Commit Graph

27 Commits

Author SHA1 Message Date
Prashant D
92eb39ee6f crush/CrushCompiler: print weight with uniform precision
Fixes: https://tracker.ceph.com/issues/48508

Signed-off-by: Prashant D <pdhange@redhat.com>
2021-03-29 14:44:49 +11:00
Sage Weil
33dee7d7bf crush/CrushWrapper: update shadow trees on update_item()
insert_item() already does this, but update_item did not.

Fixes: https://tracker.ceph.com/issues/48065
Signed-off-by: Sage Weil <sage@newdream.net>
2021-02-22 14:21:04 -06:00
Sage Weil
4d42b4c5a0 common/TextTable: default to 2 spaces separating columns
This is what other projects and libraries default to, and it is more
legible.

Signed-off-by: Sage Weil <sage@redhat.com>
2020-02-23 15:46:30 -06:00
Sage Weil
4c741c109d qa/standalone/crush/crush-choose-args: add weight-set tests
Verify we have the expected behavior for creates and moves that
maintain bucket summation, both with and without the
osd_crush_update_weight_set option enabled.

Signed-off-by: Sage Weil <sage@redhat.com>
2019-03-20 04:57:51 -05:00
Sage Weil
f20c736e99 qa/standalone/crush/crush-choose-args: fix test
- Make the initial weight-set actually consistent (summing)
- Fix the intermediate state so that it reflects a correctly
  maintained summation.

Signed-off-by: Sage Weil <sage@redhat.com>
2019-03-20 04:57:51 -05:00
xie xingguo
5bcac35213 mon/OSDMonitor: do not remove device class still referenced by ec-profiles
Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
2018-10-23 21:17:56 +08:00
xie xingguo
4bc54587a1 mon/OSDMonitor: make "ceph osd crush class rm" idempotent
Removing a non-existent device class should be generally okay.

Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
2018-10-23 21:17:56 +08:00
xie xingguo
d7ff33e9fd crush: get "ceph osd crush class create/rm" back
This reverts a27fd9d25c and
b863883ca7.

Quote form Sébastien Han:
> IIRC at some point, we were able to create a device class from the CLI.
Now it seems that the device class gets created when at least one OSD
of a particular class starts.
In ceph-ansible, we create pools after the initial monitors are up and
we want to assign a device crush class on some of them.
That's not possible at the moment since there no device class available yet.
Also, someone might want to create its own device class.
Something as crazy as running Filestore with a tmpfs osd store and
might want to isolate them.
I know it's a very limited use case, but still, it could be desired.

See also https://www.spinics.net/lists/ceph-devel/msg41152.html

Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
2018-10-16 08:45:49 +08:00
Sage Weil
662e883683 qa/standalone/crush/crush-choose-args: run mgr
The osd purge command needs a running mgr.

Fixes: d2b41d4095
Signed-off-by: Sage Weil <sage@redhat.com>
2018-08-07 11:57:05 -05:00
xie xingguo
2ee80aead8 mon/OSDMonitor: make 'osd crush class rename' idempotent
Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
2017-08-29 10:43:35 +08:00
Kefu Chai
30b5b4627c Merge pull request #16494 from asomers/bin_bash
misc: Fix bash path in shebangs

Reviewed-by: Willem Jan Withagen <wjw@digiware.nl>
Reviewed-by: Kefu Chai <kchai@redhat.com>
2017-08-27 10:14:14 +08:00
xie xingguo
d792e8d528 crush: "osd crush class rename" support
In 076a6abd80 I killed the 'class rename' command
and thought it was totally useless but I was wrong.

Consider the following user case:
(1) randomly choose some OSDs(e.g., from different hosts) and try to make them for private use only,
    say, by grouping them into 'pool1'
(2) ceph osd crush set-device-class pool1 'OSDs from (1)'
(3) ceph osd crush rule create-replicated rule_for_pool1 default host pool1
(4) ceph osd pool rename pool1 pool2
(5) ceph osd crush class rename pool1 pool2

From the above user case, we need to safely change a pool name without worrying
any risk of data migration. That is why the 'osd crush class rename' command
is still needed here.

Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
2017-08-11 08:32:39 +08:00
xie xingguo
87952fc68d crush: automatically kill dead classes
If a class is no more referenced by any devices or crush rules,
it shall be considered as dead.

This patch makes Ceph automatically recycles those dead classes,
so user does not to explicitly call 'class rm', which is unsafe
and annoying.

Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
2017-08-05 18:53:39 +08:00
xie xingguo
b863883ca7 crush: remove 'class rm' command
The current version is broken. E.g., it should only remove a class
which is never referenced by any device.

Since we now create new classes automatically, we shall automatically
recycle dead classes too. So this command is definitely unuseful.
(Actually it is weird that we keep 'class rm' without keeping the
 corresponding 'class create' command).

Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
2017-08-05 18:52:30 +08:00
xie xingguo
f1d80ff750 crush: do not automatically recycle class for 'rm-device-class'
This will prevent the current crush rule from referencing a non-existent
shadow tree and hence avoid a coredump such as below:

 0> 2017-08-05 09:54:19.943349 7f73887d6700 -1 /clove/vm/xxg/rpm/ceph/rpmbuild/BUILD/ceph-12.1.2.1/src/crush/CrushWrapper.cc: In function 'int CrushWrapper::get_rule_weight_osd_map(unsigned
 int, std::map<int, float>*)' thread 7f73887d6700 time 2017-08-05 09:54:19.941291
/clove/vm/xxg/rpm/ceph/rpmbuild/BUILD/ceph-12.1.2.1/src/crush/CrushWrapper.cc: 1631: FAILED assert(b)

 ceph version 12.1.2.1-11-gd0f812a (d0f812a3a757b319c26794f558b57770663ab324) luminous (rc)
 1: (ceph::__ceph_assert_fail(char const*, char const*, int, char const*)+0x110) [0x7f7398b66ea0]
 2: (CrushWrapper::get_rule_weight_osd_map(unsigned int, std::map<int, float, std::less<int>, std::allocator<std::pair<int const, float> > >*)+0x54e) [0x7f7398daac4e]
 3: (PGMap::get_rule_avail(OSDMap const&, int) const+0x68) [0x7f73989a6428]
 4: (PGMap::get_rules_avail(OSDMap const&, std::map<int, long, std::less<int>, std::allocator<std::pair<int const, long> > >*) const+0x35c) [0x7f73989b748c]
 5: (PGMap::encode_digest(OSDMap const&, ceph::buffer::list&, unsigned long) const+0x16) [0x7f73989b7506]
 6: (DaemonServer::send_report()+0x2a4) [0x7f73989f5474]
 7: (DaemonServer::maybe_ready(int)+0x2f9) [0x7f73989f6129]
 8: (DaemonServer::ms_dispatch(Message*)+0xce) [0x7f73989ff68e]
 9: (DispatchQueue::entry()+0x792) [0x7f7398dd2a22]
 10: (DispatchQueue::DispatchThread::entry()+0xd) [0x7f7398c1429d]
 11: (()+0x7df3) [0x7f739640cdf3]
 12: (clone()+0x6d) [0x7f73954f23ed]

Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
2017-08-05 18:44:59 +08:00
Alan Somers
3aae5ca6fd scripts: fix bash path in shebangs
/bin/bash is a Linuxism.  Other operating systems install bash to
different paths.  Use /usr/bin/env in shebangs to find bash.

Signed-off-by: Alan Somers <asomers@gmail.com>
2017-07-27 13:24:26 -06:00
Sage Weil
e469a8044c qa/standalone/crush/crush-classes: fix test
Signed-off-by: Sage Weil <sage@redhat.com>
2017-07-27 12:25:25 -04:00
xie xingguo
076a6abd80 crush: kill 'class rename'
Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
2017-07-26 22:40:50 +08:00
xie xingguo
a27fd9d25c crush: kill "class create" command
The device class is now self and automatically managed.

Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
2017-07-26 22:40:17 +08:00
xie xingguo
edd8930346 crush: allow "crush class rm" to automatically recycle shadow tree(s)
Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
2017-07-26 22:39:41 +08:00
xie xingguo
9d908c14f6 crush: rm-device-class support
Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
2017-07-26 22:39:08 +08:00
xie xingguo
32fb548797 crush: guard set-device-class
If a device has already been bounded to a class,
do not allow to change its class silently.
Require user call rm-device-class first.

Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
2017-07-26 22:34:08 +08:00
xie xingguo
e4e83a0dd7 crush: fix class_is_in_use()
A class can be considered as in-use only if it is referenced by
any of the existing crush rules.

The patch also makes the output more human readable. For example:

./bin/ceph osd crush rule create-replicated myrule default host ssd
./bin/ceph osd crush class rm ssd
Error EBUSY: class 'ssd' still referenced by crush_rule 'myrule'

Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
2017-07-26 22:31:39 +08:00
xie xingguo
f3a3180cca crush: rebuild shadow tree on "crush create-or-move/move"
This patch solves the problem below:

./bin/ceph osd crush move osd.0 root=foo rack=foo-rack host=foo-host
moved item id 0 name 'osd.0' to location {host=foo-host,rack=foo-rack,root=foo} in crush map

 ./bin/ceph osd crush rule create-replicated foo-rule foo host ssd
Error EINVAL: root foo has no devices with class ssd

Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
2017-07-26 22:30:59 +08:00
xie xingguo
10bf2a633f crush: fix "crush create-or-move/move" would drop osd's class
Was:
     ./bin/ceph osd tree
    ID CLASS WEIGHT  TYPE NAME                                        UP/DOWN REWEIGHT PRI-AFF
    -1       3.00000 root default
    -2       3.00000     host gitbuilder-ceph-rpm-centos7-amd64-basic
     0   ssd 1.00000         osd.0                                         up  1.00000 1.00000
     1   ssd 1.00000         osd.1                                         up  1.00000 1.00000
     2   ssd 1.00000         osd.2                                         up  1.00000 1.00000

    ./bin/ceph osd crush move osd.0 root=foo rack=foo-rack  host=foo-host
    moved item id 0 name 'osd.0' to location {host=foo-host,rack=foo-rack,root=foo} in crush map

     ./bin/ceph osd tree
    ID CLASS WEIGHT  TYPE NAME                                        UP/DOWN REWEIGHT PRI-AFF
    -7       1.00000 root foo
    -6       1.00000     rack foo-rack
    -5       1.00000         host foo-host
     0       1.00000             osd.0                                     up  1.00000 1.00000
    -1       2.00000 root default
    -2       2.00000     host gitbuilder-ceph-rpm-centos7-amd64-basic
     1   ssd 1.00000         osd.1                                         up  1.00000 1.00000
     2   ssd 1.00000         osd.2                                         up  1.00000 1.00000

    Now:
    ./bin/ceph osd tree
    ID CLASS WEIGHT  TYPE NAME                                        UP/DOWN REWEIGHT PRI-AFF
    -1       3.00000 root default
    -2       3.00000     host gitbuilder-ceph-rpm-centos7-amd64-basic
     0   ssd 1.00000         osd.0                                         up  1.00000 1.00000
     1   ssd 1.00000         osd.1                                         up  1.00000 1.00000
     2   ssd 1.00000         osd.2                                         up  1.00000 1.00000

    ./bin/ceph osd crush move osd.0 root=foo rack=foo-rack  host=foo-host
    moved item id 0 name 'osd.0' to location {host=foo-host,rack=foo-rack,root=foo} in crush map

    ./bin/ceph osd tree
    ID CLASS WEIGHT  TYPE NAME                                        UP/DOWN REWEIGHT PRI-AFF
    -7       1.00000 root foo
    -6       1.00000     rack foo-rack
    -5       1.00000         host foo-host
     0   ssd 1.00000             osd.0                                     up  1.00000 1.00000
    -1       2.00000 root default
    -2       2.00000     host gitbuilder-ceph-rpm-centos7-amd64-basic
     1   ssd 1.00000         osd.1                                         up  1.00000 1.00000
     2   ssd 1.00000         osd.2                                         up  1.00000 1.00000

Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
2017-07-26 22:30:26 +08:00
Sage Weil
cabad62242 qa/standalone/ceph-helpers: factor rbd pool create out of run_mon
Signed-off-by: Sage Weil <sage@redhat.com>
2017-07-24 22:11:50 -04:00
Sage Weil
71ea171604 qa: move ceph-helpers and misc src/test/*.sh tests to qa/standalone
- stop running via make check
- add teuthology yamls to run them
- disable ceph_objecstore_tool.py for now (too slow for make check, and
we can't use vstart in teuthology via a package install)
- drop cephtool tests since those are already covered by other teuthology
tests
- leave a handful of (fast!) ceph-helpers tests for make check for minimal
integration tests.

Signed-off-by: Sage Weil <sage@redhat.com>
2017-07-24 22:11:49 -04:00