Catch decode errors so osd doesn't crash on corrupt OI_ATTR or SS_ATTR
Use boost::optional<> to make current state clearer
Create next_clone as needed using head/curclone
Add equivalent logic after getting to end of scrubmap.objects
Fixes: #12738
Signed-off-by: David Zafman <dzafman@redhat.com>
in addition to testing the references to unknown name/type, we should
perform necessary test the crushmap when looking for a good crushmap
Signed-off-by: Kefu Chai <kchai@redhat.com>
_fdump wrongly returns if journal is currently unreadable, and fix some other errors also.
Fixes: #13626
Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
Add a default quota config that is applied when no specific quota is
enabled. There are both per-bucket and per-user defaults. Defaults are
enabled if either max count or max size is enabled.
This fixes issues 12912 and 12997.
Signed-off-by: Daniel Gryniewicz <dang@redhat.com>
with dmcrypt device, ceph-disk list will get Nonetype with the real device
i.e. /dev/sdb1 for /dev/dm-1, `ceph-disk --list /dev/sdb1` will get Nonetype
that because the holder we query is `dm-x`, the devices includ the path
is `sdX`, it cannot match.
if we list format plain and meet dmcrypt device, query again with the
holder (dm-x).
Also change the list_devices() parameter to let other function to use it
easily.
Signed-off-by: Vicente Cheng <freeze.bilsted@gmail.com>
We were indexing the buckets array without verifying the index was within
the [0,max_buckets) range. This could happen because a multistep rule
does not have enough buckets and has CRUSH_ITEM_NONE
for an intermediate result, which would feed in CRUSH_ITEM_NONE and
make us crash.
Fixes: #13477
Signed-off-by: Sage Weil <sage@redhat.com>
This parameter has been removed since systemd 213, so this
effects Fedora 21+, Debian Jessie, and potentially future
releases of RHEL 7.
Fixes: #13560
Backport: hammer, infernalis
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
If the request in one Requester Pays bucket succeeds the response has to
include the header 'x-amz-request-charged: requester'.
Fixes: #13427
Signed-off-by: Javier M. Mellid <jmunhoz@igalia.com>
When we remove the ready flag from osd directory, it will execute mkfs again.
It will read superblock but can not check osd id and cluster fsid correctly.
It need to decode the superblock that we read from osd.
It will always failure now, so I modify it for read osd id and fsid correctly.
http://tracker.ceph.com/issues/13586Fixes: #13586
Signed-off-by: Vicente Cheng <freeze.bilsted@gmail.com>
When copy/pasting a tests, it is easy to forget (or not know) that the
port used must be unique to allow for multiple tests to run in
parallel (make -j8). Add a reminder next to each port.
Signed-off-by: Loic Dachary <ldachary@redhat.com>
we could not reset max to 0 in Throttle using perf reset command,
such as throttle-osd_client_messages, throttle-filestore_bytes and so on.
If we call PerfCountersBuilder::add_u64, the data.type would be PERFCOUNTER_U64.
the reset function would skip it if data.type is PERFCOUNTER_U64.
Fixes: #13517
Signed-off-by: Xinze Chi <xinze@xsky.com>
