dbae92253f changed the encoding
significantly, but did not increase the transaction encoding version.
To fix this:
1) Increment the version number to 9.
2) Due to this bug, there are versions which encode the version 9
encoding incorrectly with version 8. We will detect those by trying to
first decode any version 8 encoding with the old decoding function, and
then trying with the new one if there is an exception.
Fixes: 10734
Signed-off-by: Samuel Just <sjust@redhat.com>
There was a typo within the RBD Image get_flags method and
several runtime errors relating to the new fadvise flags.
Fixes: #10782
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
Instead of testing directly using CrushTester::test(), which may break
the monitor if a malformed map is passed and we don't manage to catch
it, fork and run crushtool instead. This way, if crushtool segfaults,
the monitor (or whomever the call might be) will be protected from a
catastrophic failure.
Signed-off-by: Joao Eduardo Luis <joao@redhat.com>
This was available as a legacy mds tell command
previously: make it availably via the cleaner
asok interface. asok rather than new `tell` because
it's a low level/debug thing.
Signed-off-by: John Spray <john.spray@redhat.com>
add control files & config files
rbd-recover-tool is an offline tool for dead ceph cluster
to recever rbd image in replicated pool from all osds.
it is a simple but usefull policy to keep data secruity on ceph
Sigbed-off-by: Min Chen <minchen@ubuntukylin.com>
When kernel client receives traceless reply for request that created
new inode, it re-send a lookup request to MDS get information of the
newly created inode. (VFS expects FS' callback return an inode in this
case). This breaks one request into two requests. Other client may
modify or move to the new inode in the middle.
The fix is avoid sending traceless reply for request that created new
inode. Instead, we convert the origin request into 'lookup' request.
Signed-off-by: Yan, Zheng <zyan@redhat.com>
Completed requests may have created new file/directorie. This guarantees
completed requests are processed before any other client gets change to
modify the new files/directorie.
Signed-off-by: Yan, Zheng <zyan@redhat.com>
No need to use an extra variable to check the result of
a function returning bool if the result isn't used again.
Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>
Move _addr to upper scope to prevent dead pointer 'a' after
assignment of _addr in if-clause.
Fix for:
[src/msg/xio/XioMessenger.cc:669]: (error) Dead pointer usage.
Pointer 'a' is dead if it has been assigned '&_addr' at line 644
Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>
Seems that it doesn't work to add the cppcheck-suppress to
the define. Needs to go to the memcpy call using the define.
Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>
Fix for:
[src/test/erasure-code/TestErasureCodeIsa.cc:474] ->
[src/test/erasure-code/TestErasureCodeIsa.cc:477]:
[src/test/erasure-code/TestErasureCodeIsa.cc:601] ->
[src/test/erasure-code/TestErasureCodeIsa.cc:604]
[src/test/erasure-code/TestErasureCodeIsa.cc:728] ->
[src/test/erasure-code/TestErasureCodeIsa.cc:731]
[src/test/erasure-code/TestErasureCodeIsa.cc:854] ->
[src/test/erasure-code/TestErasureCodeIsa.cc:857]
(performance) Variable 'err' is reassigned a value
before the old one has been used.
Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>
Remove pointer deref before NULL check. These lines are already
after NULL check.
Fix for:
[src/msg/async/EventKqueue.cc:96] -> [src/msg/async/EventKqueue.cc:99]:
(warning) Possible null pointer dereference: tvp - otherwise it is
redundant to check it against null.
[src/msg/async/EventKqueue.cc:97] -> [src/msg/async/EventKqueue.cc:99]:
(warning) Possible null pointer dereference: tvp - otherwise it is
redundant to check it against null.
Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>
