I do not think that 4 is meaninful. If a value needs to be determined,
it should be configured in the configuration file.
Signed-off-by: optimistyzy <optimistyzy@gmail.com>
By default RocksDB extensively employs CRC32. It has two paths
for the checksum calculation:
* rocksdb::crc32c::Slow_CRC32,
* rocksdb::crc32c::Fast_CRC32.
The fast path depends on a run-time discovery of CPU capabilities
AND a compile-time define __SSE4_2__. Although my systems really
offer SSE4.2 support, the macro was undefined resulting in poor
RocksDB performance visible especially during WAL transactions.
The patch (awkwardly) adds the -msse4.2 to CXXFLAGS for RocksDB.
Signed-off-by: Radoslaw Zarzynski <rzarzynski@mirantis.com>
Run all of our cases with no thrashing. This would have uncovered
http://tracker.ceph.com/issues/19131 much sooner!
Signed-off-by: Sage Weil <sage@redhat.com>
Formerly masked in part by the reclaim() action, direct-delete now
substitutes for reclaim() iff its LRU lane is over its high-water
mark, and in particular, like reclaim() the destructor is certain
to see handles still interned on the FHcache when nfs-ganesha is
recycling objects from its own LRU.
Fixes: http://tracker.ceph.com/issues/19112
Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
As an ganesha FSAL backend, rgw_file should properly maintain
the atime,ctime,mtime properly against operations such as:
(read,write) for file
(create,unlink,mkdir,rmdir,rename) for dir
(setattr) for file and dir
Signed-off-by: Gui Hecheng <guihecheng@cmss.chinamobile.com>
rbd: When Ceph cluster becomes full, should allow user to remove rbd …
Reviewed-by: Sage Weil <sage@redhat.com>
Reviewed-by: Jason Dillaman <dillaman@redhat.com>
rados: we need to get the latest osdmap when pool does not exists
Reviewed-by: Sage Weil <sage@redhat.com>
Reviewed-by: Josh Durgin <jdurgin@redhat.com>
BitMapZone aggregates BmapEntries through a pointer to independently-
allocated std::vector which in turn has its own pointer to the real
data. This indirection is unnecessary as std::vector is pretty cheap
in the terms of memory overhead. For instance, on x86-64 & GCC 5.4.0
sizeof(std::vector) == 24, which translates to 3 raw pointers.
Stripping the indirection could have positive impact on processor's
caches and allow to minimize the number of memory accesses.
Signed-off-by: Radoslaw Zarzynski <rzarzynski@mirantis.com>
The snaptrimmer can block taking a write lock, which might happen due to
a conficting EC read. When the EC read completes, we need to wake up the
snaptrimmer.
Fixes: http://tracker.ceph.com/issues/19131
Signed-off-by: Sage Weil <sage@redhat.com>
I have fixed the issue cleanup #19090: Wrong hard-coded URLs
I have replaced the wrong URLs with correct URLs redirecting to the Tunables headline under Crush Map documentation.
Please add it to the documentation.
We will now only enable persistent features automatically
when ALL the monitors in the monmap are in the quorum.
#noMonitorLeftBehind
Signed-off-by: Joao Eduardo Luis <joao@suse.de>
