Loic Dachary [Sun, 17 May 2015 16:09:21 +0000 (18:09 +0200)]
erasure-code: ErasureCodeShec update default values to string
Update the ErasureCodeShec::init function to use string default
values. The default profile values for technique, ruleset-root and
ruleset-failure-domain are set for the user to see.
The error message when an invalid technique is provided is written to
the error stream to be returned to the user instead of being written to
the OSD log.
Updating other values from the profile would require deeper changes and
is left for later.
Loic Dachary [Sun, 17 May 2015 16:08:31 +0000 (18:08 +0200)]
erasure-code: ErasureCodeLrc update default values to string
Update the ErasureCodeLrc::init function to use string default values
and update the profile accordingly.
If the k,m,l values are set and generate mappings and layers entries,
those are not stored in the profile. The details of what k,m,l create
are private and not exposed to the caller.
Loic Dachary [Sun, 17 May 2015 15:58:36 +0000 (17:58 +0200)]
erasure-code: ErasureCodeIsa update default values to string
Update the ErasureCodeIsa::init function to use string default values
and update the profile accordingly.
Also fix a bug by which the plugin would incorrectly report using the
default technique when the reed_sol_van or cauchy techniques are the
only legitimate values.
Loic Dachary [Sun, 17 May 2015 15:53:55 +0000 (17:53 +0200)]
erasure-code: ErasureCode default value is a string
The to_int, to_bool prototypes are modified and the default values are
changed to string instead of int. If a default value is set, the profile
is modified to contain the default value.
The to_string helper is added so it can be used to set the default value
of a string, in a way that is consistent with to_int and to_bool.
Loic Dachary [Thu, 28 May 2015 23:34:45 +0000 (01:34 +0200)]
install-deps.sh: robust pip and wheel installation
The wheel and pip module must not only be installed in each wheelhouse
directory for tox to find. They must also be installed in the virtual
environment that populates the wheelhouse directory. Otherwise older pip
modules such as the one found by default on Ubuntu 12.04 will fail.
Python 2.7.3 on Ubuntu 12.04 also requires that distribute >= 0.7.3 is
installed although it is redundant with setuptools, otherwise it will
fail to run the wheel module.
Loic Dachary [Thu, 28 May 2015 08:35:51 +0000 (10:35 +0200)]
debian: ceph-dbg steals ceph-objectstore-tool from ceph-test-dbg
When ceph-objectstore-tool was moved from ceph-test to
ceph by 61cf5da0b51e2d9578c7b4bca85184317e30f4ca, the ceph package in
debian/control was updated accordingly, as recommended by
https://www.debian.org/doc/debian-policy/ch-relationships.html#s-replaces
The same must be done for the ceph-dbg package because
/usr/lib/debug/usr/bin/ceph-objectstore-too is no longer in
ceph-test-dbg.
Although the change was merged may 6th, 2015 8f23382064c189b657564d58c3f9d17720e891ed, teuthology jobs were not
always failing because packages were not systematically upgraded during
the installation. The missing dependencies that were responsible for
this upgrade problem were fixed by f898ec1e4e3472b0202280f09653a769fc62c8d3 on may 18th, 2015 and all
upgrade tests relying on ceph-*-dbg packages started to fail
systematically after this date.
Kefu Chai [Wed, 20 May 2015 06:24:48 +0000 (14:24 +0800)]
osdmap, mon: switch the params of print_tree()
* change from
print_tree(ostream*, Formatter*)
to
print_tree(Formatter*, ostream*)
this is more consistent with other map's print
functions. e.g.
MDSMap::print_summary(Formatter *f, ostream *out)
* and in print_tree(Formatter* f, ostream* os),
- `f` and `os` will be mutual exclusive.
- will assert(0) if both of them are NULL.
Kefu Chai [Wed, 20 May 2015 06:16:14 +0000 (14:16 +0800)]
osdmaptool: dump 'osd tree' in specified format
* so we are able to dump more info from osdmap using this tool,
and this allows us to reproduce the behavior of "ceph osd tree".
* add a test for 'osdmaptool --tree=<format> filepath'
Samuel Just [Wed, 27 May 2015 18:00:54 +0000 (11:00 -0700)]
ReplicatedPG: start_flush: use filtered snapset
Otherwise, we might send our deletes based on deleted snaps. This is
problematic since we may have trimmed the clones to which those snaps
belong, causing us to send them at an earlier snap than we used before.
The specific situation was
78:[78, 70, 63, 5a, 58, 57]:[64(63), 58(58, 57)]
with 58 already clean. To flush 64, we send:
delete@58
delete@59
copyfrom@62
Then, snap 63 is trimmed leaving us with a snapset of:
78:[78, 70, 63, 5a, 58, 57]:[58(58, 57)]
since trim_object doesn't filter the head object snapset snaps. This
isn't really a bug since in general all snapset users must be aware
that there may be trimmed snaps in snapset::snaps. However, here
it becomes a problem when we go to flush head:
delete@58 -- ignored due to snapc
delete@59 -- ignored due to snapc
copyfrom@78 -- not ignored
The base pool head is at snap seq 62, so it clones that value into
clone 78(78, 70) instead of forgetting it. What should have happened
is that we should have based our flushes on filtered snapset:
78:[78, 70, 58, 57]:[58(58, 57)]
Causing us to instead send:
delete@58 -- ignored due to snapc
delete@69 -- not ignored, causes no clone to be made
copyfrom@78 -- not ignored, updates head such that a subsequent clone
will leave 70 out of the clone snaps vector.
Fixes: 11787 Signed-off-by: Samuel Just <sjust@redhat.com>
Ilya Dryomov [Fri, 15 May 2015 18:44:27 +0000 (21:44 +0300)]
doc: fix crush-ruleset-name param description
Specified crush-ruleset-name is required to exist, implicit creation is
going to happen only if crush-ruleset-name wasn't specified on the
command line. While at it, pool-name is very much a required param.
Validate osd_pool_default_crush_{replicated_ruleset,rule} config
options, in particular when creating pools. Otherwise "ceph osd pool
create foo <pg_num>" may end up creating pools with non-existent
rulesets.
Ilya Dryomov [Thu, 21 May 2015 15:52:52 +0000 (18:52 +0300)]
OSDMap: respect default replicated ruleset config opt in build_simple()
Use id provided by osd_pool_default_crush_{replicated_ruleset,rule}
config options when creating a simple replicated ruleset for an initial
osdmap instead of always making it ruleset 0. Not doing so may leave
default created pools (currently "rbd") in a broken state with their
crush_ruleset pointing to a non-existent ruleset.
Ilya Dryomov [Fri, 22 May 2015 12:50:07 +0000 (15:50 +0300)]
tests: a couple tweaks to osd-pool-create.sh
In TEST_default_deprectated_*(), make expected/unexpected vars local
and actually check that rbd, being a default created pool, is set to
use the ruleset specified by conf.
INVALIDRULESET thing in TEST_replicated_pool() is redundant - it is
checked in TEST_replicated_pool_with_ruleset() a bit earlier.
First cut for a `ceph-release-notes` script added which looks at merge
commits and picks out issue numbers. Though this ideally suits for
backport releases workflow where the commit messages always follow a
specific pattern, it is partly useful for preparing release notes for
normal releases as well.
projects / ceph.git / log