common: extract options into yaml

extract the options in common/options.cc into separate .yaml.in
files, and preprocess them using CMake before translating them into .cc
files using a python script.

this change paves the road to render the options using sphinx, and
will allow us to further annotate the options to include more metadata.

also, a this YAML file can be consumed by applications like dashboard
and Sphinx to consume these metadata in a simpler way.

* use @variable-name@ for substituting the variables in .yaml.in file
* use cmake variable of `mgr_disabled_modules` instead of C macro
  to define `mgr_disabled_modules` in global.yaml.in
* debian/control, ceph.spec.in, win32_deps_build.sh: add python3-yaml
  as build dep
* add y2c.py (short for YAML to C++) to translate .yaml to .cc file
* common/options/*.yaml.in: extract and split options into .yaml.in
  files, the subvars in it is then replaced with CMake variables,
  and copied to the corresponding .yaml files
* include/config-h.in.cmake: remove MGR_DISABLED_MODULES, as it
  is not a CMake variable.

Signed-off-by: Kefu Chai <kchai@redhat.com>

CMakeLists.txt

@@ -487,11 +487,6 @@ if(WITH_MGR)
   set(MGR_PYTHON_LIBRARIES ${Python3_LIBRARIES})
   set(MGR_PYTHON_VERSION_MAJOR ${Python3_VERSION_MAJOR})
   set(MGR_PYTHON_VERSION_MINOR ${Python3_VERSION_MINOR})
-  # https://tracker.ceph.com/issues/45147
-  if(Python3_VERSION VERSION_GREATER_EQUAL 3.8)
-    set(MGR_DISABLED_MODULES "diskprediction_local")
-    message(STATUS "mgr module disabled for ${Python3_VERSION}: ${MGR_DISABLED_MODULES}")
-  endif()
   # Boost dependency check deferred to Boost section
 endif(WITH_MGR)
 endif(NOT WIN32)

ceph.spec.in

@@ -327,6 +327,7 @@ BuildRequires:  cunit-devel
 BuildRequires:	python%{python3_pkgversion}-setuptools
 BuildRequires:	python%{python3_pkgversion}-Cython
 BuildRequires:	python%{python3_pkgversion}-PrettyTable
+BuildRequires:	python%{python3_pkgversion}-PyYAML
 BuildRequires:	python%{python3_pkgversion}-Sphinx
 BuildRequires:  rdma-core-devel
 BuildRequires:	liblz4-devel >= 1.7
@@ -349,6 +350,7 @@ BuildRequires:	python%{python3_pkgversion}-devel
 BuildRequires:	python%{python3_pkgversion}-setuptools
 BuildRequires:	python%{python3_pkgversion}-Cython
 BuildRequires:	python%{python3_pkgversion}-prettytable
+BuildRequires:	python%{python3_pkgversion}-pyyaml
 BuildRequires:	python%{python3_pkgversion}-sphinx
 BuildRequires:	lz4-devel >= 1.7
 %endif

debian/control

@@ -98,7 +98,7 @@ Build-Depends: automake,
                python3-setuptools,
                python3-sphinx,
 # Make-Check   python3-werkzeug,
-               python3-setuptools,
+               python3-yaml,
 # Crimson      ragel,
 # Make-Check   socat,
 # Crimson      systemtap-sdt-dev,

src/CMakeLists.txt

@@ -447,6 +447,7 @@ add_subdirectory(arch)
 set(ceph_common_objs
   $<TARGET_OBJECTS:common-auth-objs>
   $<TARGET_OBJECTS:common-common-objs>
+  $<TARGET_OBJECTS:common-options-objs>
   $<TARGET_OBJECTS:common-msg-objs>
   $<TARGET_OBJECTS:common_buffer_obj>
   $<TARGET_OBJECTS:common_texttable_obj>

src/common/CMakeLists.txt

@@ -11,6 +11,8 @@ if(WIN32)
   add_library(dlfcn_win32 STATIC win32/dlfcn.cc win32/errno.cc)
 endif()
 
+add_subdirectory(options)
+
 set(common_srcs
   AsyncOpTracker.cc
   BackTrace.cc

src/common/options.h

@@ -399,5 +399,26 @@ struct Option {
   }
 };
 
-extern const std::vector<Option> ceph_options;
+constexpr unsigned long long operator"" _min (unsigned long long min) {
+  return min * 60;
+}
+constexpr unsigned long long operator"" _hr (unsigned long long hr) {
+  return hr * 60 * 60;
+}
+constexpr unsigned long long operator"" _day (unsigned long long day) {
+  return day * 24 * 60 * 60;
+}
+constexpr unsigned long long operator"" _K (unsigned long long n) {
+  return n << 10;
+}
+constexpr unsigned long long operator"" _M (unsigned long long n) {
+  return n << 20;
+}
+constexpr unsigned long long operator"" _G (unsigned long long n) {
+  return n << 30;
+}
+constexpr unsigned long long operator"" _T (unsigned long long n) {
+  return n << 40;
+}
 
+extern const std::vector<Option> ceph_options;

src/common/options/CMakeLists.txt

@@ -0,0 +1,94 @@
+set(common_options_srcs build_options.cc)
+
+# to mimic the behavior of file(CONFIGURE ...)
+file(GENERATE OUTPUT configure_file.cmake
+  CONTENT "configure_file(\${input_file} \${output_file} @ONLY)")
+function(file_configure input_file output_file)
+  set(cmake_defs
+    -D input_file=${input_file}
+    -D output_file=${output_file})
+  file(STRINGS ${input_file} subvars REGEX "@[^@]+@")
+  foreach(line ${subvars})
+    string(REGEX REPLACE ".*@([^@]+)@.*" "\\1"
+      var "${line}")
+    set(value ${${var}})
+    list(APPEND cmake_defs -D ${var}=${value})
+  endforeach()
+  add_custom_command(OUTPUT ${output_file}
+    COMMAND ${CMAKE_COMMAND} ${cmake_defs} -P configure_file.cmake
+    DEPENDS ${input_file}
+    VERBATIM)
+endfunction()
+
+function(add_options name)
+  set(yaml_in_file ${CMAKE_CURRENT_SOURCE_DIR}/${name}.yaml.in)
+  set(yaml_file ${CMAKE_CURRENT_BINARY_DIR}/${name}.yaml)
+  file_configure("${yaml_in_file}"
+    "${yaml_file}" @ONLY)
+  set(src_file "${name}_options.cc")
+  add_custom_command(
+    OUTPUT ${src_file}
+    COMMAND ${Python_EXECUTABLE} ${CMAKE_CURRENT_SOURCE_DIR}/y2c.py
+      --input ${yaml_file}
+      --output ${src_file}
+      --name ${name}
+    DEPENDS ${yaml_file})
+  list(APPEND common_options_srcs
+    ${src_file})
+  set(common_options_srcs ${common_options_srcs} PARENT_SCOPE)
+endfunction()
+
+set(osd_erasure_code_plugins "jerasure" "lrc")
+if(WITH_EC_ISA_PLUGIN)
+  list(APPEND osd_erasure_code_plugins "isa")
+endif()
+string(REPLACE ";" " " osd_erasure_code_plugins "${osd_erasure_code_plugins}")
+
+set(keyring_paths
+  "/etc/ceph/$cluster.$name.keyring"
+  "/etc/ceph/$cluster.keyring"
+  "/etc/ceph/keyring"
+  "/etc/ceph/keyring.bin")
+if(FREEBSD)
+  list(APPEND keyring_paths
+    "/usr/local/etc/ceph/$cluster.$name.keyring"
+    "/usr/local/etc/ceph/$cluster.keyring"
+    "/usr/local/etc/ceph/keyring"
+    "/usr/local/etc/ceph/keyring.bin")
+endif()
+string(REPLACE ";" "," keyring_paths "${keyring_paths}")
+
+set(ms_bind_retry_count 3)
+set(ms_bind_retry_delay 5)
+if(FREEBSD)
+  # FreeBSD does not use SO_REAUSEADDR so allow for a bit more time per default
+  set(ms_bind_retry_count 6)
+  set(ms_bind_retry_delay 6)
+endif()
+
+set(mgr_disabled_modules "")
+if(WITH_MGR)
+  # https://tracker.ceph.com/issues/45147
+  if(Python3_VERSION VERSION_GREATER_EQUAL 3.8)
+    set(mgr_disabled_modules "diskprediction_local")
+    message(STATUS "mgr module disabled for ${Python3_VERSION}: ${mgr_disabled_modules}")
+  endif()
+endif()
+
+add_options(global)
+add_options(cephfs-mirror)
+add_options(mds)
+add_options(mds-client)
+add_options(rbd)
+add_options(rbd-mirror)
+add_options(immutable-object-cache)
+
+# if set to empty string, system default luarocks package location (if exist) will be used
+set(rgw_luarocks_location "")
+if(WITH_RADOSGW_LUA_PACKAGES)
+  set(rgw_luarocks_location "/tmp/luarocks")
+endif()
+add_options(rgw)
+
+add_library(common-options-objs OBJECT
+  ${common_options_srcs})

src/common/options/build_options.cc

@@ -0,0 +1,35 @@
+// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*-
+// vim: ts=8 sw=2 smarttab
+
+#include "build_options.h"
+
+std::vector<Option> get_global_options();
+std::vector<Option> get_rgw_options();
+std::vector<Option> get_rbd_options();
+std::vector<Option> get_rbd_mirror_options();
+std::vector<Option> get_immutable_object_cache_options();
+std::vector<Option> get_mds_options();
+std::vector<Option> get_mds_client_options();
+std::vector<Option> get_cephfs_mirror_options();
+
+std::vector<Option> build_options()
+{
+  std::vector<Option> result = get_global_options();
+
+  auto ingest = [&result](std::vector<Option>&& options, const char* svc) {
+    for (auto &o : options) {
+      o.add_service(svc);
+      result.push_back(std::move(o));
+    }
+  };
+
+  ingest(get_rgw_options(), "rgw");
+  ingest(get_rbd_options(), "rbd");
+  ingest(get_rbd_mirror_options(), "rbd-mirror");
+  ingest(get_immutable_object_cache_options(), "immutable-objet-cache");
+  ingest(get_mds_options(), "mds");
+  ingest(get_mds_client_options(), "mds_client");
+  ingest(get_cephfs_mirror_options(), "cephfs-mirror");
+
+  return result;
+}

src/common/options/build_options.h

@@ -0,0 +1,8 @@
+// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*-
+// vim: ts=8 sw=2 smarttab
+
+#pragma once
+#include <vector>
+#include "common/options.h"
+
+std::vector<Option> build_options();

src/common/options/cephfs-mirror.yaml.in

@@ -0,0 +1,82 @@
+# -*- mode: YAML -*-
+---
+
+options:
+- name: cephfs_mirror_max_concurrent_directory_syncs
+  type: uint
+  level: advanced
+  desc: maximum number of concurrent snapshot synchronization threads
+  long_desc: maximum number of directory snapshots that can be synchronized concurrently
+    by cephfs-mirror daemon. Controls the number of synchronization threads.
+  default: 3
+  services:
+  - cephfs-mirror
+  min: 1
+- name: cephfs_mirror_action_update_interval
+  type: secs
+  level: advanced
+  desc: interval for driving asynchornous mirror actions
+  long_desc: Interval in seconds to process pending mirror update actions.
+  default: 2
+  services:
+  - cephfs-mirror
+  min: 1
+- name: cephfs_mirror_restart_mirror_on_blocklist_interval
+  type: secs
+  level: advanced
+  desc: interval to restart blocklisted instances
+  long_desc: Interval in seconds to restart blocklisted mirror instances. Setting
+    to zero (0) disables restarting blocklisted instances.
+  default: 30
+  services:
+  - cephfs-mirror
+  min: 0
+- name: cephfs_mirror_max_snapshot_sync_per_cycle
+  type: uint
+  level: advanced
+  desc: number of snapshots to mirror in one cycle
+  long_desc: maximum number of snapshots to mirror when a directory is picked up for
+    mirroring by worker threads.
+  default: 3
+  services:
+  - cephfs-mirror
+  min: 1
+- name: cephfs_mirror_directory_scan_interval
+  type: uint
+  level: advanced
+  desc: interval to scan directories to mirror snapshots
+  long_desc: interval in seconds to scan configured directories for snapshot mirroring.
+  default: 10
+  services:
+  - cephfs-mirror
+  min: 1
+- name: cephfs_mirror_max_consecutive_failures_per_directory
+  type: uint
+  level: advanced
+  desc: consecutive failed directory synchronization attempts before marking a directory
+    as "failed"
+  long_desc: number of consecutive snapshot synchronization failues to mark a directory
+    as "failed". failed directories are retried for synchronization less frequently.
+  default: 10
+  services:
+  - cephfs-mirror
+  min: 0
+- name: cephfs_mirror_retry_failed_directories_interval
+  type: uint
+  level: advanced
+  desc: failed directory retry interval for synchronization
+  long_desc: interval in seconds to retry synchronization for failed directories.
+  default: 60
+  services:
+  - cephfs-mirror
+  min: 1
+- name: cephfs_mirror_restart_mirror_on_failure_interval
+  type: secs
+  level: advanced
+  desc: interval to restart failed mirror instances
+  long_desc: Interval in seconds to restart failed mirror instances. Setting to zero
+    (0) disables restarting failed mirror instances.
+  default: 20
+  services:
+  - cephfs-mirror
+  min: 0

src/common/options/immutable-object-cache.yaml.in

@@ -0,0 +1,98 @@
+# -*- mode: YAML -*-
+---
+
+options:
+- name: immutable_object_cache_path
+  type: str
+  level: advanced
+  desc: immutable object cache data dir
+  default: /tmp/ceph_immutable_object_cache
+  services:
+  - immutable-objet-cache
+- name: immutable_object_cache_sock
+  type: str
+  level: advanced
+  desc: immutable object cache domain socket
+  default: /var/run/ceph/immutable_object_cache_sock
+  services:
+  - immutable-objet-cache
+- name: immutable_object_cache_max_size
+  type: size
+  level: advanced
+  desc: max immutable object cache data size
+  default: 1_G
+  services:
+  - immutable-objet-cache
+- name: immutable_object_cache_max_inflight_ops
+  type: uint
+  level: advanced
+  desc: max inflight promoting requests for immutable object cache daemon
+  default: 128
+  services:
+  - immutable-objet-cache
+- name: immutable_object_cache_client_dedicated_thread_num
+  type: uint
+  level: advanced
+  desc: immutable object cache client dedicated thread number
+  default: 2
+  services:
+  - immutable-objet-cache
+- name: immutable_object_cache_watermark
+  type: float
+  level: advanced
+  desc: immutable object cache water mark
+  default: 0.1
+  services:
+  - immutable-objet-cache
+- name: immutable_object_cache_qos_schedule_tick_min
+  type: millisecs
+  level: advanced
+  desc: minimum schedule tick for immutable object cache
+  default: 50
+  services:
+  - immutable-objet-cache
+  min: 1
+- name: immutable_object_cache_qos_iops_limit
+  type: uint
+  level: advanced
+  desc: the desired immutable object cache IO operations limit per second
+  default: 0
+  services:
+  - immutable-objet-cache
+- name: immutable_object_cache_qos_iops_burst
+  type: uint
+  level: advanced
+  desc: the desired burst limit of immutable object cache IO operations
+  default: 0
+  services:
+  - immutable-objet-cache
+- name: immutable_object_cache_qos_iops_burst_seconds
+  type: secs
+  level: advanced
+  desc: the desired burst duration in seconds of immutable object cache IO operations
+  default: 1
+  services:
+  - immutable-objet-cache
+  min: 1
+- name: immutable_object_cache_qos_bps_limit
+  type: uint
+  level: advanced
+  desc: the desired immutable object cache IO bytes limit per second
+  default: 0
+  services:
+  - immutable-objet-cache
+- name: immutable_object_cache_qos_bps_burst
+  type: uint
+  level: advanced
+  desc: the desired burst limit of immutable object cache IO bytes
+  default: 0
+  services:
+  - immutable-objet-cache
+- name: immutable_object_cache_qos_bps_burst_seconds
+  type: secs
+  level: advanced
+  desc: the desired burst duration in seconds of immutable object cache IO bytes
+  default: 1
+  services:
+  - immutable-objet-cache
+  min: 1

src/common/options/mds-client.yaml.in

@@ -0,0 +1,454 @@
+# -*- mode: YAML -*-
+---
+
+options:
+- name: client_cache_size
+  type: size
+  level: basic
+  desc: soft maximum number of directory entries in client cache
+  default: 16_K
+  services:
+  - mds_client
+- name: client_cache_mid
+  type: float
+  level: advanced
+  desc: mid-point of client cache LRU
+  default: 0.75
+  services:
+  - mds_client
+- name: client_use_random_mds
+  type: bool
+  level: dev
+  desc: issue new requests to a random active MDS
+  default: false
+  services:
+  - mds_client
+- name: client_mount_timeout
+  type: float
+  level: advanced
+  desc: timeout for mounting CephFS (seconds)
+  default: 5_min
+  services:
+  - mds_client
+- name: client_tick_interval
+  type: secs
+  level: dev
+  desc: seconds between client upkeep ticks
+  default: 1
+  services:
+  - mds_client
+- name: client_trace
+  type: str
+  level: dev
+  desc: file containing trace of client operations
+  services:
+  - mds_client
+- name: client_readahead_min
+  type: size
+  level: advanced
+  desc: minimum bytes to readahead in a file
+  default: 128_K
+  services:
+  - mds_client
+- name: client_readahead_max_bytes
+  type: size
+  level: advanced
+  desc: maximum bytes to readahead in a file (zero is unlimited)
+  default: 0
+  services:
+  - mds_client
+- name: client_readahead_max_periods
+  type: int
+  level: advanced
+  desc: maximum stripe periods to readahead in a file
+  default: 4
+  services:
+  - mds_client
+- name: client_reconnect_stale
+  type: bool
+  level: advanced
+  desc: reconnect when the session becomes stale
+  default: false
+  services:
+  - mds_client
+- name: client_snapdir
+  type: str
+  level: advanced
+  desc: pseudo directory for snapshot access to a directory
+  default: .snap
+  services:
+  - mds_client
+- name: client_mountpoint
+  type: str
+  level: advanced
+  desc: default mount-point
+  default: /
+  services:
+  - mds_client
+- name: client_mount_uid
+  type: int
+  level: advanced
+  desc: uid to mount as
+  default: -1
+  services:
+  - mds_client
+- name: client_mount_gid
+  type: int
+  level: advanced
+  desc: gid to mount as
+  default: -1
+  services:
+  - mds_client
+- name: client_notify_timeout
+  type: int
+  level: dev
+  default: 10
+  services:
+  - mds_client
+- name: osd_client_watch_timeout
+  type: int
+  level: dev
+  default: 30
+  services:
+  - mds_client
+- name: client_caps_release_delay
+  type: int
+  level: dev
+  default: 5
+  services:
+  - mds_client
+- name: client_quota_df
+  type: bool
+  level: advanced
+  desc: show quota usage for statfs (df)
+  default: true
+  services:
+  - mds_client
+- name: client_oc
+  type: bool
+  level: advanced
+  desc: enable object caching
+  default: true
+  services:
+  - mds_client
+- name: client_oc_size
+  type: size
+  level: advanced
+  desc: maximum size of object cache
+  default: 200_M
+  services:
+  - mds_client
+  flags:
+  - runtime
+- name: client_oc_max_dirty
+  type: size
+  level: advanced
+  desc: maximum size of dirty pages in object cache
+  default: 100_M
+  services:
+  - mds_client
+  flags:
+  - runtime
+- name: client_oc_target_dirty
+  type: size
+  level: advanced
+  desc: target size of dirty pages object cache
+  default: 8_M
+  services:
+  - mds_client
+  flags:
+  - runtime
+- name: client_oc_max_dirty_age
+  type: float
+  level: advanced
+  desc: maximum age of dirty pages in object cache (seconds)
+  default: 5
+  services:
+  - mds_client
+  flags:
+  - runtime
+- name: client_oc_max_objects
+  type: int
+  level: advanced
+  desc: maximum number of objects in cache
+  default: 1000
+  services:
+  - mds_client
+  flags:
+  - runtime
+- name: client_debug_getattr_caps
+  type: bool
+  level: dev
+  default: false
+  services:
+  - mds_client
+- name: client_debug_force_sync_read
+  type: bool
+  level: dev
+  default: false
+  services:
+  - mds_client
+- name: client_debug_inject_tick_delay
+  type: secs
+  level: dev
+  default: 0
+  services:
+  - mds_client
+- name: client_max_inline_size
+  type: size
+  level: dev
+  default: 4_K
+  services:
+  - mds_client
+- name: client_inject_release_failure
+  type: bool
+  level: dev
+  default: false
+  services:
+  - mds_client
+- name: client_inject_fixed_oldest_tid
+  type: bool
+  level: dev
+  default: false
+  services:
+  - mds_client
+- name: client_metadata
+  type: str
+  level: advanced
+  desc: metadata key=value comma-delimited pairs appended to session metadata
+  services:
+  - mds_client
+- name: client_acl_type
+  type: str
+  level: advanced
+  desc: ACL type to enforce (none or "posix_acl")
+  services:
+  - mds_client
+- name: client_permissions
+  type: bool
+  level: advanced
+  desc: client-enforced permission checking
+  default: true
+  services:
+  - mds_client
+- name: client_dirsize_rbytes
+  type: bool
+  level: advanced
+  desc: set the directory size as the number of file bytes recursively used
+  long_desc: This option enables a CephFS feature that stores the recursive directory
+    size (the bytes used by files in the directory and its descendents) in the st_size
+    field of the stat structure.
+  default: true
+  services:
+  - mds_client
+- name: client_force_lazyio
+  type: bool
+  level: advanced
+  default: false
+  services:
+  - mds_client
+- name: fuse_use_invalidate_cb
+  type: bool
+  level: advanced
+  desc: use fuse 2.8+ invalidate callback to keep page cache consistent
+  default: true
+  services:
+  - mds_client
+- name: fuse_disable_pagecache
+  type: bool
+  level: advanced
+  desc: disable page caching in the kernel for this FUSE mount
+  default: false
+  services:
+  - mds_client
+- name: fuse_allow_other
+  type: bool
+  level: advanced
+  desc: pass allow_other to FUSE on mount
+  default: true
+  services:
+  - mds_client
+- name: fuse_default_permissions
+  type: bool
+  level: advanced
+  desc: pass default_permisions to FUSE on mount
+  default: false
+  services:
+  - mds_client
+  flags:
+  - startup
+- name: fuse_splice_read
+  type: bool
+  level: advanced
+  desc: enable splice read to reduce the memory copies
+  default: true
+  services:
+  - mds_client
+- name: fuse_splice_write
+  type: bool
+  level: advanced
+  desc: enable splice write to reduce the memory copies
+  default: true
+  services:
+  - mds_client
+- name: fuse_splice_move
+  type: bool
+  level: advanced
+  desc: enable splice move to reduce the memory copies
+  default: true
+  services:
+  - mds_client
+- name: fuse_big_writes
+  type: bool
+  level: advanced
+  desc: big_writes is deprecated in libfuse 3.0.0
+  default: true
+  services:
+  - mds_client
+- name: fuse_max_write
+  type: size
+  level: advanced
+  desc: set the maximum number of bytes in a single write operation
+  long_desc: Set the maximum number of bytes in a single write operation that may
+    pass atomically through FUSE. The FUSE default is 128kB and may be indicated by
+    setting this option to 0.
+  default: 0
+  services:
+  - mds_client
+- name: fuse_atomic_o_trunc
+  type: bool
+  level: advanced
+  desc: pass atomic_o_trunc flag to FUSE on mount
+  default: true
+  services:
+  - mds_client
+- name: fuse_debug
+  type: bool
+  level: advanced
+  desc: enable debugging for the libfuse
+  default: false
+  services:
+  - mds_client
+  flags:
+  - no_mon_update
+  - startup
+- name: fuse_multithreaded
+  type: bool
+  level: advanced
+  desc: allow parallel processing through FUSE library
+  default: true
+  services:
+  - mds_client
+- name: fuse_require_active_mds
+  type: bool
+  level: advanced
+  desc: require active MDSs in the file system when mounting
+  default: true
+  services:
+  - mds_client
+- name: fuse_syncfs_on_mksnap
+  type: bool
+  level: advanced
+  desc: synchronize all local metadata/file changes after snapshot
+  default: true
+  services:
+  - mds_client
+- name: fuse_set_user_groups
+  type: bool
+  level: advanced
+  desc: check for ceph-fuse to consider supplementary groups for permissions
+  default: true
+  services:
+  - mds_client
+- name: client_try_dentry_invalidate
+  type: bool
+  level: dev
+  default: false
+  services:
+  - mds_client
+- name: client_die_on_failed_remount
+  type: bool
+  level: dev
+  default: false
+  services:
+  - mds_client
+- name: client_die_on_failed_dentry_invalidate
+  type: bool
+  level: advanced
+  desc: kill the client when no dentry invalidation options are available
+  long_desc: The CephFS client requires a mechanism to invalidate dentries in the
+    caller (e.g. the kernel for ceph-fuse) when capabilities must be recalled. If
+    the client cannot do this then the MDS cache cannot shrink which can cause the
+    MDS to fail.
+  default: true
+  services:
+  - mds_client
+- name: client_check_pool_perm
+  type: bool
+  level: advanced
+  desc: confirm access to inode's data pool/namespace described in file layout
+  default: true
+  services:
+  - mds_client
+- name: client_use_faked_inos
+  type: bool
+  level: dev
+  default: false
+  services:
+  - mds_client
+- name: client_fs
+  type: str
+  level: advanced
+  desc: CephFS file system name to mount
+  long_desc: Use this with ceph-fuse, or with any process that uses libcephfs.  Programs
+    using libcephfs may also pass the filesystem name into mount(), which will override
+    this setting. If no filesystem name is given in mount() or this setting, the default
+    filesystem will be mounted (usually the first created).
+  services:
+  - mds_client
+  flags:
+  - startup
+- name: client_mds_namespace
+  type: str
+  level: dev
+  services:
+  - mds_client
+  flags:
+  - startup
+- name: fake_statfs_for_testing
+  type: int
+  level: dev
+  desc: Set a value for kb and compute kb_used from total of num_bytes
+  default: 0
+  services:
+  - mds_client
+- name: debug_allow_any_pool_priority
+  type: bool
+  level: dev
+  desc: Allow any pool priority to be set to test conversion to new range
+  default: false
+  services:
+  - mds_client
+- name: client_asio_thread_count
+  type: uint
+  level: advanced
+  desc: Size of thread pool for ASIO completions
+  default: 2
+  tags:
+  - client
+  services:
+  - mds_client
+  min: 1
+- name: client_shutdown_timeout
+  type: secs
+  level: advanced
+  desc: timeout for shutting down CephFS
+  long_desc: Timeout for shutting down CephFS via unmount or shutdown.
+  default: 30
+  tags:
+  - client
+  services:
+  - mds_client
+  min: 0
+  flags:
+  - runtime

src/common/options/rbd-mirror.yaml.in

@@ -0,0 +1,210 @@
+# -*- mode: YAML -*-
+---
+
+options:
+- name: rbd_mirror_journal_commit_age
+  type: float
+  level: advanced
+  desc: commit time interval, seconds
+  default: 5
+  services:
+  - rbd-mirror
+- name: rbd_mirror_journal_poll_age
+  type: float
+  level: advanced
+  desc: maximum age (in seconds) between successive journal polls
+  default: 5
+  services:
+  - rbd-mirror
+- name: rbd_mirror_sync_point_update_age
+  type: float
+  level: advanced
+  desc: number of seconds between each update of the image sync point object number
+  default: 30
+  services:
+  - rbd-mirror
+- name: rbd_mirror_concurrent_image_syncs
+  type: uint
+  level: advanced
+  desc: maximum number of image syncs in parallel
+  default: 5
+  services:
+  - rbd-mirror
+- name: rbd_mirror_pool_replayers_refresh_interval
+  type: uint
+  level: advanced
+  desc: interval to refresh peers in rbd-mirror daemon
+  default: 30
+  services:
+  - rbd-mirror
+- name: rbd_mirror_concurrent_image_deletions
+  type: uint
+  level: advanced
+  desc: maximum number of image deletions in parallel
+  default: 1
+  services:
+  - rbd-mirror
+  min: 1
+- name: rbd_mirror_delete_retry_interval
+  type: float
+  level: advanced
+  desc: interval to check and retry the failed deletion requests
+  default: 30
+  services:
+  - rbd-mirror
+- name: rbd_mirror_image_state_check_interval
+  type: uint
+  level: advanced
+  desc: interval to get images from pool watcher and set sources in replayer
+  default: 30
+  services:
+  - rbd-mirror
+  min: 1
+- name: rbd_mirror_leader_heartbeat_interval
+  type: uint
+  level: advanced
+  desc: interval (in seconds) between mirror leader heartbeats
+  default: 5
+  services:
+  - rbd-mirror
+  min: 1
+- name: rbd_mirror_leader_max_missed_heartbeats
+  type: uint
+  level: advanced
+  desc: number of missed heartbeats for non-lock owner to attempt to acquire lock
+  default: 2
+  services:
+  - rbd-mirror
+- name: rbd_mirror_leader_max_acquire_attempts_before_break
+  type: uint
+  level: advanced
+  desc: number of failed attempts to acquire lock after missing heartbeats before
+    breaking lock
+  default: 3
+  services:
+  - rbd-mirror
+- name: rbd_mirror_image_policy_type
+  type: str
+  level: advanced
+  desc: active/active policy type for mapping images to instances
+  default: simple
+  services:
+  - rbd-mirror
+  enum_values:
+  - none
+  - simple
+- name: rbd_mirror_image_policy_migration_throttle
+  type: uint
+  level: advanced
+  desc: number of seconds after which an image can be reshuffled (migrated) again
+  default: 300
+  services:
+  - rbd-mirror
+- name: rbd_mirror_image_policy_update_throttle_interval
+  type: float
+  level: advanced
+  desc: interval (in seconds) to throttle images for mirror daemon peer updates
+  default: 1
+  services:
+  - rbd-mirror
+  min: 1
+- name: rbd_mirror_image_policy_rebalance_timeout
+  type: float
+  level: advanced
+  desc: number of seconds policy should be idle before trigerring reshuffle (rebalance)
+    of images
+  default: 0
+  services:
+  - rbd-mirror
+- name: rbd_mirror_perf_stats_prio
+  type: int
+  level: advanced
+  desc: Priority level for mirror daemon replication perf counters
+  long_desc: The daemon will send perf counter data to the manager daemon if the priority
+    is not lower than mgr_stats_threshold.
+  default: 5
+  services:
+  - rbd-mirror
+  min: 0
+  max: 11
+- name: rbd_mirror_image_perf_stats_prio
+  type: int
+  level: advanced
+  desc: Priority level for mirror daemon per-image replication perf counters
+  long_desc: The daemon will send per-image perf counter data to the manager daemon
+    if the priority is not lower than mgr_stats_threshold.
+  default: 5
+  services:
+  - rbd-mirror
+  min: 0
+  max: 11
+- name: rbd_mirror_memory_autotune
+  type: bool
+  level: dev
+  desc: Automatically tune the ratio of caches while respecting min values.
+  default: true
+  services:
+  - rbd-mirror
+  see_also:
+  - rbd_mirror_memory_target
+- name: rbd_mirror_memory_target
+  type: size
+  level: basic
+  desc: When tcmalloc and cache autotuning is enabled, try to keep this many bytes
+    mapped in memory.
+  default: 4_G
+  services:
+  - rbd-mirror
+  see_also:
+  - rbd_mirror_memory_autotune
+- name: rbd_mirror_memory_base
+  type: size
+  level: dev
+  desc: When tcmalloc and cache autotuning is enabled, estimate the minimum amount
+    of memory in bytes the rbd-mirror daemon will need.
+  default: 768_M
+  services:
+  - rbd-mirror
+  see_also:
+  - rbd_mirror_memory_autotune
+- name: rbd_mirror_memory_expected_fragmentation
+  type: float
+  level: dev
+  desc: When tcmalloc and cache autotuning is enabled, estimate the percent of memory
+    fragmentation.
+  default: 0.15
+  services:
+  - rbd-mirror
+  see_also:
+  - rbd_mirror_memory_autotune
+  min: 0
+  max: 1
+- name: rbd_mirror_memory_cache_min
+  type: size
+  level: dev
+  desc: When tcmalloc and cache autotuning is enabled, set the minimum amount of memory
+    used for cache.
+  default: 128_M
+  services:
+  - rbd-mirror
+  see_also:
+  - rbd_mirror_memory_autotune
+- name: rbd_mirror_memory_cache_resize_interval
+  type: float
+  level: dev
+  desc: When tcmalloc and cache autotuning is enabled, wait this many seconds between
+    resizing caches.
+  default: 5
+  services:
+  - rbd-mirror
+  see_also:
+  - rbd_mirror_memory_autotune
+- name: rbd_mirror_memory_cache_autotune_interval
+  type: float
+  level: dev
+  desc: The number of seconds to wait between rebalances when cache autotune is enabled.
+  default: 30
+  services:
+  - rbd-mirror
+  see_also:
+  - rbd_mirror_memory_autotune

src/common/options/rbd.yaml.in

@@ -0,0 +1,832 @@
+# -*- mode: YAML -*-
+---
+
+headers: |
+  #include <regex>
+  // rbd feature and io operation validation
+  #include "include/stringify.h"
+  #include "common/strtol.h"
+  #include "librbd/Features.h"
+  #include "librbd/io/IoOperations.h"
+options:
+- name: rbd_default_pool
+  type: str
+  level: advanced
+  desc: default pool for storing new images
+  default: rbd
+  services:
+  - rbd
+  validator: |
+    [](std::string *value, std::string *error_message) {
+      std::regex pattern("^[^@/]+$");
+      if (!std::regex_match (*value, pattern)) {
+        *value = "rbd";
+        *error_message = "invalid RBD default pool, resetting to 'rbd'";
+      }
+      return 0;
+    }
+- name: rbd_default_data_pool
+  type: str
+  level: advanced
+  desc: default pool for storing data blocks for new images
+  services:
+  - rbd
+  validator: |
+    [](std::string *value, std::string *error_message) {
+      std::regex pattern("^[^@/]*$");
+      if (!std::regex_match (*value, pattern)) {
+        *value = "";
+        *error_message = "ignoring invalid RBD data pool";
+      }
+      return 0;
+    }
+- name: rbd_default_features
+  type: str
+  level: advanced
+  desc: default v2 image features for new images
+  long_desc: 'RBD features are only applicable for v2 images. This setting accepts
+    either an integer bitmask value or comma-delimited string of RBD feature names.
+    This setting is always internally stored as an integer bitmask value. The mapping
+    between feature bitmask value and feature name is as follows: +1 -> layering,
+    +2 -> striping, +4 -> exclusive-lock, +8 -> object-map, +16 -> fast-diff, +32
+    -> deep-flatten, +64 -> journaling, +128 -> data-pool'
+  default: layering,exclusive-lock,object-map,fast-diff,deep-flatten
+  services:
+  - rbd
+  flags:
+  - runtime
+  validator: |
+    [](std::string *value, std::string *error_message) {
+      stringstream ss;
+      uint64_t features = librbd::rbd_features_from_string(*value, &ss);
+      // Leave this in integer form to avoid breaking Cinder.  Someday
+      // we would like to present this in string form instead...
+      *value = stringify(features);
+      if (ss.str().size()) {
+        return -EINVAL;
+      }
+      return 0;
+    }
+- name: rbd_op_threads
+  type: uint
+  level: advanced
+  desc: number of threads to utilize for internal processing
+  default: 1
+  services:
+  - rbd
+- name: rbd_op_thread_timeout
+  type: uint
+  level: advanced
+  desc: time in seconds for detecting a hung thread
+  default: 60
+  services:
+  - rbd
+- name: rbd_disable_zero_copy_writes
+  type: bool
+  level: advanced
+  desc: Disable the use of zero-copy writes to ensure unstable writes from clients
+    cannot cause a CRC mismatch
+  default: true
+  services:
+  - rbd
+- name: rbd_non_blocking_aio
+  type: bool
+  level: advanced
+  desc: process AIO ops from a dispatch thread to prevent blocking
+  default: true
+  services:
+  - rbd
+- name: rbd_cache
+  type: bool
+  level: advanced
+  desc: whether to enable caching (writeback unless rbd_cache_max_dirty is 0)
+  default: true
+  services:
+  - rbd
+- name: rbd_cache_policy
+  type: str
+  level: advanced
+  desc: cache policy for handling writes.
+  default: writearound
+  services:
+  - rbd
+  enum_values:
+  - writethrough
+  - writeback
+  - writearound
+- name: rbd_cache_writethrough_until_flush
+  type: bool
+  level: advanced
+  desc: whether to make writeback caching writethrough until flush is called, to be
+    sure the user of librbd will send flushes so that writeback is safe
+  default: true
+  services:
+  - rbd
+- name: rbd_cache_size
+  type: size
+  level: advanced
+  desc: cache size in bytes
+  default: 32_M
+  services:
+  - rbd
+- name: rbd_cache_max_dirty
+  type: size
+  level: advanced
+  desc: dirty limit in bytes - set to 0 for write-through caching
+  default: 24_M
+  services:
+  - rbd
+- name: rbd_cache_target_dirty
+  type: size
+  level: advanced
+  desc: target dirty limit in bytes
+  default: 16_M
+  services:
+  - rbd
+- name: rbd_cache_max_dirty_age
+  type: float
+  level: advanced
+  desc: seconds in cache before writeback starts
+  default: 1
+  services:
+  - rbd
+- name: rbd_cache_max_dirty_object
+  type: uint
+  level: advanced
+  desc: dirty limit for objects - set to 0 for auto calculate from rbd_cache_size
+  default: 0
+  services:
+  - rbd
+- name: rbd_cache_block_writes_upfront
+  type: bool
+  level: advanced
+  desc: whether to block writes to the cache before the aio_write call completes
+  default: false
+  services:
+  - rbd
+- name: rbd_parent_cache_enabled
+  type: bool
+  level: advanced
+  desc: whether to enable rbd shared ro cache
+  default: false
+  services:
+  - rbd
+- name: rbd_concurrent_management_ops
+  type: uint
+  level: advanced
+  desc: how many operations can be in flight for a management operation like deleting
+    or resizing an image
+  default: 10
+  services:
+  - rbd
+  min: 1
+- name: rbd_balance_snap_reads
+  type: bool
+  level: advanced
+  desc: distribute snap read requests to random OSD
+  default: false
+  services:
+  - rbd
+- name: rbd_localize_snap_reads
+  type: bool
+  level: advanced
+  desc: localize snap read requests to closest OSD
+  default: false
+  services:
+  - rbd
+- name: rbd_balance_parent_reads
+  type: bool
+  level: advanced
+  desc: distribute parent read requests to random OSD
+  default: false
+  services:
+  - rbd
+- name: rbd_localize_parent_reads
+  type: bool
+  level: advanced
+  desc: localize parent requests to closest OSD
+  default: false
+  services:
+  - rbd
+- name: rbd_sparse_read_threshold_bytes
+  type: size
+  level: advanced
+  desc: threshold for issuing a sparse-read
+  long_desc: minimum number of sequential bytes to read against an object before issuing
+    a sparse-read request to the cluster. 0 implies it must be a full object read
+    to issue a sparse-read, 1 implies always use sparse-read, and any value larger
+    than the maximum object size will disable sparse-read for all requests
+  default: 64_K
+  services:
+  - rbd
+- name: rbd_readahead_trigger_requests
+  type: uint
+  level: advanced
+  desc: number of sequential requests necessary to trigger readahead
+  default: 10
+  services:
+  - rbd
+- name: rbd_readahead_max_bytes
+  type: size
+  level: advanced
+  desc: set to 0 to disable readahead
+  default: 512_K
+  services:
+  - rbd
+- name: rbd_readahead_disable_after_bytes
+  type: size
+  level: advanced
+  desc: how many bytes are read in total before readahead is disabled
+  default: 50_M
+  services:
+  - rbd
+- name: rbd_clone_copy_on_read
+  type: bool
+  level: advanced
+  desc: copy-up parent image blocks to clone upon read request
+  default: false
+  services:
+  - rbd
+- name: rbd_blocklist_on_break_lock
+  type: bool
+  level: advanced
+  desc: whether to blocklist clients whose lock was broken
+  default: true
+  services:
+  - rbd
+- name: rbd_blocklist_expire_seconds
+  type: uint
+  level: advanced
+  desc: number of seconds to blocklist - set to 0 for OSD default
+  default: 0
+  services:
+  - rbd
+- name: rbd_request_timed_out_seconds
+  type: uint
+  level: advanced
+  desc: number of seconds before maintenance request times out
+  default: 30
+  services:
+  - rbd
+- name: rbd_skip_partial_discard
+  type: bool
+  level: advanced
+  desc: skip discard (zero) of unaligned extents within an object
+  default: true
+  services:
+  - rbd
+- name: rbd_discard_granularity_bytes
+  type: uint
+  level: advanced
+  desc: minimum aligned size of discard operations
+  default: 64_K
+  services:
+  - rbd
+  min: 4_K
+  max: 32_M
+  validator: |
+    [](std::string *value, std::string *error_message) {
+      uint64_t f = strict_si_cast<uint64_t>(value->c_str(), error_message);
+      if (!error_message->empty()) {
+        return -EINVAL;
+      } else if (!isp2(f)) {
+        *error_message = "value must be a power of two";
+        return -EINVAL;
+      }
+      return 0;
+    }
+- name: rbd_enable_alloc_hint
+  type: bool
+  level: advanced
+  desc: when writing a object, it will issue a hint to osd backend to indicate the
+    expected size object need
+  default: true
+  services:
+  - rbd
+- name: rbd_compression_hint
+  type: str
+  level: basic
+  desc: Compression hint to send to the OSDs during writes
+  default: none
+  services:
+  - rbd
+  enum_values:
+  - none
+  - compressible
+  - incompressible
+  flags:
+  - runtime
+- name: rbd_read_from_replica_policy
+  type: str
+  level: basic
+  desc: Read replica policy send to the OSDS during reads
+  default: default
+  services:
+  - rbd
+  enum_values:
+  - default
+  - balance
+  - localize
+  flags:
+  - runtime
+- name: rbd_tracing
+  type: bool
+  level: advanced
+  desc: true if LTTng-UST tracepoints should be enabled
+  default: false
+  services:
+  - rbd
+- name: rbd_blkin_trace_all
+  type: bool
+  level: advanced
+  desc: create a blkin trace for all RBD requests
+  default: false
+  services:
+  - rbd
+- name: rbd_validate_pool
+  type: bool
+  level: advanced
+  desc: validate empty pools for RBD compatibility
+  default: true
+  services:
+  - rbd
+- name: rbd_validate_names
+  type: bool
+  level: advanced
+  desc: validate new image names for RBD compatibility
+  default: true
+  services:
+  - rbd
+- name: rbd_invalidate_object_map_on_timeout
+  type: bool
+  level: dev
+  desc: true if object map should be invalidated when load or update timeout
+  default: true
+  services:
+  - rbd
+- name: rbd_auto_exclusive_lock_until_manual_request
+  type: bool
+  level: advanced
+  desc: automatically acquire/release exclusive lock until it is explicitly requested
+  default: true
+  services:
+  - rbd
+- name: rbd_move_to_trash_on_remove
+  type: bool
+  level: basic
+  desc: automatically move images to the trash when deleted
+  default: false
+  services:
+  - rbd
+- name: rbd_move_to_trash_on_remove_expire_seconds
+  type: uint
+  level: basic
+  desc: default number of seconds to protect deleted images in the trash
+  default: 0
+  services:
+  - rbd
+- name: rbd_move_parent_to_trash_on_remove
+  type: bool
+  level: basic
+  desc: move parent with clone format v2 children to the trash when deleted
+  default: false
+  services:
+  - rbd
+- name: rbd_mirroring_resync_after_disconnect
+  type: bool
+  level: advanced
+  desc: automatically start image resync after mirroring is disconnected due to being
+    laggy
+  default: false
+  services:
+  - rbd
+- name: rbd_mirroring_delete_delay
+  type: uint
+  level: advanced
+  desc: time-delay in seconds for rbd-mirror delete propagation
+  default: 0
+  services:
+  - rbd
+- name: rbd_mirroring_replay_delay
+  type: uint
+  level: advanced
+  desc: time-delay in seconds for rbd-mirror asynchronous replication
+  default: 0
+  services:
+  - rbd
+- name: rbd_mirroring_max_mirroring_snapshots
+  type: uint
+  level: advanced
+  desc: mirroring snapshots limit
+  default: 3
+  services:
+  - rbd
+  min: 3
+- name: rbd_default_format
+  type: uint
+  level: advanced
+  desc: default image format for new images
+  default: 2
+  services:
+  - rbd
+- name: rbd_default_order
+  type: uint
+  level: advanced
+  desc: default order (data block object size) for new images
+  default: 22
+  services:
+  - rbd
+- name: rbd_default_stripe_count
+  type: uint
+  level: advanced
+  desc: default stripe count for new images
+  default: 0
+  services:
+  - rbd
+- name: rbd_default_stripe_unit
+  type: size
+  level: advanced
+  desc: default stripe width for new images
+  default: 0
+  services:
+  - rbd
+- name: rbd_default_map_options
+  type: str
+  level: advanced
+  desc: default krbd map options
+  services:
+  - rbd
+- name: rbd_default_clone_format
+  type: str
+  level: advanced
+  desc: default internal format for handling clones
+  long_desc: This sets the internal format for tracking cloned images. The setting
+    of '1' requires attaching to protected snapshots that cannot be removed until
+    the clone is removed/flattened. The setting of '2' will allow clones to be attached
+    to any snapshot and permits removing in-use parent snapshots but requires Mimic
+    or later clients. The default setting of 'auto' will use the v2 format if the
+    cluster is configured to require mimic or later clients.
+  default: auto
+  services:
+  - rbd
+  enum_values:
+  - '1'
+  - '2'
+  - auto
+  flags:
+  - runtime
+- name: rbd_journal_order
+  type: uint
+  level: advanced
+  desc: default order (object size) for journal data objects
+  default: 24
+  services:
+  - rbd
+  min: 12
+  max: 26
+- name: rbd_journal_splay_width
+  type: uint
+  level: advanced
+  desc: number of active journal objects
+  default: 4
+  services:
+  - rbd
+- name: rbd_journal_commit_age
+  type: float
+  level: advanced
+  desc: commit time interval, seconds
+  default: 5
+  services:
+  - rbd
+- name: rbd_journal_object_writethrough_until_flush
+  type: bool
+  level: advanced
+  desc: when enabled, the rbd_journal_object_flush* configuration options are ignored
+    until the first flush so that batched journal IO is known to be safe for consistency
+  default: true
+  services:
+  - rbd
+- name: rbd_journal_object_flush_interval
+  type: uint
+  level: advanced
+  desc: maximum number of pending commits per journal object
+  default: 0
+  services:
+  - rbd
+- name: rbd_journal_object_flush_bytes
+  type: size
+  level: advanced
+  desc: maximum number of pending bytes per journal object
+  default: 1_M
+  services:
+  - rbd
+- name: rbd_journal_object_flush_age
+  type: float
+  level: advanced
+  desc: maximum age (in seconds) for pending commits
+  default: 0
+  services:
+  - rbd
+- name: rbd_journal_object_max_in_flight_appends
+  type: uint
+  level: advanced
+  desc: maximum number of in-flight appends per journal object
+  default: 0
+  services:
+  - rbd
+- name: rbd_journal_pool
+  type: str
+  level: advanced
+  desc: pool for journal objects
+  services:
+  - rbd
+- name: rbd_journal_max_payload_bytes
+  type: size
+  level: advanced
+  desc: maximum journal payload size before splitting
+  default: 16_K
+  services:
+  - rbd
+- name: rbd_journal_max_concurrent_object_sets
+  type: uint
+  level: advanced
+  desc: maximum number of object sets a journal client can be behind before it is
+    automatically unregistered
+  default: 0
+  services:
+  - rbd
+- name: rbd_qos_iops_limit
+  type: uint
+  level: advanced
+  desc: the desired limit of IO operations per second
+  default: 0
+  services:
+  - rbd
+- name: rbd_qos_bps_limit
+  type: uint
+  level: advanced
+  desc: the desired limit of IO bytes per second
+  default: 0
+  services:
+  - rbd
+- name: rbd_qos_read_iops_limit
+  type: uint
+  level: advanced
+  desc: the desired limit of read operations per second
+  default: 0
+  services:
+  - rbd
+- name: rbd_qos_write_iops_limit
+  type: uint
+  level: advanced
+  desc: the desired limit of write operations per second
+  default: 0
+  services:
+  - rbd
+- name: rbd_qos_read_bps_limit
+  type: uint
+  level: advanced
+  desc: the desired limit of read bytes per second
+  default: 0
+  services:
+  - rbd
+- name: rbd_qos_write_bps_limit
+  type: uint
+  level: advanced
+  desc: the desired limit of write bytes per second
+  default: 0
+  services:
+  - rbd
+- name: rbd_qos_iops_burst
+  type: uint
+  level: advanced
+  desc: the desired burst limit of IO operations
+  default: 0
+  services:
+  - rbd
+- name: rbd_qos_bps_burst
+  type: uint
+  level: advanced
+  desc: the desired burst limit of IO bytes
+  default: 0
+  services:
+  - rbd
+- name: rbd_qos_read_iops_burst
+  type: uint
+  level: advanced
+  desc: the desired burst limit of read operations
+  default: 0
+  services:
+  - rbd
+- name: rbd_qos_write_iops_burst
+  type: uint
+  level: advanced
+  desc: the desired burst limit of write operations
+  default: 0
+  services:
+  - rbd
+- name: rbd_qos_read_bps_burst
+  type: uint
+  level: advanced
+  desc: the desired burst limit of read bytes
+  default: 0
+  services:
+  - rbd
+- name: rbd_qos_write_bps_burst
+  type: uint
+  level: advanced
+  desc: the desired burst limit of write bytes
+  default: 0
+  services:
+  - rbd
+- name: rbd_qos_iops_burst_seconds
+  type: uint
+  level: advanced
+  desc: the desired burst duration in seconds of IO operations
+  default: 1
+  services:
+  - rbd
+  min: 1
+- name: rbd_qos_bps_burst_seconds
+  type: uint
+  level: advanced
+  desc: the desired burst duration in seconds of IO bytes
+  default: 1
+  services:
+  - rbd
+  min: 1
+- name: rbd_qos_read_iops_burst_seconds
+  type: uint
+  level: advanced
+  desc: the desired burst duration in seconds of read operations
+  default: 1
+  services:
+  - rbd
+  min: 1
+- name: rbd_qos_write_iops_burst_seconds
+  type: uint
+  level: advanced
+  desc: the desired burst duration in seconds of write operations
+  default: 1
+  services:
+  - rbd
+  min: 1
+- name: rbd_qos_read_bps_burst_seconds
+  type: uint
+  level: advanced
+  desc: the desired burst duration in seconds of read bytes
+  default: 1
+  services:
+  - rbd
+  min: 1
+- name: rbd_qos_write_bps_burst_seconds
+  type: uint
+  level: advanced
+  desc: the desired burst duration in seconds of write bytes
+  default: 1
+  services:
+  - rbd
+  min: 1
+- name: rbd_qos_schedule_tick_min
+  type: uint
+  level: advanced
+  desc: minimum schedule tick (in milliseconds) for QoS
+  default: 50
+  services:
+  - rbd
+  min: 1
+- name: rbd_qos_exclude_ops
+  type: str
+  level: advanced
+  desc: optionally exclude ops from QoS
+  long_desc: 'Optionally exclude ops from QoS. This setting accepts either an integer
+    bitmask value or comma-delimited string of op names. This setting is always internally
+    stored as an integer bitmask value. The mapping between op bitmask value and op
+    name is as follows: +1 -> read, +2 -> write, +4 -> discard, +8 -> write_same,
+    +16 -> compare_and_write'
+  services:
+  - rbd
+  flags:
+  - runtime
+  validator: |
+    [](std::string *value, std::string *error_message) {
+        ostringstream ss;
+        uint64_t exclude_ops = librbd::io::rbd_io_operations_from_string(*value, &ss);
+        // Leave this in integer form to avoid breaking Cinder.  Someday
+        // we would like to present this in string form instead...
+        *value = stringify(exclude_ops);
+        if (ss.str().size()) {
+          return -EINVAL;
+        }
+        return 0;
+    }
+- name: rbd_discard_on_zeroed_write_same
+  type: bool
+  level: advanced
+  desc: discard data on zeroed write same instead of writing zero
+  default: true
+  services:
+  - rbd
+- name: rbd_mtime_update_interval
+  type: uint
+  level: advanced
+  desc: RBD Image modify timestamp refresh interval. Set to 0 to disable modify timestamp
+    update.
+  default: 60
+  services:
+  - rbd
+  min: 0
+- name: rbd_atime_update_interval
+  type: uint
+  level: advanced
+  desc: RBD Image access timestamp refresh interval. Set to 0 to disable access timestamp
+    update.
+  default: 60
+  services:
+  - rbd
+  min: 0
+- name: rbd_io_scheduler
+  type: str
+  level: advanced
+  desc: RBD IO scheduler
+  default: simple
+  services:
+  - rbd
+  enum_values:
+  - none
+  - simple
+- name: rbd_io_scheduler_simple_max_delay
+  type: uint
+  level: advanced
+  desc: maximum io delay (in milliseconds) for simple io scheduler (if set to 0 dalay
+    is calculated based on latency stats)
+  default: 0
+  services:
+  - rbd
+  min: 0
+- name: rbd_persistent_cache_mode
+  type: str
+  level: advanced
+  desc: enable persistent write back cache for this volume
+  default: disabled
+  services:
+  - rbd
+  enum_values:
+  - disabled
+  - rwl
+  - ssd
+- name: rbd_persistent_cache_log_periodic_stats
+  type: bool
+  level: advanced
+  desc: emit periodic perf stats to debug log
+  default: false
+  services:
+  - rbd
+- name: rbd_persistent_cache_size
+  type: uint
+  level: advanced
+  desc: size of the persistent write back cache for this volume
+  default: 1_G
+  services:
+  - rbd
+  min: 1_G
+- name: rbd_persistent_cache_path
+  type: str
+  level: advanced
+  desc: location of the persistent write back cache in a DAX-enabled filesystem on
+    persistent memory
+  default: /tmp
+  services:
+  - rbd
+- name: rbd_quiesce_notification_attempts
+  type: uint
+  level: dev
+  desc: the number of quiesce notification attempts
+  default: 10
+  services:
+  - rbd
+  min: 1
+- name: rbd_default_snapshot_quiesce_mode
+  type: str
+  level: advanced
+  desc: default snapshot quiesce mode
+  default: required
+  services:
+  - rbd
+  enum_values:
+  - required
+  - ignore-error
+  - skip
+- name: rbd_plugins
+  type: str
+  level: advanced
+  desc: comma-delimited list of librbd plugins to enable
+  services:
+  - rbd
+- name: rbd_config_pool_override_update_timestamp
+  type: uint
+  level: dev
+  desc: timestamp of last update to pool-level config overrides
+  default: 0
+  services:
+  - rbd

src/common/options/y2c.py

@@ -0,0 +1,323 @@
+#!/usr/bin/python3
+
+import yaml
+import argparse
+import math
+import os
+import sys
+
+
+def type_to_cxx(t):
+    return f'Option::TYPE_{t.upper()}'
+
+
+def level_to_cxx(lv):
+    return f'Option::LEVEL_{lv.upper()}'
+
+
+def eval_str(v):
+    if v == "":
+        return v
+    v = v.strip('"').replace('"', '\\"')
+    return f'"{v}"'
+
+
+def eval_value(v, typ):
+    try:
+        if typ == 'str':
+            return eval_str(v)
+        if typ == 'float':
+            return float(v)
+        if typ in ('uint', 'int', 'size', 'secs', 'millisecs'):
+            return int(v)
+        if typ == 'bool':
+            return 'true' if v else 'false'
+        else:
+            return f'"{v}"'
+    except ValueError:
+        times = dict(_min=60,
+                     _hr=60*60,
+                     _day=24*60*60,
+                     _K=1 << 10,
+                     _M=1 << 20,
+                     _G=1 << 30,
+                     _T=1 << 40)
+        for unit, m in times.items():
+            if v.endswith(unit):
+                int(v[:-len(unit)])
+                # user defined literals
+                return v
+        raise ValueError(f'unknown value: {v}')
+
+
+def set_default(default, typ):
+    if default is None:
+        return ''
+    v = eval_value(default, typ)
+    return f'.set_default({v})\n'
+
+
+def set_daemon_default(default, typ):
+    if default is None:
+        return ''
+    v = eval_value(default, typ)
+    return f'.set_daemon_default({v})\n'
+
+
+def add_tags(tags):
+    if tags is None:
+        return ''
+    cxx = ''
+    for tag in tags:
+        v = eval_str(tag)
+        cxx += f'.add_tag({v})\n'
+    return cxx
+
+
+def add_services(services):
+    if services is None:
+        return ''
+    if len(services) == 1:
+        return f'.add_service("{services[0]}")\n'
+    else:
+        param = ', '.join(f'"{s}"' for s in services)
+        return f'.add_service({{{param}}})\n'
+
+
+def add_see_also(see_also):
+    if see_also is None:
+        return ''
+    param = ', '.join(f'"{v}"' for v in see_also)
+    return f'.add_see_also({{{param}}})\n'
+
+
+def set_desc(desc):
+    if desc is None:
+        return ''
+    v = eval_str(desc)
+    return f'.set_description({v})\n'
+
+
+def set_long_desc(desc):
+    if desc is None:
+        return ''
+    v = eval_str(desc)
+    return f'.set_long_description({v})\n'
+
+
+def set_min_max(mi, ma, typ):
+    if mi is None and ma is None:
+        return ''
+    if mi is not None and ma is not None:
+        min_v = eval_value(mi, typ)
+        max_v = eval_value(ma, typ)
+        if isinstance(min_v, str) and isinstance(max_v, int):
+            return f'.set_min_max({min_v}, {max_v}ULL)\n'
+        elif isinstance(min_v, int) and isinstance(max_v, str):
+            return f'.set_min_max({min_v}ULL, {max_v})\n'
+        else:
+            return f'.set_min_max({min_v}, {max_v})\n'
+    if mi is not None:
+        min_v = eval_value(mi, typ)
+        return f'.set_min({min_v})\n'
+    raise ValueError('set_max() is not implemented')
+
+
+def set_enum_allowed(values):
+    if values is None:
+        return ''
+    param = ', '.join(f'"{v}"' for v in values)
+    return f'.set_enum_allowed({{{param}}})'
+
+
+def add_flags(flags):
+    if flags is None:
+        return ''
+    cxx = ''
+    for flag in flags:
+        cxx += f'.set_flag(Option::FLAG_{flag.upper()})\n'
+    return cxx
+
+
+def set_validator(validator):
+    if validator is None:
+        return ''
+    validator = validator.rstrip()
+    return f'.set_validator({validator})\n'
+
+
+def add_verbatim(verbatim):
+    if verbatim is None:
+        return ''
+    return verbatim + '\n'
+
+
+def yaml_to_cxx(opt, indent):
+    name = opt['name']
+    typ = opt['type']
+    ctyp = type_to_cxx(typ)
+    level = level_to_cxx(opt['level'])
+    cxx = f'Option("{name}", {ctyp}, {level})\n'
+    cxx += set_desc(opt.get('desc'))
+    cxx += set_long_desc(opt.get('long_desc'))
+    cxx += set_default(opt.get('default'), typ)
+    cxx += set_daemon_default(opt.get('daemon_default'), typ)
+    cxx += set_min_max(opt.get('min'), opt.get('max'), typ)
+    cxx += set_enum_allowed(opt.get('enum_values'))
+    cxx += set_validator(opt.get('validator'))
+    cxx += add_flags(opt.get('flags'))
+    cxx += add_services(opt.get('services'))
+    cxx += add_tags(opt.get('tags'))
+    cxx += add_see_also(opt.get('see_also'))
+    verbatim = add_verbatim(opt.get('verbatim'))
+    cxx += verbatim
+    if verbatim:
+        cxx += '\n'
+    else:
+        cxx = cxx.rstrip()
+    cxx += ',\n'
+    if indent > 0:
+        indented = []
+        for line in cxx.split('\n'):
+            if line:
+                indented.append(' ' * indent + line + '\n')
+        cxx = ''.join(indented)
+    return cxx
+
+
+TEMPLATE_CC = '''#include "common/options.h"
+{headers}
+
+std::vector<Option> get_{name}_options() {{
+  return std::vector<Option>({{
+@body@
+  }});
+}}
+'''
+
+
+def translate(opts):
+    if opts.raw:
+        prelude, epilogue = '', ''
+    else:
+        prelude, epilogue = TEMPLATE_CC.split('@body@')
+
+    if opts.name:
+        name = opts.name
+    else:
+        name = os.path.split(opts.input)[-1]
+        name = name.rsplit('.', 1)[0]
+    name = name.replace('-', '_')
+    with open(opts.input) as infile, open(opts.output, 'w') as outfile:
+        yml = yaml.safe_load(infile)
+        headers = yml.get('headers', '')
+        outfile.write(prelude.format(name=name, headers=headers))
+        options = yml['options']
+        for option in options:
+            try:
+                cxx = yaml_to_cxx(option, opts.indent)
+                outfile.write(cxx)
+                outfile.write('\n')
+            except ValueError as e:
+                print(f'failed to translate option "{name}": {e}',
+                      file=sys.stderr)
+                return 1
+        outfile.write(epilogue.replace("}}", "}"))
+
+
+def readable_size(value, typ):
+    times = dict(T=1 << 40,
+                 G=1 << 30,
+                 M=1 << 20,
+                 K=1 << 10)
+    if isinstance(value, str):
+        value = value.strip('"')
+    try:
+        v = int(value)
+        if v == 0:
+            return 0
+        for unit, m in times.items():
+            if v % m == 0:
+                v = int(v / m)
+                return f'{v}_{unit}'
+        return v
+    except ValueError:
+        return value
+
+
+def readable_duration(value, typ):
+    times = dict(day=24*60*60,
+                 hr=60*60,
+                 min=60)
+    if isinstance(value, str):
+        value = value.strip('"')
+    try:
+        v = float(value)
+        if math.floor(v) != v:
+            return v
+        v = int(v)
+        if v == 0:
+            return 0
+        for unit, m in times.items():
+            if v % m == 0:
+                v = int(v / m)
+                return f'{v}_{unit}'
+        return v
+    except ValueError:
+        return value
+
+
+def readable_millisecs(value, typ):
+    return int(value)
+
+
+def readable(opts):
+    with open(opts.input) as infile, open(opts.output, 'w') as outfile:
+        yml = yaml.safe_load(infile)
+        options = yml['options']
+        for option in options:
+            typ = option['type']
+            if typ in ('size', 'uint'):
+                do_readable = readable_size
+            elif typ in ('float', 'int', 'secs'):
+                do_readable = readable_duration
+            elif typ == 'millisecs':
+                do_readable = readable_millisecs
+            else:
+                continue
+            for field in ['default', 'min', 'max', 'daemon_default']:
+                v = option.get(field)
+                if v is not None:
+                    option[field] = do_readable(v, typ)
+        yml['options'] = options
+        yaml.dump(yml, outfile, sort_keys=False, indent=2)
+
+
+def main():
+    parser = argparse.ArgumentParser(
+        formatter_class=argparse.ArgumentDefaultsHelpFormatter)
+    parser.add_argument('-i', '--input', dest='input',
+                        default='options.yaml',
+                        help='the YAML file to be processed')
+    parser.add_argument('-o', '--output', dest='output',
+                        default='options',
+                        help='the path to the generated .cc file')
+    parser.add_argument('--indent', type=int,
+                        default=4,
+                        help='the number of spaces added before each line')
+    parser.add_argument('--name',
+                        help='the name of the option group')
+    parser.add_argument('--raw', action='store_true',
+                        help='output the array without the full function')
+    parser.add_argument('--op', choices=('readable', 'translate'),
+                        default='translate',
+                        help='operation to perform.')
+    opts = parser.parse_args(sys.argv[1:])
+    if opts.op == 'translate':
+        translate(opts)
+    else:
+        readable(opts)
+
+
+if __name__ == '__main__':
+    main()

src/crimson/CMakeLists.txt

@@ -118,7 +118,8 @@ add_library(crimson-common STATIC
   ${PROJECT_SOURCE_DIR}/src/osd/PGPeeringEvent.cc
   ${PROJECT_SOURCE_DIR}/src/xxHash/xxhash.c
   ${crimson_common_srcs}
-  $<TARGET_OBJECTS:common_mountcephfs_objs>)
+  $<TARGET_OBJECTS:common_mountcephfs_objs>
+  $<TARGET_OBJECTS:common-options-objs>)
 
 target_compile_definitions(crimson-common PRIVATE
   "CMAKE_INSTALL_LIBDIR=\"${CMAKE_INSTALL_LIBDIR}\""

src/include/config-h.in.cmake

@@ -324,9 +324,6 @@
 
 #cmakedefine MGR_PYTHON_EXECUTABLE "@MGR_PYTHON_EXECUTABLE@"
 
-/* the default value of "mgr_disabled_module" option */
-#cmakedefine MGR_DISABLED_MODULES "@MGR_DISABLED_MODULES@"
-
 /* Define to 1 if you have the `getprogname' function. */
 #cmakedefine HAVE_GETPROGNAME 1
 

win32_deps_build.sh

@@ -68,14 +68,16 @@ mkdir -p $depsSrcDir
 case "$OS" in
     ubuntu)
         sudo apt-get update
-        sudo apt-get -y install mingw-w64 cmake pkg-config python3-dev python3-pip \
+        sudo apt-get -y install mingw-w64 cmake pkg-config \
+            python3-dev python3-pip python3-yaml \
                 autoconf libtool ninja-build zip
         sudo python3 -m pip install cython
         ;;
     suse)
         for PKG in mingw64-cross-gcc-c++ mingw64-libgcc_s_seh1 mingw64-libstdc++6 \
                 cmake pkgconf python3-devel autoconf libtool ninja zip \
-                python3-Cython gcc patch wget git; do
+                python3-Cython python3-PyYAML \
+                gcc patch wget git; do
             rpm -q $PKG >/dev/null || zypper -n install $PKG
         done
         ;;