neorados/test: Support erasure coded pools

The test framework in common test can now support erasure-coded pools
and be more easily extended generally.

Signed-off-by: Adam Emerson <aemerson@redhat.com>

neorados: Take `claim_append`ed `buffer::list`s by value

Since both const lvalue and rvalue make sense, instead of duplicating
everything, just take a value and move from it.

Signed-off-by: Adam Emerson <aemerson@redhat.com>

test/neorados: Rounding out test coverage, part 1

This includes cls, cmd, and read_operations.

Signed-off-by: Adam Emerson <aemerson@redhat.com>

neorados: Implement `ReadOp::checksum`

Rather than using `buffer::list`s for initial values and hash results,
serialize to/from the expected types.

Signed-off-by: Adam Emerson <aemerson@redhat.com>

test/neorados: common_tests gets functions for bufferlists

Mostly for getting things into and out of them conveniently.

Signed-off-by: Adam Emerson <aemerson@redhat.com>

neorados: Improve `cmp_omap` argument structure

A map from strings to value/op pairs is a bit much and a bit annoying
for what's essentially a list of assertions. Just use a vector.

Also put op in the middle so it matches `cmpxattr`.

Signed-off-by: Adam Emerson <aemerson@redhat.com>

neorados: Change `cmpxattr_op` to `cmp_op` and use in `Op::cmp_omap`

Omap and Xattr comparison use the same values, but in the Xattr case
it's eight bits wide.

Rename the enum to `cmp_op` to reflect this, and open-code encoding
the assertions so we don't have to allocate an intermediate structure.

Signed-off-by: Adam Emerson <aemerson@redhat.com>

neorados: Fix `Op::cmpext` error and unfound behavior

A failure will produce the output `osd_errc::cmpext_failed`. On
failure, the `unfound` out parameter, if provided, will be set to the
index of the first nonmatching character; On success, it will be set
to `-1`.

Signed-off-by: Adam Emerson <aemerson@redhat.com>

common/ceph_json: `encode_json()` for `std::array`

No decode, though, since JSON arrays can vary in length and
`std::array` cannot.

Signed-off-by: Adam Emerson <aemerson@redhat.com>

neorados: Op builder functions should chain

Make every operation function return the (Read/Write)Op, for chaining.

Signed-off-by: Adam C. Emerson <aemerson@redhat.com>

neorados: Use `asio::any_completion_handler`

As we'd like to reduce (and eliminate) internal Ceph dependencies to
the extent possible, now that Boost.Asio has a type-erased handler
type, let's use it.

Signed-off-by: Adam C. Emerson <aemerson@redhat.com>

cmake: drop BOOST_ASIO_USE_TS_EXECUTOR_AS_DEFAULT

this define was added for backward compatibility before boost 1.74

https://www.boost.org/doc/libs/1_74_0/doc/html/boost_asio/std_executors.html

Signed-off-by: Casey Bodley <cbodley@redhat.com>

submodule: update spawn submodule for any_io_executor

Signed-off-by: Casey Bodley <cbodley@redhat.com>

test/common: test_blocked_completion uses boost::asio::post()

Signed-off-by: Casey Bodley <cbodley@redhat.com>

librbd: use strand template and make_strand()

Signed-off-by: Casey Bodley <cbodley@redhat.com>
Signed-off-by: Adam C. Emerson <aemerson@redhat.com>

neorados: use strand template and make_strand()

Signed-off-by: Casey Bodley <cbodley@redhat.com>

common/async: SharedMutex uses free function post

Signed-off-by: Casey Bodley <cbodley@redhat.com>

common/async: completion uses free functions for defer/dispatch/post

Signed-off-by: Casey Bodley <cbodley@redhat.com>

osdc: replace asio::io_context::strand with asio::strand<>

Signed-off-by: Casey Bodley <cbodley@redhat.com>

osdc: Catch exceptions thrown in CLS client decoders

And return them to the client by setting the error code and result in
the vector and returning an error from the operation as a whole.

Signed-off-by: Adam C. Emerson <aemerson@redhat.com>

test/neorados: Harness and convenience for Neorados tests

Google Test does not support C++ coroutines, so kludge together a test
harness that supports coroutines reasonably well.

Also add a couple utility functions.

Signed-off-by: Adam C. Emerson <aemerson@redhat.com>

common/async: Don't get allocator from moved-from handler

In the `destroy_` functions of `CompletionImpl` we were getting the
associated allocator after moving out of the handler into the call to
`bind_and_forward`. This was triggering a crash on null-pointer access
in operations made with `co_composed`.

Signed-off-by: Adam C. Emerson <aemerson@redhat.com>

neorados: Use C++20 Coroutines in exerciser/demo

Just as a demonstration to see how well they work and how to put
things together with them.

Signed-off-by: Adam C. Emerson <aemerson@redhat.com>

neorados: Fix logic error in exerciser

Use `min` rather than `max` when deciding how much we need to read in
`read()`.

Signed-off-by: Adam C. Emerson <aemerson@redhat.com>

neorados: Ensure we don't have dangling references

Apparently in some cases we were using references to objects on the
caller stack that happen to no longer exist if the caller's stack goes
away.

C++20 Coroutines tickled this bug.

Signed-off-by: Adam C. Emerson <aemerson@redhat.com>

neorados: Don't call dispatch inside with_osdmap

The lock will continue to be held over the 'dispatch' with C++20
coroutines.

Signed-off-by: Adam C. Emerson <aemerson@redhat.com>

neorados: Switch to async_initiate

This form of completion handling is compatible with C++20 Completions
and generally more flexible stuff.

Signed-off-by: Adam C. Emerson <aemerson@redhat.com>

neorados: Remove 'broken out' versions of calls

Previously we had versions of all the calls that took an int64_t and
optional key and namespace.

This didn't really offer much benefit and doubled the maintenance
burden for changing anything.

As such just make the IOContext constructor non-explicit, and make all
its mutators return the same IOContext so people can supply them
builder-style if they want to.

Only wrap some things in optional, when it actually makes it easier.

Signed-off-by: Adam C. Emerson <aemerson@redhat.com>

Merge pull request #53296 from phlogistonjohn/jjm-fix-git-ls-files

cephadm: remove duplicates when listing files in scan

Reviewed-by: Adam King <adking@redhat.com>

Merge pull request #52690 from phlogistonjohn/jjm-cephadm-daemon-ident

cephadm: refactor code to make use of DaemonIdentity type

Reviewed-by: Adam King <adking@redhat.com>

Merge pull request #53450 from zdover23/wip-doc-2023-09-14-man-ceph-monstore-tool

doc/man: edit ceph-monstore-tool.rst

Reviewed-by: Anthony D'Atri <anthony.datri@gmail.com>

Merge pull request #53426 from ronen-fr/wip-rf-fshut

osd: cleaning stop_for_fast_shutdown()

Reviewed-by: Radoslaw Zarzynski <rzarzyns@redhat.com>
Reviewed-by: Samuel Just <sjust@redhat.com>

Merge PR #53070 into main

* refs/pull/53070/head:
PendingReleaseNotes: add note about 'm' and 'M' periods
doc/cephfs: add note about monthly scheduling and retention period
qa: update minute period multiplier to m
qa: change minute duration specifier from M to m
mgr/snap_schedule: change minute duration specifier from M to m

Reviewed-by: Dhairya Parmar <dparmar@redhat.com>
Reviewed-by: Neeraj Pratap Singh <neesingh@redhat.com>
Reviewed-by: Venky Shankar <vshankar@redhat.com>

Merge pull request #53433 from myoungwon/fix-53232

crimson/tools/perf_staged_fltree: fix compile error

Reviewed-by: Yingxin Cheng <yingxin.cheng@intel.com>
Reviewed-by: Aravind Ramesh <aravind.ramesh@wdc.com>

Merge PR #51891 into main

* refs/pull/51891/head:
qa: fix cephfs-top tests for the new fields
cephfs-top: include the missing fields in --dump output

Reviewed-by: Neeraj Pratap Singh <neesingh@redhat.com>

Merge pull request #53444 from zdover23/wip-doc-2023-09-14-architecture-4-of-x

doc/architecture: edit "High Avail. Monitors"

Reviewed-by: Anthony D'Atri <anthony.datri@gmail.com>

doc/man: edit ceph-monstore-tool.rst

Clean up the English in doc/man/8/ceph-monstore-tool.rst. Put all
commands in the imperative mood; make sentences more direct.

Follows https://github.com/ceph/ceph/pull/52860.

Signed-off-by: Zac Dover <zac.dover@proton.me>

doc/architecture: edit "High Avail. Monitors"

Improve the sentence structure in the "High Availability Monitors"
section of doc/architecture.rst.

Co-authored-by: Anthony D'Atri <anthony.datri@gmail.com>
Signed-off-by: Zac Dover <zac.dover@proton.me>

Merge pull request #53178 from aclamk/wip-aclamk-bs-esb-5-enabled

bluestore: Elastic Shared Blobs 5 - enabled, final

Reviewed-by: Igor Fedotov <ifedotov@suse.com>
Reviewed-by: Radoslaw Zarzynski <rzarzyns@redhat.com

cephadm: fix an error running iscsi services

Signed-off-by: John Mulligan <jmulligan@redhat.com>

Merge pull request #52793 from linuxbox2/wip-fix-test-librgw-xattr

rgw/file:  fix build of librgw_file_xattr.cc unit test

Merge pull request #51368 from dang/wip-dang-zipper-motr

Fix MOTR build, and move MOTR and DAOS to proper location

Merge pull request #53391 from shreyanshjain7174/bucket_quota

RadosGW API: incorrect bucket quota in response to HEAD /{bucket}/?usage

Reviewed-by: Ali Maredia <amaredia@redhat.com>

crimson/tools/perf_staged_fltree: fix compile error

Signed-off-by: Myoungwon Oh <myoungwon.oh@samsung.com>

Merge pull request #53423 from zdover23/wip-doc-2023-09-12-architecture-3-of-x

doc/architecture: edit "Cluster Map"

Reviewed-by: Anthony D'Atri <anthony.datri@gmail.com>

Merge PR #53153 into main

* refs/pull/53153/head:
Revert "mds: align quota.max_bytes to 4MB or 4KB"

Reviewed-by: Xiubo Li <xiubli@redhat.com>

Merge pull request #53370 from wowly/rgw-cache-nginx-if-match-header

examples: fixed errors with the if-match header in the examples of rgw-cache nginx configs

Reviewed-by: Casey Bodley <cbodley@redhat.com>

doc/architecture: edit "Cluster Map"

Edit the section "Cluster Map" in doc/architecture.rst.

Co-authored-by: Anthony D'Atri <anthony.datri@gmail.com>
Signed-off-by: Zac Dover <zac.dover@proton.me>

Merge pull request #53422 from zdover23/wip-doc-2023-09-12-glossary-primary-affinity

doc/glossary: add "primary affinity" to glossary

osd: cleaning stop_for_fast_shutdown()

Removed unsued variables to prevent compiler warnings.
Protected the shard lock.

Signed-off-by: Ronen Friedman <rfriedma@redhat.com>

Merge pull request #52407 from rhcs-dashboard/osd-op-scraper

os/bluestore: osd scrape read/write operations

os/bluestore: osd scrape read/write operations

To run:
```bash
../src/test/objectstore/allocsim/ops_scraper.py --ceph_bin_path bin/ceph
```

Example:
```bash
$ ../src/test/objectstore/allocsim/ops_scraper.py --ceph_bin_path bin/ceph                                                                           [13:06:39]
collecting from osd. 0
op_type | offset/extent | name | pgid | initiated_at | who
-------------------------------------------------------
write 0~4194304 4:c4b97d2c:::be151868021c187609fb97e4c22ae1f9:head 4.3 2023-07-12T13:06:34.959082+0200 client.4831.0:351
write 0~4194304 4:d392075e:::b068dce954dca8ca56d72e02e87f39c8:head 4.b 2023-07-12T13:06:35.093439+0200 client.4831.0:358
write 0~4194304 4:bc1ac643:::7b815bd25a40695c891dc06bdf2b65be:head 4.1d 2023-07-12T13:06:35.128097+0200 client.4831.0:360
write 0~4194304 4:d0d21bd5:::cc531199dd2707370bb461cce82bdef6:head 4.b 2023-07-12T13:06:36.023082+0200 client.4831.0:367
write 0~4194304 4:a9abc92c:::0703d6f43d14bc33d69f94ecb96c522c:head 4.15 2023-07-12T13:06:36.367886+0200 client.4831.0:369
write 0~4194304 4:9b4684ea:::d5856dc0ecad96f811012e0937d5e0ac:head 4.19 2023-07-12T13:06:36.654584+0200 client.4831.0:370
write 0~4194304 4:d15b8c87:::c3344e5de8c8936876965d9fb056ad89:head 4.b 2023-07-12T13:06:36.835690+0200 client.4831.0:371
write 0~4194304 4:ad41de2a:::8a1e2743b577a67d6bc8a1d514391b53:head 4.15 2023-07-12T13:06:37.102741+0200 client.4831.0:373
write 0~4194304 4:fc35d54f:::39cd961e7e33680f4bfd1ca0fe43dda4:head 4.1f 2023-07-12T13:06:38.047612+0200 client.4831.0:377

collecting from osd. 1
op_type | offset/extent | name | pgid | initiated_at | who
-------------------------------------------------------
write 0~4194304 4:4c9ed2e8:::86206ccc35d8aceaf78670a751a6b550:head 4.12 2023-07-12T13:06:34.771423+0200 client.4831.0:340
write 0~4194304 4:106d9524:::fe7534cc604b0eb50498cb23aaca9161:head 4.8 2023-07-12T13:06:35.128105+0200 client.4831.0:359
write 0~4194304 4:a6e3fdb1:::2692491bfc4c584ca5b1efcb896936c2:head 4.5 2023-07-12T13:06:35.567070+0200 client.4831.0:363
write 0~4194304 4:2d9a471a:::ba51255646d00c496c2638bad03e0869:head 4.14 2023-07-12T13:06:35.568290+0200 client.4831.0:364
write 0~4194304 4:2b1381b1:::c58fa31e5d3a4f5550ee143bcb07a49f:head 4.14 2023-07-12T13:06:35.764699+0200 client.4831.0:365
write 0~4194304 4:a22e158f:::b2116e08700da18595b96d58120fe81c:head 4.5 2023-07-12T13:06:36.023101+0200 client.4831.0:366
```

Signed-off-by: Pere Diaz Bou <pere-altea@hotmail.com>

Merge pull request #53419 from AliMasarweh/wip-alimasa-persistent-q-test

RGW | bucket notification: adding more sleeping time to prevent test from failing

Reviewed-by: yuvalif <ylifshit@redhat.com>

qa: fix cephfs-top tests for the new fields

Fixes: https://tracker.ceph.com/issues/61397
Signed-off-by: Jos Collin <jcollin@redhat.com>

cephfs-top: include the missing fields in --dump output

Fixes: https://tracker.ceph.com/issues/61397
Signed-off-by: Jos Collin <jcollin@redhat.com>

doc/glossary: add "primary affinity" to glossary

Add "Primary Affinity" to the glossary and link to the information in
the RADOS documentation that (more or less) explains it.

Signed-off-by: Zac Dover <zac.dover@proton.me>

RGW | bucket notification: adding more sleeping time to prevent test from failing

Signed-off-by: Ali Masarwa <ali.saed.masarwa@gmail.com>

PendingReleaseNotes: add note about 'm' and 'M' periods

Signed-off-by: Milind Changire <mchangir@redhat.com>

doc/cephfs: add note about monthly scheduling and retention period

Fixes: https://tracker.ceph.com/issues/62494
Signed-off-by: Milind Changire <mchangir@redhat.com>

qa: update minute period multiplier to m

Signed-off-by: Milind Changire <mchangir@redhat.com>

qa: change minute duration specifier from M to m

Fixes: https://tracker.ceph.com/issues/62494
Signed-off-by: Milind Changire <mchangir@redhat.com>

mgr/snap_schedule: change minute duration specifier from M to m

Problem:
As per the issue tracker, the period spec specifier 'M' is not
consistent with what is used elsewhere, like the period specifiers
displayed in the 'ceph status' command output.
The 'M' period specifier is used as a 'minute' level period specifier by
the cephfs team.
The issue reporter suggests to use 'M' as a 'month' period specifier.

Solution:
Since the 'minute' level period specifer, 'M', is used internally by
the development team, it is failrly easy to swap the 'minute' ('M')
level and 'month' ('m') level period specifers to finally mean that 'm'
implies 'minute' level period and 'M' implies 'month' level period.
Also, since this is the first time that somebody has ever reported that
neither the 'M' nor the 'm' level specifiers work in production, it is a
good idea to fix them once and for all.

Fixes: https://tracker.ceph.com/issues/62494
Signed-off-by: Milind Changire <mchangir@redhat.com>

Merge pull request #53408 from roysahar/nvmeof_fix_omap_state_object_contains_None

nvmeof-gw: omap object name contains None as string due to generated template to ceph-nvmeof.conf

Merge pull request #53232 from myoungwon/wip-enable-rbm-tests

test/crimson/seastore/rbm: add sub-tests regarding RBM to the existing tests

Reviewed-by: Yingxin Cheng <yingxin.cheng@intel.com>
Reviewed-by: Liu-Chunmei <chunmei.liu@intel.com>

crimson/os/seastore/object_data_handler: consider a RBM case when checking if write can be merged

RBM's paddr always indicates physical address, which means it doesn't have the dealayed.
So, this commit adds a condition that checks if given paddr is used for ongoing write.

Signed-off-by: Myoungwon Oh <myoungwon.oh@samsung.com>

Merge pull request #53305 from xxhdx1985126/wip-seastore-onode-erase-replay

crimson/os/seastore/onode_manager: populate value recorders of onodes to be erased

Reviewed-by: Yingxin Cheng <yingxin.cheng@intel.com>

nvmeof-gw: omap object name contains None due to generated template contains string None

Signed-off-by: Roy Sahar <royswi@gmail.com>

Merge pull request #53344 from ljflores/wip-tracker-62761

common: add CephContext parameter to tracing::Tracer::init() in !HAVE_JAEGER branch

cephadm: convert _write_iscsi_unit_poststop_commands to to use a DaemonIdentity argument

Signed-off-by: John Mulligan <jmulligan@redhat.com>

cephadm: convert _write_osd_unit_poststop_commands to to use a DaemonIdentity argument

Signed-off-by: John Mulligan <jmulligan@redhat.com>

cephadm: convert _write_iscsi_unit_run_commands to to use a DaemonIdentity argument

Signed-off-by: John Mulligan <jmulligan@redhat.com>

cephadm: convert _write_osd_unit_run_commands to to use a DaemonIdentity argument

Signed-off-by: John Mulligan <jmulligan@redhat.com>

cephadm: convert _dispatch_deploy to to use a DaemonIdentity argument

Signed-off-by: John Mulligan <jmulligan@redhat.com>

cephadm: convert get_deployment_type to to use a DaemonIdentity argument

Signed-off-by: John Mulligan <jmulligan@redhat.com>

cephadm: convert get_deployment_container to to use a DaemonIdentity argument

Signed-off-by: John Mulligan <jmulligan@redhat.com>

cephadm: convert get_container to to use a DaemonIdentity argument

Signed-off-by: John Mulligan <jmulligan@redhat.com>

cephadm: add daemon_name property to DaemonIdentity

Add a convenience property to DaemonIdentity and DaemonSubIdentity that
produces a name in the ceph daemon naming convention: <type>.<id>.

Signed-off-by: John Mulligan <jmulligan@redhat.com>

cephadm: convert deploy_daemon to to use a DaemonIdentity argument

Signed-off-by: John Mulligan <jmulligan@redhat.com>

cephadm: convert deploy_daemon_units to to use a DaemonIdentity argument

Signed-off-by: John Mulligan <jmulligan@redhat.com>

cephadm: convert get_daemon_args to use DaemonIdentity argument

Signed-off-by: John Mulligan <jmulligan@redhat.com>

cephadm: assert that DaemonIdentity has valid values

Signed-off-by: John Mulligan <jmulligan@redhat.com>

cephadm: convert get_container_mounts to use a DaemonIdentity argument

Convert get_container_mounts to use a DaemonIdentity argument, and in
order to do so: split get_container_mounts_for_type out of
get_container_mounts. get_container_mounts will take a DaemonIdentity
and safely assume that the fsid, daemon_type, daemon_id triple is
available whereas get_container_mounts_for_type has the functionality
of only requiring the fsid and daemon_type.

Signed-off-by: John Mulligan <jmulligan@redhat.com>

cephadm: convert get_container_binds to use a DaemonIdentity argument

Signed-off-by: John Mulligan <jmulligan@redhat.com>

cephadm: convert _write_custom_conf_files to use a DaemonIdentity arg

Signed-off-by: John Mulligan <jmulligan@redhat.com>

src/common: add context to tracing::Tracer::init

Followup to https://github.com/ceph/ceph/pull/50948. This wasn't originally
caught since the centos 8 default build passed, but it did fail the
crimson build: https://shaman.ceph.com/builds/ceph/wip-yuri3-testing-2023-08-15-0955/b5259484dbc61e8573f27e692560c6c107b9d4c1/

Fixes: https://tracker.ceph.com/issues/62761
Signed-off-by: Laura Flores <lflores@ibm.com>

cephadm: convert create_daemon_dirs to use a DaemonIdentity argument

Signed-off-by: John Mulligan <jmulligan@redhat.com>

cephadm: replace get_data_dir & make_data_dir args with DaemonIdentity

Signed-off-by: John Mulligan <jmulligan@redhat.com>

cephadm: split get_unit_name into two functions

The previous version of get_unit_name was effectively overloaded,
so that you could pass either an fsid, daemon_type, & daemon_id OR
pass an fsid and a "daemon_type" only that wasn't actually a daemon_type
but rather the instance name of the systemd service. Separate the
two use cases into two functions, making the naming more correct and
the code more audit-able overall.

Signed-off-by: John Mulligan <jmulligan@redhat.com>

cephadm: add unit_name property to DaemonIdentity

Replace some uses of get_unit_name with the new method.

Signed-off-by: John Mulligan <jmulligan@redhat.com>

cephadm: use DaemonIdentity in arguments to CephContainer.for_daemon

Previously, the CephContainer.for_daemon took separate arguments for
fsid, daemon_type, & daemon_id. Now that we have a dedicated type
for encapsulating those values, switch for_daemon to take
a DaemonIdentity instead.

Signed-off-by: John Mulligan <jmulligan@redhat.com>

cephadm: split DaemonSubIdentity class off from DaemonIdentity

Split DaemonSubIdentity into it's own subclass of DaemonIdentity
in order to better use the type system for validation and clarity.
DaemonIdentity loses the `subcomponent` field so that there's no
mistaking a DaemonIdentity with a DaemonSubIdentity.
This will make a lot more sense once there are more uses of
DaemonSubIdentity in the codebase.

Signed-off-by: John Mulligan <jmulligan@redhat.com>

Merge pull request #53390 from rishabh-d-dave/cephfs-doc-admin

doc/cephfs: write cephfs commands fully in docs

Reviewed-by: Anthony D'Atri <anthony.datri@gmail.com>
Reviewed-by: Zac Dover <zac.dover@proton.me>
Reviewed-by: Dhairya Parmar <dparmar@redhat.com>

Merge pull request #53361 from ronen-fr/wip-rf-moveit

rgw/test: qualifying 'move'

Reviewed-by: Mark Kogan <mkogan@redhat.com>
Reviewed-by: Yuval Lifshitz <ylifshit@redhat.com>

Merge pull request #53369 from yuvalif/wip-yuval-62784

rgw/notifications: allow cross tenant notification management

reviewed-by: mattbenjamin

Merge pull request #52043 from adk3798/tcmu-entrypoint

cephadm: run tcmu-runner through script to do restart on failure

Reviewed-by: John Mulligan <jmulligan@redhat.com>

Merge pull request #52881 from adk3798/upgrade-test-start-squid

qa/cephadm: start upgrade tests from quincy

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

RadosGW API: incorrect bucket quota in response to HEAD /{bucket}/?usage

When we try to get the bucket usage via various methods, through curl or while accessing rgw api endpoint at HEAD /{bucket}/?usage doesn't return the updated information. The endpoint was always returning the user quota and not the actual bucket quota which we see after querying the endpoint.

Fixes: https://tracker.ceph.com/issues/62737
Signed-off-by: shreyanshjain7174 <ssanchet@redhat.com>

Merge pull request #52708 from vshankar/wip-62236

qa: move nfs (mgr/nfs) related tests to fs suite

Reviewed-by: Adam King <adking@redhat.com>

Merge pull request #53343 from markhpc/wip-rocksdb-compression

common/options: Set LZ4 compression for bluestore RocksDB.

Merge pull request #53072 from adk3798/tcmu-custom-configs

cephadm: make custom_configs work for tcmu-runner container

Reviewed-by: John Mulligan <jmulligan@redhat.com>