The commit 289b7f4bcdd changed a behavior of SD_OP_CLUSTER_INFO. The
changed op returns SD_RES_WAIT_FOR_FORMAT if the cluster is not
formatted yet. But this behavior conflicts with ongoing fixed vnode
feature. This patch resolves the conflict by changing a way of
detecting cluster format in dog
At Fri, 9 Jan 2015 16:37:13 +0900,
Hitoshi Mitake wrote:
LTTng is the de facto standard of tracing framework. This patchset
employes it for performance analysis and debugging of sheepdog. It
would be useful for analyzing bottlenecks.
Hitoshi Mitake (6):
configure: enable LTTng with
At Fri, 09 Jan 2015 18:00:36 +0900,
Saeki Masaki wrote:
On 2015/01/09 17:18, Hitoshi Mitake wrote:
The commit 289b7f4bcdd changed a behavior of SD_OP_CLUSTER_INFO. The
changed op returns SD_RES_WAIT_FOR_FORMAT if the cluster is not
formatted yet. But this behavior conflicts with ongoing
See http://jenkins.sheepdog-project.org:8080/job/sheepdog-build/598/changes
Changes:
[mitake.hitoshi] sheep, dog: detect a cluster is formatted or not with
cinfo.ctime
[mitake.hitoshi] configure: enable LTTng with --enable-lttng-ust option
[mitake.hitoshi] sheep: use _exit(2) instead of
On 2015/01/09 17:18, Hitoshi Mitake wrote:
The commit 289b7f4bcdd changed a behavior of SD_OP_CLUSTER_INFO. The
changed op returns SD_RES_WAIT_FOR_FORMAT if the cluster is not
formatted yet. But this behavior conflicts with ongoing fixed vnode
feature. This patch resolves the conflict by
Qemu backend driver for sheepdog used a hard-coded value
'SD_DATA_OBJ_SIZE' when you created VDI with qemu-img
command.
Then, this patch add operation SD_OP_GET_CLUSTER_DEFAULT
to get a cluster default value of block_size_shift
for creating a VDI.
Signed-off-by: Teruaki Ishizaki