Display benchmark results for the default erasure code plugins, in a tab
separated CSV file. The first two column contain the amount of KB
that were coded or decoded, for a given combination of parameters
displayed in the following fields.
seconds KB plugin k m work. iter. size eras.
1.2 10 example 2 1 encode 10 1024 0
0.5 10 example 2 1 decode 10 1024 1
It can be used as input for a human readable report. It is also intented
to be used to show if a given version of an erasure code plugin performs
better than another.
The last column ( not shown above for brievety ) is the exact command
that was run to produce the result so it can be copy / pasted to
reproduce them or to profile.
Only the jerasure techniques mentionned in
https://www.usenix.org/legacy/events/fast09/tech/full_papers/plank/plank_html/
are benchmarked, the others are assumed to be less interesting.
Signed-off-by: Loic Dachary <loic@dachary.org>
Implement the ceph_erasure_code_benchmark utility to:
* load an erasure code plugin
* loop over the encode/decode function using the parameters from the
command line
* print the number of bytes encoded/decoded and the time to process
When decoding, random chunks ( as set with --erasures ) are lost on each
run.
For instance:
$ ceph_erasure_code_benchmark \
--plugin jerasure \
--parameter erasure-code-directory=.libs \
--parameter erasure-code-technique=reed_sol_van \
--parameter erasure-code-k=2 \
--parameter erasure-code-m=2 \
--workload decode \
--erasures 2 \
--iterations 1000
0.964759 1048576
shows 1GB is decoded in 1second.
It is intended to be used by other scripts to present a human readable
output or detect performance regressions.
Signed-off-by: Loic Dachary <loic@dachary.org>
The XOR based example is ten times slower than it could because it uses
the buffer::ptr[] operator. Use a temporary char * instead. It performs
as well as jerasure Reed Solomon when decoding with a single erasure:
$ ceph_erasure_code_benchmark \
--plugin example --parameter erasure-code-directory=.libs \
--parameter erasure-code-technique=example \
--parameter erasure-code-k=2 --parameter erasure-code-m=1 \
--erasure 1 --workload decode --iterations 5000
8.095007 5GB
$ ceph_erasure_code_benchmark \
--plugin jerasure --parameter erasure-code-directory=.libs \
--parameter erasure-code-technique=reed_sol_van \
--parameter erasure-code-k=10 --parameter erasure-code-m=6 \
--erasure 1 --workload decode --iterations 5000
7.870990 5GB
Signed-off-by: Loic Dachary <loic@dachary.org>
When profiling, tools such as valgrind --tool=callgrind require that the
dynamically loaded libraries are not dlclosed so they can collect usage
information.
The public ErasureCodePluginRegistry::disable_dlclose boolean is introduced
for this purpose.
Signed-off-by: Loic Dachary <loic@dachary.org>
ceph-mds --reset-journal didn't work; it would deadlock waiting for
the osdmap. Comparing the init code in the Dumper (that worked) with
that in the Resetter (that didn't), I noticed the lock had to be
released before waiting for the osdmap.
Now the resetter works. However, both the resetter and the dumper
fail an assertion after they've performed their task; I didn't look
into it:
../../src/msg/SimpleMessenger.cc: In function 'void SimpleMessenger::reaper()' t
hread 7fdc188d27c0 time 2013-12-19 04:48:16.930895
../../src/msg/SimpleMessenger.cc: 230: FAILED assert(!cleared)
ceph version 0.72.1-6-g6bca44e (6bca44ec129d11f1c4f38357db8ae435616f2c7c)
1: (SimpleMessenger::reaper()+0x706) [0x880da6]
2: (SimpleMessenger::wait()+0x36f) [0x88180f]
3: (Resetter::reset()+0x714) [0x56e664]
4: (main()+0x1359) [0x562769]
5: (__libc_start_main()+0xf5) [0x3632e21b45]
6: /l/tmp/build/ceph/build/src/ceph-mds() [0x564e49]
NOTE: a copy of the executable, or `objdump -rdS <executable>` is needed to int
erpret this.
2013-12-19 04:48:16.934093 7fdc188d27c0 -1 ../../src/msg/SimpleMessenger.cc: In
function 'void SimpleMessenger::reaper()' thread 7fdc188d27c0 time 2013-12-19 04
:48:16.930895
../../src/msg/SimpleMessenger.cc: 230: FAILED assert(!cleared)
Signed-off-by: Alexandre Oliva <oliva@gnu.org>
Because the gitbuilder build environment does not read the build depends
from the packages and needs to be updated independantly.
Signed-off-by: Loic Dachary <loic@dachary.org>
make check runs vstart.sh to setup a cluster from source and it misses
the python-argparse library as well as uuidgen otherwise it will fail
with:
./vstart.sh: 460: uuidgen: not found
...
Error: Import No module named argparse
Signed-off-by: Loic Dachary <loic@dachary.org>
Allow specifying 'rep', 'raid4' and 'erasure'.
Only allow setting type 'erasure' if all up osds support erasure codes.
Signed-off-by: Joao Eduardo Luis <joao.luis@inktank.com>
If OSDMap has the EC feature set, then update our superblock to
reflect as such, making our on-disk format incompatible with previous
OSDs without EC support.
Fixes: 6028
Signed-off-by: Joao Eduardo Luis <joao.luis@inktank.com>
Add the osd's features to the osd's extra info field in the OSDMap
so we can track which OSDs are able to deal with Erasure Codes.
This will allow us to decide whether or not we are ready to set EC
whenever the user asks us to set EC on a pool -- which shall be
handled by a subsequent commit.
Fixes: 6028
Signed-off-by: Joao Eduardo Luis <joao.luis@inktank.com>
If we have a pool with type Erasure, then we consider we require
CEPH_FEATURE_OSD_ERASURE_CODES.
Signed-off-by: Joao Eduardo Luis <joao.luis@inktank.com>
crush: make set_chooseleaf_tries work with firstn chooseleaf, too
Reviewed-by: Samuel Just <sam.just@inktank.com>
Reviewed-by: Loic Dachary <loic@dachary.com>
The error messages are intentional when error conditions are
created. They will create false positive in the gitbuilder parser when
the string error is found.
The --debug-crush flag is detected to allow the caller to reset the
verbosity level.
Signed-off-by: Loic Dachary <loic@dachary.org>
Include the tunables profile, and flags indicating whether it is optimal,
legacy, or requires certain features.
Signed-off-by: Sage Weil <sage@inktank.com>
