RepoMirrors/ceph
1
0
Fork 0
mirror of https://github.com/ceph/ceph synced 2025-03-11 02:39:05 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge branch 'wip-3610' into next

Browse Source
This commit is contained in:
Sam Lang 2012-12-14 09:00:24 -10:00
commit b63940caa1
19 changed files with 1691 additions and 202 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

185
doc/architecture.rst
View File

@ -146,15 +146,30 @@ appropriate placement groups in the secondary and tertiary OSDs (as many OSDs as
additional replicas), and responds to the client once it has confirmed the
object was stored successfully.
.. ditaa:: +--------+ Write +--------------+ Replica 1 +----------------+
| Client |*-------------->| Primary OSD |*---------------->| Secondary OSD |
| |<--------------*| |<----------------*| |
+--------+ Write Ack +--------------+ Replica 1 Ack +----------------+
^ *
| | Replica 2 +----------------+
| +----------------------->| Tertiary OSD |
+--------------------------*| |
Replica 2 Ack +----------------+
.. ditaa::
+----------+
| Client |
| |
+----------+
* ^
Write (1) | | Ack (6)
| |
v *
+-------------+
| Primary OSD |
| |
+-------------+
* ^ ^ *
Write (2) | | | | Write (3)
+------+ | | +------+
| +------+ +------+ |
| | Ack (4) Ack (5)| |
v * * v
+---------------+ +---------------+
| Secondary OSD | | Tertiary OSD |
| | | |
+---------------+ +---------------+
Since any network device has a limit to the number of concurrent connections it
@ -222,82 +237,84 @@ of striping:
If you anticipate large images sizes, large S3 or Swift objects (video), or
large CephFS files, you may see considerable read/write performance improvements
by striping client data over mulitple objects within an object set. Significant
write performance occurs when the client writes the stripe units to their
corresponding objects simultaneously. Since objects get mapped to different
placement groups and further mapped to different OSDs, each write occurs
simultaneously at the maximum write speed. So the stripe count may serve as a
proxy for the multiple of the performance improvement. Read performance is
similarly affected. However, setting up connections between the client and the
OSDs and the network latency also play a role in the overall performance.
large CephFS directories, you may see considerable read/write performance
improvements by striping client data over mulitple objects within an object set.
Significant write performance occurs when the client writes the stripe units to
their corresponding objects in parallel. Since objects get mapped to different
placement groups and further mapped to different OSDs, each write occurs in
parallel at the maximum write speed. A write to a single disk would be limited
by the head movement (e.g. 6ms per seek) and bandwidth of that one device (e.g.
100MB/s). By spreading that write over multiple objects (which map to different
placement groups and OSDs) Ceph can reduce the number of seeks per drive and
combine the throughput of multiple drives to achieve much faster write (or read)
speeds.
In the following diagram, client data gets striped across an object set
(``object set 1`` in the following diagram) consisting of 4 objects, where the
first stripe unit is ``stripe 0`` in ``object 0``, and the fourth stripe unit is
``stripe 3`` in ``object 3``. After writing the fourth stripe, the client
determines if the object set is full. If the object set is not full, the client
begins writing a stripe to the first object again (``object 0`` in the following
diagram). If the object set is full, the client creates a new object set
(``object set 2`` in the following diagram), and begins writing to the first
stripe (``stripe 4``) in the first object in the new object set (``object 4`` in
the diagram below).
first stripe unit is ``stripe unit 0`` in ``object 0``, and the fourth stripe
unit is ``stripe unit 3`` in ``object 3``. After writing the fourth stripe, the
client determines if the object set is full. If the object set is not full, the
client begins writing a stripe to the first object again (``object 0`` in the
following diagram). If the object set is full, the client creates a new object
set (``object set 2`` in the following diagram), and begins writing to the first
stripe (``stripe unit 16``) in the first object in the new object set (``object
4`` in the diagram below).
.. ditaa::
+---------------+
| Client Data |
| Format |
| cCCC |
+---------------+
|
+-----------------+--------+--------+-----------------+
| | | | +--\
v v v v |
/-----------\ /-----------\ /-----------\ /-----------\ |
| Begin cCCC| | Begin cCCC| | Begin cCCC| | Begin cCCC| |
| Object 0 | | Object 1 | | Object 2 | | Object 3 | |
+-----------+ +-----------+ +-----------+ +-----------+ |
| stripe | | stripe | | stripe | | stripe | |
| unit 0 | | unit 1 | | unit 2 | | unit 3 | |
+-----------+ +-----------+ +-----------+ +-----------+ |
| stripe | | stripe | | stripe | | stripe | +-\
| unit 4 | | unit 5 | | unit 6 | | unit 7 | | Object
+-----------+ +-----------+ +-----------+ +-----------+ +- Set
| stripe | | stripe | | stripe | | stripe | | 1
| unit 8 | | unit 9 | | unit 10 | | unit 11 | +-/
+-----------+ +-----------+ +-----------+ +-----------+ |
| stripe | | stripe | | stripe | | stripe | |
| unit 12 | | unit 13 | | unit 14 | | unit 15 | |
+-----------+ +-----------+ +-----------+ +-----------+ |
| End cCCC | | End cCCC | | End cCCC | | End cCCC | |
| Object 0 | | Object 1 | | Object 2 | | Object 3 | |
\-----------/ \-----------/ \-----------/ \-----------/ |
|
+--/
+---------------+
| Client Data |
| Format |
| cCCC |
+---------------+
|
+-----------------+--------+--------+-----------------+
| | | | +--\
v v v v |
/-----------\ /-----------\ /-----------\ /-----------\ |
| Begin cCCC| | Begin cCCC| | Begin cCCC| | Begin cCCC| |
| Object 0 | | Object 1 | | Object 2 | | Object 3 | |
+-----------+ +-----------+ +-----------+ +-----------+ |
| stripe | | stripe | | stripe | | stripe | |
| unit 0 | | unit 1 | | unit 2 | | unit 3 | |
+-----------+ +-----------+ +-----------+ +-----------+ |
| stripe | | stripe | | stripe | | stripe | +-\
| unit 4 | | unit 5 | | unit 6 | | unit 7 | | Object
+-----------+ +-----------+ +-----------+ +-----------+ +- Set
| stripe | | stripe | | stripe | | stripe | | 1
| unit 8 | | unit 9 | | unit 10 | | unit 11 | +-/
+-----------+ +-----------+ +-----------+ +-----------+ |
| stripe | | stripe | | stripe | | stripe | |
| unit 12 | | unit 13 | | unit 14 | | unit 15 | |
+-----------+ +-----------+ +-----------+ +-----------+ |
| End cCCC | | End cCCC | | End cCCC | | End cCCC | |
| Object 0 | | Object 1 | | Object 2 | | Object 3 | |
\-----------/ \-----------/ \-----------/ \-----------/ |
|
+--/
+--\
|
/-----------\ /-----------\ /-----------\ /-----------\ |
| Begin cCCC| | Begin cCCC| | Begin cCCC| | Begin cCCC| |
| Object 4 | | Object 5 | | Object 6 | | Object 7 | |
+-----------+ +-----------+ +-----------+ +-----------+ |
| stripe | | stripe | | stripe | | stripe | |
| unit 15 | | unit 16 | | unit 17 | | unit 18 | |
+-----------+ +-----------+ +-----------+ +-----------+ |
| stripe | | stripe | | stripe | | stripe | +-\
| unit 19 | | unit 20 | | unit 21 | | unit 22 | | Object
+-----------+ +-----------+ +-----------+ +-----------+ +- Set
| stripe | | stripe | | stripe | | stripe | | 2
| unit 23 | | unit 24 | | unit 25 | | unit 26 | +-/
+-----------+ +-----------+ +-----------+ +-----------+ |
| stripe | | stripe | | stripe | | stripe | |
| unit 27 | | unit 28 | | unit 29 | | unit 30 | |
+-----------+ +-----------+ +-----------+ +-----------+ |
| End cCCC | | End cCCC | | End cCCC | | End cCCC | |
| Object 4 | | Object 5 | | Object 6 | | Object 7 | |
\-----------/ \-----------/ \-----------/ \-----------/ |
|
+--/
+--\
|
/-----------\ /-----------\ /-----------\ /-----------\ |
| Begin cCCC| | Begin cCCC| | Begin cCCC| | Begin cCCC| |
| Object 4 | | Object 5 | | Object 6 | | Object 7 | |
+-----------+ +-----------+ +-----------+ +-----------+ |
| stripe | | stripe | | stripe | | stripe | |
| unit 16 | | unit 17 | | unit 18 | | unit 19 | |
+-----------+ +-----------+ +-----------+ +-----------+ |
| stripe | | stripe | | stripe | | stripe | +-\
| unit 20 | | unit 21 | | unit 22 | | unit 23 | | Object
+-----------+ +-----------+ +-----------+ +-----------+ +- Set
| stripe | | stripe | | stripe | | stripe | | 2
| unit 24 | | unit 25 | | unit 26 | | unit 27 | +-/
+-----------+ +-----------+ +-----------+ +-----------+ |
| stripe | | stripe | | stripe | | stripe | |
| unit 28 | | unit 29 | | unit 30 | | unit 31 | |
+-----------+ +-----------+ +-----------+ +-----------+ |
| End cCCC | | End cCCC | | End cCCC | | End cCCC | |
| Object 4 | | Object 5 | | Object 6 | | Object 7 | |
\-----------/ \-----------/ \-----------/ \-----------/ |
|
+--/
Three important variables determine how Ceph stripes data:
@ -306,9 +323,9 @@ Three important variables determine how Ceph stripes data:
enough to accomodate many stripe units, and should be a multiple of
the stripe unit.
- **Stripe Unit:** Stripes have a configurable unit size (e.g., 64kb).
- **Stripe Width:** Stripes have a configurable unit size (e.g., 64kb).
The Ceph client divides the data it will write to objects into equally
sized stripe units, except for the last stripe unit. A stripe unit,
sized stripe units, except for the last stripe unit. A stripe width,
should be a fraction of the Object Size so that an object may contain
many stripe units.
@ -347,7 +364,11 @@ storage disk. See `How Ceph Scales`_ for details.
get mapped to placement groups in the same pool. So they use the same CRUSH
map and the same access controls.
.. tip:: The objects Ceph stores in the Object Store are not striped.
.. tip:: The objects Ceph stores in the Object Store are not striped. RGW, RBD
and CephFS automatically stripe their data over multiple RADOS objects.
Clients that write directly to the Object Store via ``librados`` must
peform the the striping (and parallel I/O) for themselves to obtain these
benefits.
Data Consistency

23
doc/install/os-recommendations.rst
View File

@ -24,7 +24,8 @@ glibc
- **syncfs(2)**: For non-btrfs filesystems such as XFS and ext4 where
more than one ``ceph-osd`` daemon is used on a single server, Ceph
performs signficantly better with the ``syncfs(2)`` system call
(added in kernel 2.6.39 and glibc 2.14).
(added in kernel 2.6.39 and glibc 2.14). New versions of Ceph (v0.55 and
later) do not depend on glibc support.
Platforms
@ -49,12 +50,12 @@ Argonaut (0.48)
+----------+----------+--------------------+--------------+---------+------------+
| Ubuntu | 12.10 | Quantal Quetzal | linux-3.5.4 | 2 | B |
+----------+----------+--------------------+--------------+---------+------------+
| Debian | 6.0 | Squeeze | linux-2.6.32 | 1, 2 | B |
| Debian | 6.0 | Squeeze | linux-2.6.32 | 1, 2, 3 | B |
+----------+----------+--------------------+--------------+---------+------------+
| Debian | 7.0 | Wheezy | linux-3.2.0 | 1, 2 | B |
| Debian | 7.0 | Wheezy | linux-3.2.0 | 1, 2, 3 | B |
+----------+----------+--------------------+--------------+---------+------------+
Bobtail (0.55)
Bobtail (0.56)
--------------
+----------+----------+--------------------+--------------+---------+------------+
@ -62,17 +63,17 @@ Bobtail (0.55)
+==========+==========+====================+==============+=========+============+
| Ubuntu | 11.04 | Natty Narwhal | linux-2.6.38 | 1, 2, 3 | B |
+----------+----------+--------------------+--------------+---------+------------+
| Ubuntu | 11.10 | Oneric Ocelot | linux-3.0.0 | 1, 2, 3 | B |
| Ubuntu | 11.10 | Oneric Ocelot | linux-3.0.0 | 1, 2 | B |
+----------+----------+--------------------+--------------+---------+------------+
| Ubuntu | 12.04 | Precise Pangolin | linux-3.2.0 | 1, 2 | B, I, C |
+----------+----------+--------------------+--------------+---------+------------+
| Ubuntu | 12.10 | Quantal Quetzal | linux-3.5.4 | 2 | B |
+----------+----------+--------------------+--------------+---------+------------+
| Debian | 6.0 | Squeeze | linux-2.6.32 | 1, 2 | B |
| Debian | 6.0 | Squeeze | linux-2.6.32 | 1, 2, 3 | B |
+----------+----------+--------------------+--------------+---------+------------+
| Debian | 7.0 | Wheezy | linux-3.2.0 | 1, 2 | B |
+----------+----------+--------------------+--------------+---------+------------+
| CentOS | 6.3 | N/A | linux-2.6.32 | 1, 2, 3 | B, I |
| CentOS | 6.3 | N/A | linux-2.6.32 | 1, 2 | B, I |
+----------+----------+--------------------+--------------+---------+------------+
| Fedora | 17.0 | Beefy Miracle | linux-3.3.4 | 1, 2 | B |
+----------+----------+--------------------+--------------+---------+------------+
@ -93,10 +94,10 @@ Notes
for kernel client (kernel RBD or the Ceph file system). Upgrade to a
recommended kernel.
- **3**: The installed version of ``glibc`` does not support the
``syncfs(2)`` system call. Putting multiple ``ceph-osd`` daemons
using ``XFS`` or ``ext4`` on the same host will not perform as well as
they could.
- **3**: The default kernel or installed version of ``glibc`` does not
support the ``syncfs(2)`` system call. Putting multiple
``ceph-osd`` daemons using ``XFS`` or ``ext4`` on the same host will
not perform as well as they could.
Testing
-------

12
doc/rados/configuration/filestore-config-ref.rst
View File

@ -205,7 +205,7 @@ Journal
``filestore journal parallel``
:Description:
:Description: Enables parallel journaling, default for btrfs.
:Type: Boolean
:Required: No
:Default: ``false``
@ -213,7 +213,7 @@ Journal
``filestore journal writeahead``
:Description:
:Description: Enables writeahead journaling, default for xfs.
:Type: Boolean
:Required: No
:Default: ``false``
@ -221,7 +221,7 @@ Journal
``filestore journal trailing``
:Description:
:Description: Deprecated, never use.
:Type: Boolean
:Required: No
:Default: ``false``
@ -233,7 +233,7 @@ Misc
``filestore merge threshold``
:Description:
:Description: Min number of files in a subdir before merging into parent
:Type: Integer
:Required: No
:Default: ``10``
@ -241,7 +241,7 @@ Misc
``filestore split multiple``
:Description:
:Description: filestore_split_multiple*filestore_merge_threshold*16 is the max files in a subdir before splitting into child directories.
:Type: Integer
:Required: No
:Default: ``2``
@ -249,7 +249,7 @@ Misc
``filestore update to``
:Description:
:Description: Limits filestore auto upgrade to specified version.
:Type: Integer
:Required: No
:Default: ``1000``

20
doc/rados/configuration/ms-ref.rst
View File

@ -5,7 +5,7 @@
``ms tcp nodelay``
:Description:
:Description: Disables nagle's algorithm on messenger tcp sessions.
:Type: Boolean
:Required: No
:Default: ``true``
@ -13,7 +13,7 @@
``ms initial backoff``
:Description:
:Description: The initial time to wait before reconnecting on a fault.
:Type: Double
:Required: No
:Default: ``.2``
@ -21,7 +21,7 @@
``ms max backoff``
:Description:
:Description: The maximum time to wait before reconnecting on a fault.
:Type: Double
:Required: No
:Default: ``15.0``
@ -29,7 +29,7 @@
``ms nocrc``
:Description:
:Description: Disables crc on network messages. May increase performance if cpu limited.
:Type: Boolean
:Required: No
:Default: ``false``
@ -37,7 +37,7 @@
``ms die on bad msg``
:Description:
:Description: Debug option; do not configure.
:Type: Boolean
:Required: No
:Default: ``false``
@ -45,7 +45,7 @@
``ms dispatch throttle bytes``
:Description:
:Description: Throttles total size of messages waiting to be dispatched.
:Type: 64-bit Unsigned Integer
:Required: No
:Default: ``100 << 20``
@ -53,7 +53,7 @@
``ms bind ipv6``
:Description:
:Description: Enable if you want your daemons to bind to IPv6 address instead of IPv4 ones. (Not required if you specify a daemon or cluster IP.)
:Type: Boolean
:Required: No
:Default: ``false``
@ -61,7 +61,7 @@
``ms rwthread stack bytes``
:Description:
:Description: Debug option for stack size; do not configure.
:Type: 64-bit Unsigned Integer
:Required: No
:Default: ``1024 << 10``
@ -69,7 +69,7 @@
``ms tcp read timeout``
:Description:
:Description: Controls how long (in seconds) the messenger will wait before closing an idle connection.
:Type: 64-bit Unsigned Integer
:Required: No
:Default: ``900``
@ -77,7 +77,7 @@
``ms inject socket failures``
:Description:
:Description: Debug option; do not configure.
:Type: 64-bit Unsigned Integer
:Required: No
:Default: ``0``

23
doc/rados/deployment/mkcephfs.rst
View File

@ -19,7 +19,7 @@ Enter a password for the root user.
On the admin host, generate an ``ssh`` key without specifying a passphrase
and use the default locations. ::
sudo -s
sudo -i
ssh-keygen
Generating public/private key pair.
Enter file in which to save the key (/root/.ssh/id_rsa):
@ -27,14 +27,8 @@ and use the default locations. ::
Enter same passphrase again:
Your identification has been saved in /root/.ssh/id_rsa.
Your public key has been saved in /root/.ssh/id_rsa.pub.
You may use RSA or DSA keys. Once you generate your keys, copy them to each
OSD host. For example::
ssh-copy-id root@myserver01
ssh-copy-id root@myserver02
Modify your ``~/.ssh/config`` file to login as ``root``, as follows::
Modify your ``/root/.ssh/config`` file to login as ``root``, as follows::
Host myserver01
Hostname myserver01.fully-qualified-domain.com
@ -43,6 +37,11 @@ Modify your ``~/.ssh/config`` file to login as ``root``, as follows::
Hostname myserver02.fully-qualified-domain.com
User root
You may use RSA or DSA keys. Once you generate your keys, copy them to each
OSD host. For example::
ssh-copy-id root@myserver01
ssh-copy-id root@myserver02
Copy Configuration File to All Hosts
====================================
@ -55,9 +54,9 @@ if you are using ``mkcephfs`` to deploy Ceph.
::
ssh myserver01 sudo tee /etc/ceph/ceph.conf < /etc/ceph/ceph.conf
ssh myserver02 sudo tee /etc/ceph/ceph.conf < /etc/ceph/ceph.conf
ssh myserver03 sudo tee /etc/ceph/ceph.conf < /etc/ceph/ceph.conf
sudo ssh myserver01 tee /etc/ceph/ceph.conf < /etc/ceph/ceph.conf
sudo ssh myserver02 tee /etc/ceph/ceph.conf < /etc/ceph/ceph.conf
sudo ssh myserver03 tee /etc/ceph/ceph.conf < /etc/ceph/ceph.conf
Create the Default Directories
@ -115,7 +114,7 @@ root password. See `Authentication`_ when running with ``cephx`` enabled.
When you start or stop your cluster, you will not have to use ``sudo`` or
provide passwords. For example::
service ceph -a start
sudo service ceph -a start
See `Operating a Cluster`_ for details.

5
doc/rados/operations/add-or-rm-osds.rst
View File

@ -107,10 +107,9 @@ hard disks than older hosts in the cluster (i.e., they may have greater weight).
ssh {new-osd} sudo tee /etc/ceph/ceph.conf < /etc/ceph/ceph.conf
#. Create the OSD. ::
#. Create the OSD. If no UUID is given, it will be set automatically when the OSD starts up. ::
ceph osd create {osd-num}
ceph osd create 123 #for example
ceph osd create [{uuid}]
#. Initialize the OSD data directory. ::

6
doc/rados/operations/control.rst
View File

@ -151,10 +151,10 @@ Mark an OSD as lost. This may result in permanent data loss. Use with caution. :
ceph osd lost [--yes-i-really-mean-it]
Create a new OSD. If no ID is given, a new ID is automatically selected
if possible. ::
Create a new OSD. If no UUID is given, it will be set automatically when the OSD
starts up. ::
ceph osd create [{id}]
ceph osd create [{uuid}]
Remove the given OSD(s). ::

1507
doc/radosgw/admin/adminops.rst
View File

File diff suppressed because it is too large Load Diff

4
doc/radosgw/config.rst
View File

@ -113,7 +113,7 @@ See `<IfModule> Directive`_ for additional details.
.. _<IfModule> Directive: http://httpd.apache.org/docs/2.2/mod/core.html#ifmodule
Finally, you should configure Apache to allow encoded slashes, provide paths for
log files and to trun off server signatures. ::
log files and to turn off server signatures. ::
<VirtualHost *:80>
...
@ -171,7 +171,7 @@ Add to Ceph Keyring Entries
Once you have created a keyring and key for RADOS GW, add it as an entry in
the Ceph keyring. For example::
ceph -k /etc/ceph/ceph.keyring auth add client.radosgw.gateway -i /etc/ceph/keyring.radosgw.gateway
sudo ceph -k /etc/ceph/ceph.keyring auth add client.radosgw.gateway -i /etc/ceph/keyring.radosgw.gateway
Restart Services and Start the RADOS Gateway

6
doc/radosgw/s3/python.rst
View File

@ -32,9 +32,9 @@ This also prints out the bucket name and creation date of each bucket.
.. code-block:: python
for bucket in conn.get_all_buckets():
print "{name}\t{created}".format(
name = bucket.name,
created = bucket.creation_date,
print "{name}\t{created}".format(
name = bucket.name,
created = bucket.creation_date,
)
The output will look something like this::

2
qa/run_xfstests.sh
View File

@ -49,7 +49,7 @@ XFS_MKFS_OPTIONS="-l su=32k"
# until we can work through getting them all passing reliably.
TESTS="1-9 11-15 17 19-21 26-29 31-34 41 46-48 50-54 56 61 63-67 69-70 74-76"
TESTS="${TESTS} 78 79 84-89 91-92 100 103 105 108 110 116-121 124 126"
TESTS="${TESTS} 129-135 137-141 164-167 179 182-184 186-190 192 194"
TESTS="${TESTS} 129-135 137-141 164-167 182 184 186-190 192 194"
TESTS="${TESTS} 196 199 201 203 214-216 220-227 234 236-238 241 243-249"
TESTS="${TESTS} 253 257-259 261 262 269 273 275 277 278 280 285 286"
# 275 was the highest available test as of 4/10/12.

6
qa/run_xfstests_qemu.sh
View File

@ -3,5 +3,7 @@
mkdir /tmp/cephtest
wget https://raw.github.com/ceph/ceph/master/qa/run_xfstests.sh
chmod +x run_xfstests.sh
# tests excluded require extra packages for advanced acl and quota support
./run_xfstests.sh -c 1 -f xfs -t /dev/vdb -s /dev/vdc 1-26 28-49 51-63 65-83 85-233 235-291
# tests excluded fail in the current testing vm regardless of whether
# rbd is used
./run_xfstests.sh -c 1 -f xfs -t /dev/vdb -s /dev/vdc 1-17 19-26 28-49 51-61 63 66-79 83 85-105 108-110 112-135 137-170 172-204 206-217 220-227 230-231 233 235-241 243-249 251-262 264-278 281-286 288-289

3
src/Makefile.am
View File

@ -228,10 +228,12 @@ bin_DEBUGPROGRAMS += smalliobench
smalliobenchfs_SOURCES = test/bench/small_io_bench_fs.cc test/bench/filestore_backend.cc test/bench/detailed_stat_collector.cc test/bench/bencher.cc
smalliobenchfs_LDADD = librados.la -lboost_program_options $(LIBOS_LDA) $(LIBGLOBAL_LDA)
smalliobenchfs_CXXFLAGS = ${CRYPTO_CXXFLAGS} ${AM_CXXFLAGS}
bin_DEBUGPROGRAMS += smalliobenchfs
smalliobenchdumb_SOURCES = test/bench/small_io_bench_dumb.cc test/bench/dumb_backend.cc test/bench/detailed_stat_collector.cc test/bench/bencher.cc
smalliobenchdumb_LDADD = librados.la -lboost_program_options $(LIBOS_LDA) $(LIBGLOBAL_LDA)
smalliobenchdumb_CXXFLAGS = ${CRYPTO_CXXFLAGS} ${AM_CXXFLAGS}
bin_DEBUGPROGRAMS += smalliobenchdumb
tpbench_SOURCES = test/bench/tp_bench.cc test/bench/detailed_stat_collector.cc
@ -862,6 +864,7 @@ test_mon_workloadgen_SOURCES = \
osdc/Objecter.cc \
osdc/Striper.cc
test_mon_workloadgen_LDADD = $(LIBOS_LDA) $(LIBGLOBAL_LDA)
test_mon_workloadgen_CXXFLAGS = ${CRYPTO_CXXFLAGS} ${AM_CXXFLAGS}
bin_DEBUGPROGRAMS += test_mon_workloadgen
test_rados_api_io_SOURCES = test/librados/io.cc test/librados/test.cc

6
src/client/Client.cc
View File

@ -3656,6 +3656,12 @@ void Client::flush_cap_releases()
void Client::tick()
{
if (cct->_conf->client_debug_inject_tick_delay > 0) {
sleep(cct->_conf->client_debug_inject_tick_delay);
assert(0 == cct->_conf->set_val("client_debug_inject_tick_delay", "0"));
cct->_conf->apply_changes(NULL);
}
ldout(cct, 21) << "tick" << dendl;
tick_event = new C_C_Tick(this);
timer.add_event_after(cct->_conf->client_tick_interval, tick_event);

1
src/common/config_opts.h
View File

@ -183,6 +183,7 @@ OPTION(client_oc_target_dirty, OPT_INT, 1024*1024* 8) // target dirty (keep this
OPTION(client_oc_max_dirty_age, OPT_DOUBLE, 5.0) // max age in cache before writeback
OPTION(client_oc_max_objects, OPT_INT, 1000) // max objects in cache
OPTION(client_debug_force_sync_read, OPT_BOOL, false) // always read synchronously (go to osds)
OPTION(client_debug_inject_tick_delay, OPT_INT, 0) // delay the client tick for a number of seconds
// note: the max amount of "in flight" dirty data is roughly (max - target)
OPTION(fuse_use_invalidate_cb, OPT_BOOL, false) // use fuse 2.8+ invalidate callback to keep page cache consistent
OPTION(fuse_big_writes, OPT_BOOL, true)

6
src/os/FileStore.cc
View File

@ -1227,16 +1227,14 @@ int FileStore::_detect_fs()
} else {
dout(0) << "mount syncfs(2) syscall supported by glibc BUT NOT the kernel" << dendl;
}
#endif
#ifdef SYS_syncfs
#elif defined(SYS_syncfs)
if (syscall(SYS_syncfs, fd) == 0) {
dout(0) << "mount syscall(SYS_syncfs, fd) fully supported" << dendl;
have_syncfs = true;
} else {
dout(0) << "mount syscall(SYS_syncfs, fd) supported by libc BUT NOT the kernel" << dendl;
}
#endif
#ifdef __NR_syncfs
#elif defined(__NR_syncfs)
if (syscall(__NR_syncfs, fd) == 0) {
dout(0) << "mount syscall(__NR_syncfs, fd) fully supported" << dendl;
have_syncfs = true;

5
src/rgw/rgw_cache.h
View File

@ -268,6 +268,11 @@ int RGWCache<T>::get_obj(void *ctx, void **handle, rgw_obj& obj, bufferlist& obl
return r;
}
if (obl.length() == end + 1) {
/* in this case, most likely object contains more data, we can't cache it */
return r;
}
bufferptr p(r);
bufferlist& bl = info.data;
bl.clear();

23
src/rgw/rgw_tools.cc
View File

@ -10,7 +10,7 @@
#define dout_subsys ceph_subsys_rgw
#define READ_CHUNK_LEN (16 * 1024)
#define READ_CHUNK_LEN (512 * 1024)
static map<string, string> ext_mime_map;
@ -41,25 +41,24 @@ int rgw_get_obj(RGWRados *rgwstore, void *ctx, rgw_bucket& bucket, string& key,
bufferlist::iterator iter;
int request_len = READ_CHUNK_LEN;
rgw_obj obj(bucket, key);
ret = rgwstore->prepare_get_obj(ctx, obj, NULL, NULL, pattrs, NULL,
NULL, NULL, NULL, NULL, NULL, NULL, &handle, &err);
if (ret < 0)
return ret;
do {
ret = rgwstore->get_obj(ctx, &handle, obj, bl, 0, request_len - 1);
ret = rgwstore->prepare_get_obj(ctx, obj, NULL, NULL, pattrs, NULL,
NULL, NULL, NULL, NULL, NULL, NULL, &handle, &err);
if (ret < 0)
goto done;
return ret;
ret = rgwstore->get_obj(ctx, &handle, obj, bl, 0, request_len - 1);
rgwstore->finish_get_obj(&handle);
if (ret < 0)
return ret;
if (ret < request_len)
break;
bl.clear();
request_len *= 2;
} while (true);
ret = 0;
done:
rgwstore->finish_get_obj(&handle);
return ret;
return 0;
}
void parse_mime_map_line(const char *start, const char *end)

50
src/test/libcephfs/caps.cc
View File

@ -26,60 +26,14 @@
#include <sys/xattr.h>
#include <signal.h>
void do_sigusr1(int s) {}
// wait_and_suspend() forks the process, waits for the
// child to signal SIGUSR1, suspends the child with SIGSTOP
// sleeps for s seconds, and then unsuspends the child,
// waits for the child to exit, and then returns the exit code
// of the child
static int _wait_and_suspend(int s) {
int fpid = fork();
if (fpid != 0) {
// wait for child to signal
signal(SIGUSR1, &do_sigusr1);
sigset_t set;
sigaddset(&set, SIGUSR1);
int sig;
sigwait(&set, &sig);
// fork and suspend child, sleep for 20 secs, and resume
kill(fpid, SIGSTOP);
sleep(s);
kill(fpid, SIGCONT);
int status;
wait(&status);
if (WIFEXITED(status))
return WEXITSTATUS(status);
return 1;
}
return -1;
}
// signal_for_suspend sends the parent the SIGUSR1 signal
// and sleeps for 1 second so that it can be suspended at the
// point of the call
static void _signal_for_suspend() {
kill(getppid(), SIGUSR1);
}
TEST(Caps, ReadZero) {
int w = _wait_and_suspend(20);
if (w >= 0) {
ASSERT_EQ(0, w);
return;
}
pid_t mypid = getpid();
int mypid = getpid();
struct ceph_mount_info *cmount;
ASSERT_EQ(0, ceph_create(&cmount, NULL));
ASSERT_EQ(0, ceph_conf_read_file(cmount, NULL));
ASSERT_EQ(0, ceph_mount(cmount, "/"));
ASSERT_EQ(0, ceph_conf_set(cmount, "client_cache_size", "10"));
int i = 0;
for(; i < 30; ++i) {
@ -114,7 +68,7 @@ TEST(Caps, ReadZero) {
ASSERT_EQ(expect, caps & expect);
}
_signal_for_suspend();
ASSERT_EQ(0, ceph_conf_set(cmount, "client_debug_inject_tick_delay", "20"));
for(i = 0; i < 30; ++i) {