run-tox-ceph-disk and run-tox-ceph-detect-init are already added as
test, so "ctest" and "make {test,check}" will run them without extra
settings.
Signed-off-by: Kefu Chai <kchai@redhat.com>
please note "make test" is used by cmake to run tests, so we cannot just
repurpose it to *build* them.
* AddCephTest.cmake: depends on "tests"
* CMakeLists.txt: let "check" depend on "tests"
* src/CMakeLists.txt: update the run-tox tests
* run-make-check.sh: use "make tests" and "ctest" instead of "make check"
* ceph-detect-init/CMakeLists.txt: let "tests" depend on
"ceph-detect-init"
* ceph-disk/CMakeLists.txt: let "tests" depend on "ceph-disk"
Signed-off-by: Kefu Chai <kchai@redhat.com>
because RGWCoroutine::wakeup() calls RGWCoroutinesStack::wakeup(), the
stack must also stay alive
Fixes: http://tracker.ceph.com/issues/16666
Signed-off-by: Casey Bodley <cbodley@redhat.com>
sadly, sh evalutes `[ -d ]` to true. as it takes "-d" as a non-empty
string as true.
this fixes following failure
```
2016-07-12T23:22:02.839 INFO:teuthology.orchestra.run.mira084.stderr:cp:
missing destination file operand after ‘.’
2016-07-12T23:22:02.839 INFO:teuthology.orchestra.run.mira084.stderr:Try
'cp --help' for more information.
```
see
http://pulpito.ceph.com/kchai-2016-07-12_23:09:35-rados-wip-kefu-testing2---basic-mira/311334/
Signed-off-by: Kefu Chai <kchai@redhat.com>
Make the last_write check in bh_write_adjacencies() match corresponding
check in flush().
Fixes: http://tracker.ceph.com/issues/16610
Signed-off-by: Yan, Zheng <zyan@redhat.com>
this follows the pattern in ceph-disk. this enables us to run
ceph-detect-init/run-tox.sh from the ${CMAKE_BINARY_DIRECTORY}
as well.
Signed-off-by: Kefu Chai <kchai@redhat.com>
If a snaprealm has no child/parent snaprelam, and the snaprealm inode
is not in the cache while client reconnects. The snaprealm does not
get properly removed from MDCache::reconnected_snaplrealm. This causes
incorrect "unconnected snaprealm xxx" warning
Signed-off-by: Yan, Zheng <zyan@redhat.com>
This patch includes following changes
- assign flush tid to snapcap flush
- remove session's flushing_capsnaps list. add inode with snapcap
flushes to session's flushing_caps list instead.
- when reconnecting to MDS, re-send one inode's snapcap flushes and
cap flushes at the same time.
Signed-off-by: Yan, Zheng <zyan@redhat.com>
For a snapshoted direcotry whose snaprealm parents are being opened,
MDS does not know if the directory is purgeable. So MDS can't skip
committing dirfrags of the directory. But if the direcotry is purgeale,
some dirfrags could have already been deleted during MDS failover.
Committing them could return -ENOENT.
Signed-off-by: Yan, Zheng <zyan@redhat.com>
File recovery may update inode and trigger inode COW. MDS relies on
client caps to setup CInode::client_need_snapflush. But for a given
client, the reconnected caps may not include the flushing caps.
(Before MDS failover, client released and flushed some caps at the
same time. When MDS recovers, client re-send the cap flush and send
cap reconnect to the MDS.) This may cause later snapflush to get
dropped.
Signed-off-by: Yan, Zheng <zyan@redhat.com>
Previous commit extended cap_reconnect_t to include snapflush
related information. This information are needed in various
places
Signed-off-by: Yan, Zheng <zyan@redhat.com>
Treat snap inodes that need flush in the same way as open files.
When MDS recovers, this make sure that journal replay bring snap
inodes that need flush into the cache
Signed-off-by: Yan, Zheng <zyan@redhat.com>
As a side effect, the divergent_priors set is no longer
necessary, remove it. Also, ceph_objectstore_tool needed
to be updated to handle the new format.
Signed-off-by: Samuel Just <sjust@redhat.com>
