreos00.disk > error: Failed to clone vol from coreos_2023 > error: failed to iterate RBD snapshot coreos_2023@base: Operation not > permitted > > Reported on libvirt-users list: > > https://www.redhat.com/archives/libvirt-users/2019-April/msg00060.html > > Signed-off-by: Michal Privoznik <mprivozn@redhat.com> > --- > src/storage/storage_backend_rbd.c | 7 +------ > 1 file changed, 1 insertion(+), 6 deletions(-) > > diff --git a/src/storage/storage_backend_rbd.c b/src/storage/storage_backend_rbd.c > index f8c968e682..3fe188551c 100644 > --- a/src/storage/storage_backend_rbd.c > +++ b/src/storage/storage_backend_rbd.c > @@ -1031,14 +1031,9 @@ virStorageBackendRBDIterateCb(uint64_t offset ATTRIBUTE_UNUSED, > { > /* > * Just set that there is a diff for this snapshot, we do not care where > - * > - * When it returns a negative number the rbd_diff_iterate() function will stop > - * > - * That's why we return -1, meaning that there is a difference and we can stop > - * searching any further. > */ > *(int*) arg = 1; > - return -1; > + return 0; > } You could also tweak "virStorageBackendRBDSnapshotFindNoDiff" to ignore the error if "diff > 0". That would still allow it to fail-fast once a delta is found. > static int > -- > 2.21.0 > > -- > libvir-list mailing list > libvir-list@redhat.com > https://www.redhat.com/mailman/listinfo/libvir-list -- Jason -- libvir-list mailing list libvir-list@redhat.com https://www.redhat.com/mailman/listinfo/libvir-list
On 4/29/19 3:13 PM, Jason Dillaman wrote: > On Mon, Apr 29, 2019 at 8:57 AM Michal Privoznik <mprivozn@redhat.com> wrote: >> >> When cloning an RBD volume we try to find a snapshot which is >> not different to the image we're trying to clone. This boils down >> to calling rbd_diff_iterate() or rbd_diff_iterate2() on systems >> with newer ceph. These two are passed a callback - >> virStorageBackendRBDIterateCb() which mistakenly states that if >> a negative value is returned from the callback, the iteration is >> stopped but no error is reported. Well, that is not the case. >> I've looked into librbd sources and found that a negative value >> does mean stop of iteration but it also means an error. That's >> why any attempt to clone an RBD image results in the following >> error: >> >> virsh # vol-clone --pool rbd_image_root coreos_2023 coreos00.disk >> error: Failed to clone vol from coreos_2023 >> error: failed to iterate RBD snapshot coreos_2023@base: Operation not >> permitted >> >> Reported on libvirt-users list: >> >> https://www.redhat.com/archives/libvirt-users/2019-April/msg00060.html >> >> Signed-off-by: Michal Privoznik <mprivozn@redhat.com> >> --- >> src/storage/storage_backend_rbd.c | 7 +------ >> 1 file changed, 1 insertion(+), 6 deletions(-) >> >> diff --git a/src/storage/storage_backend_rbd.c b/src/storage/storage_backend_rbd.c >> index f8c968e682..3fe188551c 100644 >> --- a/src/storage/storage_backend_rbd.c >> +++ b/src/storage/storage_backend_rbd.c >> @@ -1031,14 +1031,9 @@ virStorageBackendRBDIterateCb(uint64_t offset ATTRIBUTE_UNUSED, >> { >> /* >> * Just set that there is a diff for this snapshot, we do not care where >> - * >> - * When it returns a negative number the rbd_diff_iterate() function will stop >> - * >> - * That's why we return -1, meaning that there is a difference and we can stop >> - * searching any further. >> */ >> *(int*) arg = 1; >> - return -1; >> + return 0; >> } > > You could also tweak "virStorageBackendRBDSnapshotFindNoDiff" to > ignore the error if "diff > 0". That would still allow it to fail-fast > once a delta is found. Good catch. Let me post a v2. Michal -- libvir-list mailing list libvir-list@redhat.com https://www.redhat.com/mailman/listinfo/libvir-list
© 2016 - 2024 Red Hat, Inc.