1 // -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*-
2 // vim: ts=8 sw=2 smarttab
4 * Ceph - scalable distributed file system
6 * Copyright (C) 2004-2006 Sage Weil <sage@newdream.net>
8 * This is free software; you can redistribute it and/or
9 * modify it under the terms of the GNU Lesser General Public
10 * License version 2.1, as published by the Free Software
11 * Foundation. See file COPYING.
15 /* note: no header guard */
16 OPTION(host
, OPT_STR
, "") // "" means that ceph will use short hostname
17 OPTION(fsid
, OPT_UUID
, uuid_d())
18 OPTION(public_addr
, OPT_ADDR
, entity_addr_t())
19 OPTION(cluster_addr
, OPT_ADDR
, entity_addr_t())
20 OPTION(public_network
, OPT_STR
, "")
21 OPTION(cluster_network
, OPT_STR
, "")
22 OPTION(num_client
, OPT_INT
, 1)
23 OPTION(monmap
, OPT_STR
, "")
24 OPTION(mon_host
, OPT_STR
, "")
25 OPTION(mon_dns_srv_name
, OPT_STR
, "ceph-mon")
26 OPTION(lockdep
, OPT_BOOL
, false)
27 OPTION(lockdep_force_backtrace
, OPT_BOOL
, false) // always gather current backtrace at every lock
28 OPTION(run_dir
, OPT_STR
, "/var/run/ceph") // the "/var/run/ceph" dir, created on daemon startup
29 OPTION(admin_socket
, OPT_STR
, "$run_dir/$cluster-$name.asok") // default changed by common_preinit()
30 OPTION(admin_socket_mode
, OPT_STR
, "") // permission bits to set for admin socket file, e.g., "0775", "0755"
31 OPTION(crushtool
, OPT_STR
, "crushtool") // crushtool utility path
33 OPTION(daemonize
, OPT_BOOL
, false) // default changed by common_preinit()
34 OPTION(setuser
, OPT_STR
, "") // uid or user name
35 OPTION(setgroup
, OPT_STR
, "") // gid or group name
36 OPTION(setuser_match_path
, OPT_STR
, "") // make setuser/group conditional on this path matching ownership
37 OPTION(pid_file
, OPT_STR
, "") // default changed by common_preinit()
38 OPTION(chdir
, OPT_STR
, "/")
39 OPTION(max_open_files
, OPT_LONGLONG
, 0)
40 OPTION(restapi_log_level
, OPT_STR
, "") // default set by Python code
41 OPTION(restapi_base_url
, OPT_STR
, "") // "
42 OPTION(fatal_signal_handlers
, OPT_BOOL
, true)
43 SAFE_OPTION(erasure_code_dir
, OPT_STR
, CEPH_PKGLIBDIR
"/erasure-code") // default location for erasure-code plugins
45 OPTION(log_file
, OPT_STR
, "/var/log/ceph/$cluster-$name.log") // default changed by common_preinit()
46 OPTION(log_max_new
, OPT_INT
, 1000) // default changed by common_preinit()
47 OPTION(log_max_recent
, OPT_INT
, 10000) // default changed by common_preinit()
48 OPTION(log_to_stderr
, OPT_BOOL
, true) // default changed by common_preinit()
49 OPTION(err_to_stderr
, OPT_BOOL
, true) // default changed by common_preinit()
50 OPTION(log_to_syslog
, OPT_BOOL
, false)
51 OPTION(err_to_syslog
, OPT_BOOL
, false)
52 OPTION(log_flush_on_exit
, OPT_BOOL
, true) // default changed by common_preinit()
53 OPTION(log_stop_at_utilization
, OPT_FLOAT
, .97) // stop logging at (near) full
54 OPTION(log_to_graylog
, OPT_BOOL
, false)
55 OPTION(err_to_graylog
, OPT_BOOL
, false)
56 OPTION(log_graylog_host
, OPT_STR
, "127.0.0.1")
57 OPTION(log_graylog_port
, OPT_INT
, 12201)
59 // options will take k/v pairs, or single-item that will be assumed as general
60 // default for all, regardless of channel.
61 // e.g., "info" would be taken as the same as "default=info"
62 // also, "default=daemon audit=local0" would mean
63 // "default all to 'daemon', override 'audit' with 'local0'
64 OPTION(clog_to_monitors
, OPT_STR
, "default=true")
65 OPTION(clog_to_syslog
, OPT_STR
, "false")
66 OPTION(clog_to_syslog_level
, OPT_STR
, "info") // this level and above
67 OPTION(clog_to_syslog_facility
, OPT_STR
, "default=daemon audit=local0")
68 OPTION(clog_to_graylog
, OPT_STR
, "false")
69 OPTION(clog_to_graylog_host
, OPT_STR
, "127.0.0.1")
70 OPTION(clog_to_graylog_port
, OPT_STR
, "12201")
72 OPTION(mon_cluster_log_to_syslog
, OPT_STR
, "default=false")
73 OPTION(mon_cluster_log_to_syslog_level
, OPT_STR
, "info") // this level and above
74 OPTION(mon_cluster_log_to_syslog_facility
, OPT_STR
, "daemon")
75 OPTION(mon_cluster_log_file
, OPT_STR
,
76 "default=/var/log/ceph/$cluster.$channel.log cluster=/var/log/ceph/$cluster.log")
77 OPTION(mon_cluster_log_file_level
, OPT_STR
, "info")
78 OPTION(mon_cluster_log_to_graylog
, OPT_STR
, "false")
79 OPTION(mon_cluster_log_to_graylog_host
, OPT_STR
, "127.0.0.1")
80 OPTION(mon_cluster_log_to_graylog_port
, OPT_STR
, "12201")
82 OPTION(enable_experimental_unrecoverable_data_corrupting_features
, OPT_STR
, "")
84 SAFE_OPTION(plugin_dir
, OPT_STR
, CEPH_PKGLIBDIR
)
86 OPTION(xio_trace_mempool
, OPT_BOOL
, false) // mempool allocation counters
87 OPTION(xio_trace_msgcnt
, OPT_BOOL
, false) // incoming/outgoing msg counters
88 OPTION(xio_trace_xcon
, OPT_BOOL
, false) // Xio message encode/decode trace
89 OPTION(xio_queue_depth
, OPT_INT
, 128) // depth of Accelio msg queue
90 OPTION(xio_mp_min
, OPT_INT
, 128) // default min mempool size
91 OPTION(xio_mp_max_64
, OPT_INT
, 65536) // max 64-byte chunks (buffer is 40)
92 OPTION(xio_mp_max_256
, OPT_INT
, 8192) // max 256-byte chunks
93 OPTION(xio_mp_max_1k
, OPT_INT
, 8192) // max 1K chunks
94 OPTION(xio_mp_max_page
, OPT_INT
, 4096) // max 1K chunks
95 OPTION(xio_mp_max_hint
, OPT_INT
, 4096) // max size-hint chunks
96 OPTION(xio_portal_threads
, OPT_INT
, 2) // xio portal threads per messenger
97 OPTION(xio_max_conns_per_portal
, OPT_INT
, 32) // max xio_connections per portal/ctx
98 OPTION(xio_transport_type
, OPT_STR
, "rdma") // xio transport type: {rdma or tcp}
99 OPTION(xio_max_send_inline
, OPT_INT
, 512) // xio maximum threshold to send inline
101 OPTION(compressor_zlib_isal
, OPT_BOOL
, false)
102 OPTION(compressor_zlib_level
, OPT_INT
, 5) //regular zlib compression level, not applicable to isa-l optimized version
104 OPTION(async_compressor_enabled
, OPT_BOOL
, false)
105 OPTION(async_compressor_type
, OPT_STR
, "snappy")
106 OPTION(async_compressor_threads
, OPT_INT
, 2)
107 OPTION(async_compressor_thread_timeout
, OPT_INT
, 5)
108 OPTION(async_compressor_thread_suicide_timeout
, OPT_INT
, 30)
110 OPTION(plugin_crypto_accelerator
, OPT_STR
, "crypto_isal")
112 OPTION(mempool_debug
, OPT_BOOL
, false)
115 SUBSYS(lockdep
, 0, 1)
116 SUBSYS(context
, 0, 1)
119 SUBSYS(mds_balancer
, 1, 5)
120 SUBSYS(mds_locker
, 1, 5)
121 SUBSYS(mds_log
, 1, 5)
122 SUBSYS(mds_log_expire
, 1, 5)
123 SUBSYS(mds_migrator
, 1, 5)
127 SUBSYS(striper
, 0, 1)
128 SUBSYS(objecter
, 0, 1)
131 SUBSYS(rbd_mirror
, 0, 5)
132 SUBSYS(rbd_replay
, 0, 5)
133 SUBSYS(journaler
, 0, 5)
134 SUBSYS(objectcacher
, 0, 5)
137 SUBSYS(optracker
, 0, 5)
138 SUBSYS(objclass
, 0, 5)
139 SUBSYS(filestore
, 1, 3)
140 SUBSYS(journal
, 1, 3)
148 SUBSYS(finisher
, 1, 1)
149 SUBSYS(heartbeatmap
, 1, 5)
150 SUBSYS(perfcounter
, 1, 5)
151 SUBSYS(rgw
, 1, 5) // log level for the Rados gateway
152 SUBSYS(civetweb
, 1, 10)
153 SUBSYS(javaclient
, 1, 5)
155 SUBSYS(throttle
, 1, 1)
158 SUBSYS(compressor
, 1, 5)
159 SUBSYS(bluestore
, 1, 5)
163 SUBSYS(rocksdb
, 4, 5)
164 SUBSYS(leveldb
, 4, 5)
166 SUBSYS(kinetic
, 1, 5)
171 SUBSYS(eventtrace
, 1, 5)
173 OPTION(key
, OPT_STR
, "")
174 OPTION(keyfile
, OPT_STR
, "")
175 OPTION(keyring
, OPT_STR
,
176 // default changed by common_preinit() for mds and osd
177 "/etc/ceph/$cluster.$name.keyring,/etc/ceph/$cluster.keyring,/etc/ceph/keyring,/etc/ceph/keyring.bin,"
178 #if defined(__FreeBSD)
179 "/usr/local/etc/ceph/$cluster.$name.keyring,/usr/local/etc/ceph/$cluster.keyring,"
180 "/usr/local/etc/ceph/keyring,/usr/local/etc/ceph/keyring.bin,"
183 OPTION(heartbeat_interval
, OPT_INT
, 5)
184 OPTION(heartbeat_file
, OPT_STR
, "")
185 OPTION(heartbeat_inject_failure
, OPT_INT
, 0) // force an unhealthy heartbeat for N seconds
186 OPTION(perf
, OPT_BOOL
, true) // enable internal perf counters
188 SAFE_OPTION(ms_type
, OPT_STR
, "async+posix") // messenger backend. It will be modified in runtime, so use SAFE_OPTION
189 OPTION(ms_public_type
, OPT_STR
, "") // messenger backend
190 OPTION(ms_cluster_type
, OPT_STR
, "") // messenger backend
191 OPTION(ms_tcp_nodelay
, OPT_BOOL
, true)
192 OPTION(ms_tcp_rcvbuf
, OPT_INT
, 0)
193 OPTION(ms_tcp_prefetch_max_size
, OPT_INT
, 4096) // max prefetch size, we limit this to avoid extra memcpy
194 OPTION(ms_initial_backoff
, OPT_DOUBLE
, .2)
195 OPTION(ms_max_backoff
, OPT_DOUBLE
, 15.0)
196 OPTION(ms_crc_data
, OPT_BOOL
, true)
197 OPTION(ms_crc_header
, OPT_BOOL
, true)
198 OPTION(ms_die_on_bad_msg
, OPT_BOOL
, false)
199 OPTION(ms_die_on_unhandled_msg
, OPT_BOOL
, false)
200 OPTION(ms_die_on_old_message
, OPT_BOOL
, false) // assert if we get a dup incoming message and shouldn't have (may be triggered by pre-541cd3c64be0dfa04e8a2df39422e0eb9541a428 code)
201 OPTION(ms_die_on_skipped_message
, OPT_BOOL
, false) // assert if we skip a seq (kernel client does this intentionally)
202 OPTION(ms_dispatch_throttle_bytes
, OPT_U64
, 100 << 20)
203 OPTION(ms_bind_ipv6
, OPT_BOOL
, false)
204 OPTION(ms_bind_port_min
, OPT_INT
, 6800)
205 OPTION(ms_bind_port_max
, OPT_INT
, 7300)
206 #if !defined(__FreeBSD__)
207 OPTION(ms_bind_retry_count
, OPT_INT
, 3) // If binding fails, how many times do we retry to bind
208 OPTION(ms_bind_retry_delay
, OPT_INT
, 5) // Delay between attemps to bind
210 // FreeBSD does not use SO_REAUSEADDR so allow for a bit more time per default
211 OPTION(ms_bind_retry_count
, OPT_INT
, 6) // If binding fails, how many times do we retry to bind
212 OPTION(ms_bind_retry_delay
, OPT_INT
, 6) // Delay between attemps to bind
214 OPTION(ms_bind_before_connect
, OPT_BOOL
, true)
215 OPTION(ms_rwthread_stack_bytes
, OPT_U64
, 1024 << 10)
216 OPTION(ms_tcp_read_timeout
, OPT_U64
, 900)
217 OPTION(ms_pq_max_tokens_per_priority
, OPT_U64
, 16777216)
218 OPTION(ms_pq_min_cost
, OPT_U64
, 65536)
219 OPTION(ms_inject_socket_failures
, OPT_U64
, 0)
220 SAFE_OPTION(ms_inject_delay_type
, OPT_STR
, "") // "osd mds mon client" allowed
221 OPTION(ms_inject_delay_msg_type
, OPT_STR
, "") // the type of message to delay, as returned by Message::get_type_name(). This is an additional restriction on the general type filter ms_inject_delay_type.
222 OPTION(ms_inject_delay_max
, OPT_DOUBLE
, 1) // seconds
223 OPTION(ms_inject_delay_probability
, OPT_DOUBLE
, 0) // range [0, 1]
224 OPTION(ms_inject_internal_delays
, OPT_DOUBLE
, 0) // seconds
225 OPTION(ms_dump_on_send
, OPT_BOOL
, false) // hexdump msg to log on send
226 OPTION(ms_dump_corrupt_message_level
, OPT_INT
, 1) // debug level to hexdump undecodeable messages at
227 OPTION(ms_async_op_threads
, OPT_U64
, 3) // number of worker processing threads for async messenger created on init
228 OPTION(ms_async_max_op_threads
, OPT_U64
, 5) // max number of worker processing threads for async messenger
229 OPTION(ms_async_set_affinity
, OPT_BOOL
, true)
230 // example: ms_async_affinity_cores = 0,1
231 // The number of coreset is expected to equal to ms_async_op_threads, otherwise
232 // extra op threads will loop ms_async_affinity_cores again.
233 // If ms_async_affinity_cores is empty, all threads will be bind to current running
235 OPTION(ms_async_affinity_cores
, OPT_STR
, "")
236 OPTION(ms_async_send_inline
, OPT_BOOL
, false)
237 OPTION(ms_async_rdma_device_name
, OPT_STR
, "")
238 OPTION(ms_async_rdma_enable_hugepage
, OPT_BOOL
, false)
239 OPTION(ms_async_rdma_buffer_size
, OPT_INT
, 128 << 10)
240 OPTION(ms_async_rdma_send_buffers
, OPT_U32
, 1024)
241 OPTION(ms_async_rdma_receive_buffers
, OPT_U32
, 1024)
242 OPTION(ms_async_rdma_port_num
, OPT_U32
, 1)
243 OPTION(ms_async_rdma_polling_us
, OPT_U32
, 1000)
244 OPTION(ms_async_rdma_local_gid
, OPT_STR
, "") // GID format: "fe80:0000:0000:0000:7efe:90ff:fe72:6efe", no zero folding
245 OPTION(ms_async_rdma_roce_ver
, OPT_INT
, 1) // 0=RoCEv1, 1=RoCEv2, 2=RoCEv1.5
246 OPTION(ms_async_rdma_sl
, OPT_INT
, 3) // in RoCE, this means PCP
248 OPTION(ms_dpdk_port_id
, OPT_INT
, 0)
249 SAFE_OPTION(ms_dpdk_coremask
, OPT_STR
, "1") // it is modified in unittest so that use SAFE_OPTION to declare
250 OPTION(ms_dpdk_memory_channel
, OPT_STR
, "4")
251 OPTION(ms_dpdk_hugepages
, OPT_STR
, "")
252 OPTION(ms_dpdk_pmd
, OPT_STR
, "")
253 SAFE_OPTION(ms_dpdk_host_ipv4_addr
, OPT_STR
, "")
254 SAFE_OPTION(ms_dpdk_gateway_ipv4_addr
, OPT_STR
, "")
255 SAFE_OPTION(ms_dpdk_netmask_ipv4_addr
, OPT_STR
, "")
256 OPTION(ms_dpdk_lro
, OPT_BOOL
, true)
257 OPTION(ms_dpdk_hw_flow_control
, OPT_BOOL
, true)
258 // Weighing of a hardware network queue relative to a software queue (0=no work, 1= equal share)")
259 OPTION(ms_dpdk_hw_queue_weight
, OPT_FLOAT
, 1)
260 OPTION(ms_dpdk_debug_allow_loopback
, OPT_BOOL
, false)
261 OPTION(ms_dpdk_rx_buffer_count_per_core
, OPT_INT
, 8192)
263 OPTION(inject_early_sigterm
, OPT_BOOL
, false)
265 OPTION(mon_data
, OPT_STR
, "/var/lib/ceph/mon/$cluster-$id")
266 OPTION(mon_initial_members
, OPT_STR
, "") // list of initial cluster mon ids; if specified, need majority to form initial quorum and create new cluster
267 OPTION(mon_sync_fs_threshold
, OPT_INT
, 5) // sync() when writing this many objects; 0 to disable.
268 OPTION(mon_compact_on_start
, OPT_BOOL
, false) // compact leveldb on ceph-mon start
269 OPTION(mon_compact_on_bootstrap
, OPT_BOOL
, false) // trigger leveldb compaction on bootstrap
270 OPTION(mon_compact_on_trim
, OPT_BOOL
, true) // compact (a prefix) when we trim old states
271 OPTION(mon_osd_cache_size
, OPT_INT
, 10) // the size of osdmaps cache, not to rely on underlying store's cache
273 OPTION(mon_cpu_threads
, OPT_INT
, 4)
274 OPTION(mon_osd_mapping_pgs_per_chunk
, OPT_INT
, 4096)
275 OPTION(mon_tick_interval
, OPT_INT
, 5)
276 OPTION(mon_session_timeout
, OPT_INT
, 300) // must send keepalive or subscribe
277 OPTION(mon_subscribe_interval
, OPT_DOUBLE
, 24*3600) // for legacy clients only
278 OPTION(mon_delta_reset_interval
, OPT_DOUBLE
, 10) // seconds of inactivity before we reset the pg delta to 0
279 OPTION(mon_osd_laggy_halflife
, OPT_INT
, 60*60) // (seconds) how quickly our laggy estimations decay
280 OPTION(mon_osd_laggy_weight
, OPT_DOUBLE
, .3) // weight for new 'samples's in laggy estimations
281 OPTION(mon_osd_laggy_max_interval
, OPT_INT
, 300) // maximum value of laggy_interval in laggy estimations
282 OPTION(mon_osd_adjust_heartbeat_grace
, OPT_BOOL
, true) // true if we should scale based on laggy estimations
283 OPTION(mon_osd_adjust_down_out_interval
, OPT_BOOL
, true) // true if we should scale based on laggy estimations
284 OPTION(mon_osd_auto_mark_in
, OPT_BOOL
, false) // mark any booting osds 'in'
285 OPTION(mon_osd_auto_mark_auto_out_in
, OPT_BOOL
, true) // mark booting auto-marked-out osds 'in'
286 OPTION(mon_osd_auto_mark_new_in
, OPT_BOOL
, true) // mark booting new osds 'in'
287 OPTION(mon_osd_down_out_interval
, OPT_INT
, 600) // seconds
288 OPTION(mon_osd_down_out_subtree_limit
, OPT_STR
, "rack") // smallest crush unit/type that we will not automatically mark out
289 OPTION(mon_osd_min_up_ratio
, OPT_DOUBLE
, .3) // min osds required to be up to mark things down
290 OPTION(mon_osd_min_in_ratio
, OPT_DOUBLE
, .75) // min osds required to be in to mark things out
291 OPTION(mon_osd_max_op_age
, OPT_DOUBLE
, 32) // max op age before we get concerned (make it a power of 2)
292 OPTION(mon_osd_max_split_count
, OPT_INT
, 32) // largest number of PGs per "involved" OSD to let split create
293 OPTION(mon_osd_allow_primary_temp
, OPT_BOOL
, false) // allow primary_temp to be set in the osdmap
294 OPTION(mon_osd_allow_primary_affinity
, OPT_BOOL
, false) // allow primary_affinity to be set in the osdmap
295 OPTION(mon_osd_prime_pg_temp
, OPT_BOOL
, true) // prime osdmap with pg mapping changes
296 OPTION(mon_osd_prime_pg_temp_max_time
, OPT_FLOAT
, .5) // max time to spend priming
297 OPTION(mon_osd_prime_pg_temp_max_estimate
, OPT_FLOAT
, .25) // max estimate of pg total before we do all pgs in parallel
298 OPTION(mon_osd_pool_ec_fast_read
, OPT_BOOL
, false) // whether turn on fast read on the pool or not
299 OPTION(mon_stat_smooth_intervals
, OPT_INT
, 6) // smooth stats over last N PGMap maps
300 OPTION(mon_election_timeout
, OPT_FLOAT
, 5) // on election proposer, max waiting time for all ACKs
301 OPTION(mon_lease
, OPT_FLOAT
, 5) // lease interval
302 OPTION(mon_lease_renew_interval_factor
, OPT_FLOAT
, .6) // on leader, to renew the lease
303 OPTION(mon_lease_ack_timeout_factor
, OPT_FLOAT
, 2.0) // on leader, if lease isn't acked by all peons
304 OPTION(mon_accept_timeout_factor
, OPT_FLOAT
, 2.0) // on leader, if paxos update isn't accepted
306 OPTION(mon_clock_drift_allowed
, OPT_FLOAT
, .050) // allowed clock drift between monitors
307 OPTION(mon_clock_drift_warn_backoff
, OPT_FLOAT
, 5) // exponential backoff for clock drift warnings
308 OPTION(mon_timecheck_interval
, OPT_FLOAT
, 300.0) // on leader, timecheck (clock drift check) interval (seconds)
309 OPTION(mon_timecheck_skew_interval
, OPT_FLOAT
, 30.0) // on leader, timecheck (clock drift check) interval when in presence of a skew (seconds)
310 OPTION(mon_pg_stuck_threshold
, OPT_INT
, 300) // number of seconds after which pgs can be considered inactive, unclean, or stale (see doc/control.rst under dump_stuck for more info)
311 OPTION(mon_pg_min_inactive
, OPT_U64
, 1) // the number of PGs which have to be inactive longer than 'mon_pg_stuck_threshold' before health goes into ERR. 0 means disabled, never go into ERR.
312 OPTION(mon_pg_warn_min_per_osd
, OPT_INT
, 30) // min # pgs per (in) osd before we warn the admin
313 OPTION(mon_pg_warn_max_per_osd
, OPT_INT
, 300) // max # pgs per (in) osd before we warn the admin
314 OPTION(mon_pg_warn_max_object_skew
, OPT_FLOAT
, 10.0) // max skew few average in objects per pg
315 OPTION(mon_pg_warn_min_objects
, OPT_INT
, 10000) // do not warn below this object #
316 OPTION(mon_pg_warn_min_pool_objects
, OPT_INT
, 1000) // do not warn on pools below this object #
317 OPTION(mon_pg_check_down_all_threshold
, OPT_FLOAT
, .5) // threshold of down osds after which we check all pgs
318 OPTION(mon_cache_target_full_warn_ratio
, OPT_FLOAT
, .66) // position between pool cache_target_full and max where we start warning
319 OPTION(mon_osd_full_ratio
, OPT_FLOAT
, .95) // what % full makes an OSD "full"
320 OPTION(mon_osd_backfillfull_ratio
, OPT_FLOAT
, .90) // what % full makes an OSD backfill full (backfill halted)
321 OPTION(mon_osd_nearfull_ratio
, OPT_FLOAT
, .85) // what % full makes an OSD near full
322 OPTION(mon_osd_initial_require_min_compat_client
, OPT_STR
, "hammer")
323 OPTION(mon_allow_pool_delete
, OPT_BOOL
, false) // allow pool deletion
324 OPTION(mon_fake_pool_delete
, OPT_BOOL
, false) // fake pool deletion (add _DELETED suffix)
325 OPTION(mon_globalid_prealloc
, OPT_U32
, 10000) // how many globalids to prealloc
326 OPTION(mon_osd_report_timeout
, OPT_INT
, 900) // grace period before declaring unresponsive OSDs dead
327 OPTION(mon_force_standby_active
, OPT_BOOL
, true) // should mons force standby-replay mds to be active
328 OPTION(mon_warn_on_legacy_crush_tunables
, OPT_BOOL
, true) // warn if crush tunables are too old (older than mon_min_crush_required_version)
329 OPTION(mon_crush_min_required_version
, OPT_STR
, "firefly")
330 OPTION(mon_warn_on_crush_straw_calc_version_zero
, OPT_BOOL
, true) // warn if crush straw_calc_version==0
331 OPTION(mon_warn_on_osd_down_out_interval_zero
, OPT_BOOL
, true) // warn if 'mon_osd_down_out_interval == 0'
332 OPTION(mon_warn_on_cache_pools_without_hit_sets
, OPT_BOOL
, true)
333 OPTION(mon_warn_osd_usage_min_max_delta
, OPT_FLOAT
, .40) // warn if difference between min and max OSD utilizations exceeds specified amount
334 OPTION(mon_min_osdmap_epochs
, OPT_INT
, 500)
335 OPTION(mon_max_pgmap_epochs
, OPT_INT
, 500)
336 OPTION(mon_max_log_epochs
, OPT_INT
, 500)
337 OPTION(mon_max_mdsmap_epochs
, OPT_INT
, 500)
338 OPTION(mon_max_osd
, OPT_INT
, 10000)
339 OPTION(mon_probe_timeout
, OPT_DOUBLE
, 2.0)
340 OPTION(mon_slurp_timeout
, OPT_DOUBLE
, 10.0)
341 OPTION(mon_slurp_bytes
, OPT_INT
, 256*1024) // limit size of slurp messages
342 OPTION(mon_client_bytes
, OPT_U64
, 100ul << 20) // client msg data allowed in memory (in bytes)
343 OPTION(mon_mgr_proxy_client_bytes_ratio
, OPT_FLOAT
, .3) // ratio of mon_client_bytes that can be consumed by proxied mgr commands before we error out to client
344 OPTION(mon_daemon_bytes
, OPT_U64
, 400ul << 20) // mds, osd message memory cap (in bytes)
345 OPTION(mon_max_log_entries_per_event
, OPT_INT
, 4096)
346 OPTION(mon_reweight_min_pgs_per_osd
, OPT_U64
, 10) // min pgs per osd for reweight-by-pg command
347 OPTION(mon_reweight_min_bytes_per_osd
, OPT_U64
, 100*1024*1024) // min bytes per osd for reweight-by-utilization command
348 OPTION(mon_reweight_max_osds
, OPT_INT
, 4) // max osds to change per reweight-by-* command
349 OPTION(mon_reweight_max_change
, OPT_DOUBLE
, 0.05)
350 OPTION(mon_health_data_update_interval
, OPT_FLOAT
, 60.0)
351 OPTION(mon_health_to_clog
, OPT_BOOL
, true)
352 OPTION(mon_health_to_clog_interval
, OPT_INT
, 3600)
353 OPTION(mon_health_to_clog_tick_interval
, OPT_DOUBLE
, 60.0)
354 OPTION(mon_data_avail_crit
, OPT_INT
, 5)
355 OPTION(mon_data_avail_warn
, OPT_INT
, 30)
356 OPTION(mon_data_size_warn
, OPT_U64
, 15*1024*1024*1024) // issue a warning when the monitor's data store goes over 15GB (in bytes)
357 OPTION(mon_warn_not_scrubbed
, OPT_INT
, 0)
358 OPTION(mon_warn_not_deep_scrubbed
, OPT_INT
, 0)
359 OPTION(mon_scrub_interval
, OPT_INT
, 3600*24) // once a day
360 OPTION(mon_scrub_timeout
, OPT_INT
, 60*5) // let's give it 5 minutes; why not.
361 OPTION(mon_scrub_max_keys
, OPT_INT
, 100) // max number of keys to scrub each time
362 OPTION(mon_scrub_inject_crc_mismatch
, OPT_DOUBLE
, 0.0) // probability of injected crc mismatch [0.0, 1.0]
363 OPTION(mon_scrub_inject_missing_keys
, OPT_DOUBLE
, 0.0) // probability of injected missing keys [0.0, 1.0]
364 OPTION(mon_config_key_max_entry_size
, OPT_INT
, 4096) // max num bytes per config-key entry
365 OPTION(mon_sync_timeout
, OPT_DOUBLE
, 60.0)
366 OPTION(mon_sync_max_payload_size
, OPT_U32
, 1048576) // max size for a sync chunk payload (say, 1MB)
367 OPTION(mon_sync_debug
, OPT_BOOL
, false) // enable sync-specific debug
368 OPTION(mon_sync_debug_leader
, OPT_INT
, -1) // monitor to be used as the sync leader
369 OPTION(mon_sync_debug_provider
, OPT_INT
, -1) // monitor to be used as the sync provider
370 OPTION(mon_sync_debug_provider_fallback
, OPT_INT
, -1) // monitor to be used as fallback if sync provider fails
371 OPTION(mon_inject_sync_get_chunk_delay
, OPT_DOUBLE
, 0) // inject N second delay on each get_chunk request
372 OPTION(mon_osd_min_down_reporters
, OPT_INT
, 2) // number of OSDs from different subtrees who need to report a down OSD for it to count
373 OPTION(mon_osd_reporter_subtree_level
, OPT_STR
, "host") // in which level of parent bucket the reporters are counted
374 OPTION(mon_osd_force_trim_to
, OPT_INT
, 0) // force mon to trim maps to this point, regardless of min_last_epoch_clean (dangerous, use with care)
375 OPTION(mon_mds_force_trim_to
, OPT_INT
, 0) // force mon to trim mdsmaps to this point (dangerous, use with care)
376 OPTION(mon_mds_skip_sanity
, OPT_BOOL
, false) // skip safety assertions on FSMap (in case of bugs where we want to continue anyway)
378 // monitor debug options
379 OPTION(mon_debug_deprecated_as_obsolete
, OPT_BOOL
, false) // consider deprecated commands as obsolete
382 OPTION(mon_debug_dump_transactions
, OPT_BOOL
, false)
383 OPTION(mon_debug_dump_json
, OPT_BOOL
, false)
384 OPTION(mon_debug_dump_location
, OPT_STR
, "/var/log/ceph/$cluster-$name.tdump")
385 OPTION(mon_debug_no_require_luminous
, OPT_BOOL
, false)
386 OPTION(mon_inject_transaction_delay_max
, OPT_DOUBLE
, 10.0) // seconds
387 OPTION(mon_inject_transaction_delay_probability
, OPT_DOUBLE
, 0) // range [0, 1]
389 OPTION(mon_sync_provider_kill_at
, OPT_INT
, 0) // kill the sync provider at a specific point in the work flow
390 OPTION(mon_sync_requester_kill_at
, OPT_INT
, 0) // kill the sync requester at a specific point in the work flow
391 OPTION(mon_force_quorum_join
, OPT_BOOL
, false) // force monitor to join quorum even if it has been previously removed from the map
392 OPTION(mon_keyvaluedb
, OPT_STR
, "rocksdb") // type of keyvaluedb backend
394 // UNSAFE -- TESTING ONLY! Allows addition of a cache tier with preexisting snaps
395 OPTION(mon_debug_unsafe_allow_tier_with_nonempty_snaps
, OPT_BOOL
, false)
397 OPTION(paxos_stash_full_interval
, OPT_INT
, 25) // how often (in commits) to stash a full copy of the PaxosService state
398 OPTION(paxos_max_join_drift
, OPT_INT
, 10) // max paxos iterations before we must first sync the monitor stores
399 OPTION(paxos_propose_interval
, OPT_DOUBLE
, 1.0) // gather updates for this long before proposing a map update
400 OPTION(paxos_min_wait
, OPT_DOUBLE
, 0.05) // min time to gather updates for after period of inactivity
401 OPTION(paxos_min
, OPT_INT
, 500) // minimum number of paxos states to keep around
402 OPTION(paxos_trim_min
, OPT_INT
, 250) // number of extra proposals tolerated before trimming
403 OPTION(paxos_trim_max
, OPT_INT
, 500) // max number of extra proposals to trim at a time
404 OPTION(paxos_service_trim_min
, OPT_INT
, 250) // minimum amount of versions to trigger a trim (0 disables it)
405 OPTION(paxos_service_trim_max
, OPT_INT
, 500) // maximum amount of versions to trim during a single proposal (0 disables it)
406 OPTION(paxos_kill_at
, OPT_INT
, 0)
407 OPTION(auth_cluster_required
, OPT_STR
, "cephx") // required of mon, mds, osd daemons
408 OPTION(auth_service_required
, OPT_STR
, "cephx") // required by daemons of clients
409 OPTION(auth_client_required
, OPT_STR
, "cephx, none") // what clients require of daemons
410 OPTION(auth_supported
, OPT_STR
, "") // deprecated; default value for above if they are not defined.
411 OPTION(max_rotating_auth_attempts
, OPT_INT
, 10)
412 OPTION(cephx_require_signatures
, OPT_BOOL
, false) // If true, don't talk to Cephx partners if they don't support message signing; off by default
413 OPTION(cephx_cluster_require_signatures
, OPT_BOOL
, false)
414 OPTION(cephx_service_require_signatures
, OPT_BOOL
, false)
415 OPTION(cephx_sign_messages
, OPT_BOOL
, true) // Default to signing session messages if supported
416 OPTION(auth_mon_ticket_ttl
, OPT_DOUBLE
, 60*60*12)
417 OPTION(auth_service_ticket_ttl
, OPT_DOUBLE
, 60*60)
418 OPTION(auth_debug
, OPT_BOOL
, false) // if true, assert when weird things happen
419 OPTION(mon_client_hunt_parallel
, OPT_U32
, 2) // how many mons to try to connect to in parallel during hunt
420 OPTION(mon_client_hunt_interval
, OPT_DOUBLE
, 3.0) // try new mon every N seconds until we connect
421 OPTION(mon_client_ping_interval
, OPT_DOUBLE
, 10.0) // ping every N seconds
422 OPTION(mon_client_ping_timeout
, OPT_DOUBLE
, 30.0) // fail if we don't hear back
423 OPTION(mon_client_hunt_interval_backoff
, OPT_DOUBLE
, 2.0) // each time we reconnect to a monitor, double our timeout
424 OPTION(mon_client_hunt_interval_max_multiple
, OPT_DOUBLE
, 10.0) // up to a max of 10*default (30 seconds)
425 OPTION(mon_client_max_log_entries_per_message
, OPT_INT
, 1000)
426 OPTION(mon_max_pool_pg_num
, OPT_INT
, 65536)
427 OPTION(mon_pool_quota_warn_threshold
, OPT_INT
, 0) // percent of quota at which to issue warnings
428 OPTION(mon_pool_quota_crit_threshold
, OPT_INT
, 0) // percent of quota at which to issue errors
429 OPTION(client_cache_size
, OPT_INT
, 16384)
430 OPTION(client_cache_mid
, OPT_FLOAT
, .75)
431 OPTION(client_use_random_mds
, OPT_BOOL
, false)
432 OPTION(client_mount_timeout
, OPT_DOUBLE
, 300.0)
433 OPTION(client_tick_interval
, OPT_DOUBLE
, 1.0)
434 OPTION(client_trace
, OPT_STR
, "")
435 OPTION(client_readahead_min
, OPT_LONGLONG
, 128*1024) // readahead at _least_ this much.
436 OPTION(client_readahead_max_bytes
, OPT_LONGLONG
, 0) // default unlimited
437 OPTION(client_readahead_max_periods
, OPT_LONGLONG
, 4) // as multiple of file layout period (object size * num stripes)
438 OPTION(client_reconnect_stale
, OPT_BOOL
, false) // automatically reconnect stale session
439 OPTION(client_snapdir
, OPT_STR
, ".snap")
440 OPTION(client_mountpoint
, OPT_STR
, "/")
441 OPTION(client_mount_uid
, OPT_INT
, -1)
442 OPTION(client_mount_gid
, OPT_INT
, -1)
443 OPTION(client_notify_timeout
, OPT_INT
, 10) // in seconds
444 OPTION(osd_client_watch_timeout
, OPT_INT
, 30) // in seconds
445 OPTION(client_caps_release_delay
, OPT_INT
, 5) // in seconds
446 OPTION(client_quota_df
, OPT_BOOL
, true) // use quota for df on subdir mounts
447 OPTION(client_oc
, OPT_BOOL
, true)
448 OPTION(client_oc_size
, OPT_INT
, 1024*1024* 200) // MB * n
449 OPTION(client_oc_max_dirty
, OPT_INT
, 1024*1024* 100) // MB * n (dirty OR tx.. bigish)
450 OPTION(client_oc_target_dirty
, OPT_INT
, 1024*1024* 8) // target dirty (keep this smallish)
451 OPTION(client_oc_max_dirty_age
, OPT_DOUBLE
, 5.0) // max age in cache before writeback
452 OPTION(client_oc_max_objects
, OPT_INT
, 1000) // max objects in cache
453 OPTION(client_debug_getattr_caps
, OPT_BOOL
, false) // check if MDS reply contains wanted caps
454 OPTION(client_debug_force_sync_read
, OPT_BOOL
, false) // always read synchronously (go to osds)
455 OPTION(client_debug_inject_tick_delay
, OPT_INT
, 0) // delay the client tick for a number of seconds
456 OPTION(client_max_inline_size
, OPT_U64
, 4096)
457 OPTION(client_inject_release_failure
, OPT_BOOL
, false) // synthetic client bug for testing
458 OPTION(client_inject_fixed_oldest_tid
, OPT_BOOL
, false) // synthetic client bug for testing
459 OPTION(client_metadata
, OPT_STR
, "")
460 OPTION(client_acl_type
, OPT_STR
, "")
461 OPTION(client_permissions
, OPT_BOOL
, true)
462 OPTION(client_dirsize_rbytes
, OPT_BOOL
, true)
464 // note: the max amount of "in flight" dirty data is roughly (max - target)
465 OPTION(fuse_use_invalidate_cb
, OPT_BOOL
, true) // use fuse 2.8+ invalidate callback to keep page cache consistent
466 OPTION(fuse_disable_pagecache
, OPT_BOOL
, false)
467 OPTION(fuse_allow_other
, OPT_BOOL
, true)
468 OPTION(fuse_default_permissions
, OPT_BOOL
, false)
469 OPTION(fuse_big_writes
, OPT_BOOL
, true)
470 OPTION(fuse_atomic_o_trunc
, OPT_BOOL
, true)
471 OPTION(fuse_debug
, OPT_BOOL
, false)
472 OPTION(fuse_multithreaded
, OPT_BOOL
, true)
473 OPTION(fuse_require_active_mds
, OPT_BOOL
, true) // if ceph_fuse requires active mds server
474 OPTION(fuse_syncfs_on_mksnap
, OPT_BOOL
, true)
475 OPTION(fuse_set_user_groups
, OPT_BOOL
, false) // if ceph_fuse fills in group lists or not
477 OPTION(client_try_dentry_invalidate
, OPT_BOOL
, true) // the client should try to use dentry invaldation instead of remounting, on kernels it believes that will work for
478 OPTION(client_die_on_failed_remount
, OPT_BOOL
, true)
479 OPTION(client_check_pool_perm
, OPT_BOOL
, true)
480 OPTION(client_use_faked_inos
, OPT_BOOL
, false)
481 OPTION(client_mds_namespace
, OPT_STR
, "")
483 OPTION(crush_location
, OPT_STR
, "") // whitespace-separated list of key=value pairs describing crush location
484 OPTION(crush_location_hook
, OPT_STR
, "")
485 OPTION(crush_location_hook_timeout
, OPT_INT
, 10)
487 OPTION(objecter_tick_interval
, OPT_DOUBLE
, 5.0)
488 OPTION(objecter_timeout
, OPT_DOUBLE
, 10.0) // before we ask for a map
489 OPTION(objecter_inflight_op_bytes
, OPT_U64
, 1024*1024*100) // max in-flight data (both directions)
490 OPTION(objecter_inflight_ops
, OPT_U64
, 1024) // max in-flight ios
491 OPTION(objecter_completion_locks_per_session
, OPT_U64
, 32) // num of completion locks per each session, for serializing same object responses
492 OPTION(objecter_inject_no_watch_ping
, OPT_BOOL
, false) // suppress watch pings
493 OPTION(objecter_retry_writes_after_first_reply
, OPT_BOOL
, false) // ignore the first reply for each write, and resend the osd op instead
494 OPTION(objecter_debug_inject_relock_delay
, OPT_BOOL
, false)
496 // Max number of deletes at once in a single Filer::purge call
497 OPTION(filer_max_purge_ops
, OPT_U32
, 10)
498 // Max number of truncate at once in a single Filer::truncate call
499 OPTION(filer_max_truncate_ops
, OPT_U32
, 128)
501 OPTION(journaler_write_head_interval
, OPT_INT
, 15)
502 OPTION(journaler_prefetch_periods
, OPT_INT
, 10) // * journal object size
503 OPTION(journaler_prezero_periods
, OPT_INT
, 5) // * journal object size
504 OPTION(mds_data
, OPT_STR
, "/var/lib/ceph/mds/$cluster-$id")
505 OPTION(mds_max_file_size
, OPT_U64
, 1ULL << 40) // Used when creating new CephFS. Change with 'ceph mds set max_file_size <size>' afterwards
506 // max xattr kv pairs size for each dir/file
507 OPTION(mds_max_xattr_pairs_size
, OPT_U32
, 64 << 10)
508 OPTION(mds_cache_size
, OPT_INT
, 100000)
509 OPTION(mds_cache_mid
, OPT_FLOAT
, .7)
510 OPTION(mds_max_file_recover
, OPT_U32
, 32)
511 OPTION(mds_dir_max_commit_size
, OPT_INT
, 10) // MB
512 OPTION(mds_dir_keys_per_op
, OPT_INT
, 16384)
513 OPTION(mds_decay_halflife
, OPT_FLOAT
, 5)
514 OPTION(mds_beacon_interval
, OPT_FLOAT
, 4)
515 OPTION(mds_beacon_grace
, OPT_FLOAT
, 15)
516 OPTION(mds_enforce_unique_name
, OPT_BOOL
, true)
517 OPTION(mds_blacklist_interval
, OPT_FLOAT
, 24.0*60.0) // how long to blacklist failed nodes
518 OPTION(mds_session_timeout
, OPT_FLOAT
, 60) // cap bits and leases time out if client idle
519 OPTION(mds_sessionmap_keys_per_op
, OPT_U32
, 1024) // how many sessions should I try to load/store in a single OMAP operation?
520 OPTION(mds_revoke_cap_timeout
, OPT_FLOAT
, 60) // detect clients which aren't revoking caps
521 OPTION(mds_recall_state_timeout
, OPT_FLOAT
, 60) // detect clients which aren't trimming caps
522 OPTION(mds_freeze_tree_timeout
, OPT_FLOAT
, 30) // detecting freeze tree deadlock
523 OPTION(mds_session_autoclose
, OPT_FLOAT
, 300) // autoclose idle session
524 OPTION(mds_health_summarize_threshold
, OPT_INT
, 10) // collapse N-client health metrics to a single 'many'
525 OPTION(mds_health_cache_threshold
, OPT_FLOAT
, 1.5) // warn on cache size if it exceeds mds_cache_size by this factor
526 OPTION(mds_reconnect_timeout
, OPT_FLOAT
, 45) // seconds to wait for clients during mds restart
527 // make it (mds_session_timeout - mds_beacon_grace)
528 OPTION(mds_tick_interval
, OPT_FLOAT
, 5)
529 OPTION(mds_dirstat_min_interval
, OPT_FLOAT
, 1) // try to avoid propagating more often than this
530 OPTION(mds_scatter_nudge_interval
, OPT_FLOAT
, 5) // how quickly dirstat changes propagate up the hierarchy
531 OPTION(mds_client_prealloc_inos
, OPT_INT
, 1000)
532 OPTION(mds_early_reply
, OPT_BOOL
, true)
533 OPTION(mds_default_dir_hash
, OPT_INT
, CEPH_STR_HASH_RJENKINS
)
534 OPTION(mds_log_pause
, OPT_BOOL
, false)
535 OPTION(mds_log_skip_corrupt_events
, OPT_BOOL
, false)
536 OPTION(mds_log_max_events
, OPT_INT
, -1)
537 OPTION(mds_log_events_per_segment
, OPT_INT
, 1024)
538 OPTION(mds_log_segment_size
, OPT_INT
, 0) // segment size for mds log, default to default file_layout_t
539 OPTION(mds_log_max_segments
, OPT_U32
, 30)
540 OPTION(mds_log_max_expiring
, OPT_INT
, 20)
541 OPTION(mds_bal_export_pin
, OPT_BOOL
, true) // allow clients to pin directory trees to ranks
542 OPTION(mds_bal_sample_interval
, OPT_DOUBLE
, 3.0) // every 3 seconds
543 OPTION(mds_bal_replicate_threshold
, OPT_FLOAT
, 8000)
544 OPTION(mds_bal_unreplicate_threshold
, OPT_FLOAT
, 0)
545 OPTION(mds_bal_frag
, OPT_BOOL
, true)
546 OPTION(mds_bal_split_size
, OPT_INT
, 10000)
547 OPTION(mds_bal_split_rd
, OPT_FLOAT
, 25000)
548 OPTION(mds_bal_split_wr
, OPT_FLOAT
, 10000)
549 OPTION(mds_bal_split_bits
, OPT_INT
, 3)
550 OPTION(mds_bal_merge_size
, OPT_INT
, 50)
551 OPTION(mds_bal_interval
, OPT_INT
, 10) // seconds
552 OPTION(mds_bal_fragment_interval
, OPT_INT
, 5) // seconds
553 OPTION(mds_bal_fragment_size_max
, OPT_INT
, 10000*10) // order of magnitude higher than split size
554 OPTION(mds_bal_fragment_fast_factor
, OPT_FLOAT
, 1.5) // multiple of size_max that triggers immediate split
555 OPTION(mds_bal_idle_threshold
, OPT_FLOAT
, 0)
556 OPTION(mds_bal_max
, OPT_INT
, -1)
557 OPTION(mds_bal_max_until
, OPT_INT
, -1)
558 OPTION(mds_bal_mode
, OPT_INT
, 0)
559 OPTION(mds_bal_min_rebalance
, OPT_FLOAT
, .1) // must be this much above average before we export anything
560 OPTION(mds_bal_min_start
, OPT_FLOAT
, .2) // if we need less than this, we don't do anything
561 OPTION(mds_bal_need_min
, OPT_FLOAT
, .8) // take within this range of what we need
562 OPTION(mds_bal_need_max
, OPT_FLOAT
, 1.2)
563 OPTION(mds_bal_midchunk
, OPT_FLOAT
, .3) // any sub bigger than this taken in full
564 OPTION(mds_bal_minchunk
, OPT_FLOAT
, .001) // never take anything smaller than this
565 OPTION(mds_bal_target_decay
, OPT_DOUBLE
, 10.0) // target decay half-life in MDSMap (2x larger is approx. 2x slower)
566 OPTION(mds_replay_interval
, OPT_FLOAT
, 1.0) // time to wait before starting replay again
567 OPTION(mds_shutdown_check
, OPT_INT
, 0)
568 OPTION(mds_thrash_exports
, OPT_INT
, 0)
569 OPTION(mds_thrash_fragments
, OPT_INT
, 0)
570 OPTION(mds_dump_cache_on_map
, OPT_BOOL
, false)
571 OPTION(mds_dump_cache_after_rejoin
, OPT_BOOL
, false)
572 OPTION(mds_verify_scatter
, OPT_BOOL
, false)
573 OPTION(mds_debug_scatterstat
, OPT_BOOL
, false)
574 OPTION(mds_debug_frag
, OPT_BOOL
, false)
575 OPTION(mds_debug_auth_pins
, OPT_BOOL
, false)
576 OPTION(mds_debug_subtrees
, OPT_BOOL
, false)
577 OPTION(mds_kill_mdstable_at
, OPT_INT
, 0)
578 OPTION(mds_kill_export_at
, OPT_INT
, 0)
579 OPTION(mds_kill_import_at
, OPT_INT
, 0)
580 OPTION(mds_kill_link_at
, OPT_INT
, 0)
581 OPTION(mds_kill_rename_at
, OPT_INT
, 0)
582 OPTION(mds_kill_openc_at
, OPT_INT
, 0)
583 OPTION(mds_kill_journal_at
, OPT_INT
, 0)
584 OPTION(mds_kill_journal_expire_at
, OPT_INT
, 0)
585 OPTION(mds_kill_journal_replay_at
, OPT_INT
, 0)
586 OPTION(mds_journal_format
, OPT_U32
, 1) // Default to most recent JOURNAL_FORMAT_*
587 OPTION(mds_kill_create_at
, OPT_INT
, 0)
588 OPTION(mds_inject_traceless_reply_probability
, OPT_DOUBLE
, 0) /* percentage
589 of MDS modify replies to skip sending the
590 client a trace on [0-1]*/
591 OPTION(mds_wipe_sessions
, OPT_BOOL
, 0)
592 OPTION(mds_wipe_ino_prealloc
, OPT_BOOL
, 0)
593 OPTION(mds_skip_ino
, OPT_INT
, 0)
594 OPTION(mds_standby_for_name
, OPT_STR
, "")
595 OPTION(mds_standby_for_rank
, OPT_INT
, -1)
596 OPTION(mds_standby_for_fscid
, OPT_INT
, -1)
597 OPTION(mds_standby_replay
, OPT_BOOL
, false)
598 OPTION(mds_enable_op_tracker
, OPT_BOOL
, true) // enable/disable MDS op tracking
599 OPTION(mds_op_history_size
, OPT_U32
, 20) // Max number of completed ops to track
600 OPTION(mds_op_history_duration
, OPT_U32
, 600) // Oldest completed op to track
601 OPTION(mds_op_complaint_time
, OPT_FLOAT
, 30) // how many seconds old makes an op complaint-worthy
602 OPTION(mds_op_log_threshold
, OPT_INT
, 5) // how many op log messages to show in one go
603 OPTION(mds_snap_min_uid
, OPT_U32
, 0) // The minimum UID required to create a snapshot
604 OPTION(mds_snap_max_uid
, OPT_U32
, 4294967294) // The maximum UID allowed to create a snapshot
605 OPTION(mds_snap_rstat
, OPT_BOOL
, false) // enable/disbale nested stat for snapshot
606 OPTION(mds_verify_backtrace
, OPT_U32
, 1)
607 // detect clients which aren't trimming completed requests
608 OPTION(mds_max_completed_flushes
, OPT_U32
, 100000)
609 OPTION(mds_max_completed_requests
, OPT_U32
, 100000)
611 OPTION(mds_action_on_write_error
, OPT_U32
, 1) // 0: ignore; 1: force readonly; 2: crash
612 OPTION(mds_mon_shutdown_timeout
, OPT_DOUBLE
, 5)
614 // Maximum number of concurrent stray files to purge
615 OPTION(mds_max_purge_files
, OPT_U32
, 64)
616 // Maximum number of concurrent RADOS ops to issue in purging
617 OPTION(mds_max_purge_ops
, OPT_U32
, 8192)
618 // Maximum number of concurrent RADOS ops to issue in purging, scaled by PG count
619 OPTION(mds_max_purge_ops_per_pg
, OPT_FLOAT
, 0.5)
621 OPTION(mds_purge_queue_busy_flush_period
, OPT_FLOAT
, 1.0)
623 OPTION(mds_root_ino_uid
, OPT_INT
, 0) // The UID of / on new filesystems
624 OPTION(mds_root_ino_gid
, OPT_INT
, 0) // The GID of / on new filesystems
626 OPTION(mds_max_scrub_ops_in_progress
, OPT_INT
, 5) // the number of simultaneous scrubs allowed
628 // Maximum number of damaged frags/dentries before whole MDS rank goes damaged
629 OPTION(mds_damage_table_max_entries
, OPT_INT
, 10000)
631 // verify backend can support configured max object name length
632 OPTION(osd_check_max_object_name_len_on_startup
, OPT_BOOL
, true)
634 // Maximum number of backfills to or from a single osd
635 OPTION(osd_max_backfills
, OPT_U64
, 1)
637 // Minimum recovery priority (255 = max, smaller = lower)
638 OPTION(osd_min_recovery_priority
, OPT_INT
, 0)
640 // Seconds to wait before retrying refused backfills
641 OPTION(osd_backfill_retry_interval
, OPT_DOUBLE
, 30.0)
643 // Seconds to wait before retrying refused recovery
644 OPTION(osd_recovery_retry_interval
, OPT_DOUBLE
, 30.0)
646 // max agent flush ops
647 OPTION(osd_agent_max_ops
, OPT_INT
, 4)
648 OPTION(osd_agent_max_low_ops
, OPT_INT
, 2)
649 OPTION(osd_agent_min_evict_effort
, OPT_FLOAT
, .1)
650 OPTION(osd_agent_quantize_effort
, OPT_FLOAT
, .1)
651 OPTION(osd_agent_delay_time
, OPT_FLOAT
, 5.0)
653 // osd ignore history.last_epoch_started in find_best_info
654 OPTION(osd_find_best_info_ignore_history_les
, OPT_BOOL
, false)
656 // decay atime and hist histograms after how many objects go by
657 OPTION(osd_agent_hist_halflife
, OPT_INT
, 1000)
659 // must be this amount over the threshold to enable,
660 // this amount below the threshold to disable.
661 OPTION(osd_agent_slop
, OPT_FLOAT
, .02)
663 OPTION(osd_uuid
, OPT_UUID
, uuid_d())
664 OPTION(osd_data
, OPT_STR
, "/var/lib/ceph/osd/$cluster-$id")
665 OPTION(osd_journal
, OPT_STR
, "/var/lib/ceph/osd/$cluster-$id/journal")
666 OPTION(osd_journal_size
, OPT_INT
, 5120) // in mb
667 OPTION(osd_journal_flush_on_shutdown
, OPT_BOOL
, true) // Flush journal to data store on shutdown
668 // flags for specific control purpose during osd mount() process.
669 // e.g., can be 1 to skip over replaying journal
670 // or 2 to skip over mounting omap or 3 to skip over both.
671 // This might be helpful in case the journal is totally corrupted
672 // and we still want to bring the osd daemon back normally, etc.
673 OPTION(osd_os_flags
, OPT_U32
, 0)
674 OPTION(osd_max_write_size
, OPT_INT
, 90)
675 OPTION(osd_max_pgls
, OPT_U64
, 1024) // max number of pgls entries to return
676 OPTION(osd_client_message_size_cap
, OPT_U64
, 500*1024L*1024L) // client data allowed in-memory (in bytes)
677 OPTION(osd_client_message_cap
, OPT_U64
, 100) // num client messages allowed in-memory
678 OPTION(osd_pg_bits
, OPT_INT
, 6) // bits per osd
679 OPTION(osd_pgp_bits
, OPT_INT
, 6) // bits per osd
680 OPTION(osd_crush_chooseleaf_type
, OPT_INT
, 1) // 1 = host
681 OPTION(osd_pool_use_gmt_hitset
, OPT_BOOL
, true) // try to use gmt for hitset archive names if all osds in cluster support it.
682 OPTION(osd_crush_update_on_start
, OPT_BOOL
, true)
683 OPTION(osd_crush_initial_weight
, OPT_DOUBLE
, -1) // if >=0, the initial weight is for newly added osds.
684 OPTION(osd_pool_default_crush_rule
, OPT_INT
, -1) // deprecated for osd_pool_default_crush_replicated_ruleset
685 OPTION(osd_pool_default_crush_replicated_ruleset
, OPT_INT
, CEPH_DEFAULT_CRUSH_REPLICATED_RULESET
)
686 OPTION(osd_pool_erasure_code_stripe_unit
, OPT_U32
, 4096) // in bytes
687 OPTION(osd_pool_default_size
, OPT_INT
, 3)
688 OPTION(osd_pool_default_min_size
, OPT_INT
, 0) // 0 means no specific default; ceph will use size-size/2
689 OPTION(osd_pool_default_pg_num
, OPT_INT
, 8) // number of PGs for new pools. Configure in global or mon section of ceph.conf
690 OPTION(osd_pool_default_pgp_num
, OPT_INT
, 8) // number of PGs for placement purposes. Should be equal to pg_num
691 OPTION(osd_pool_default_erasure_code_profile
,
694 "technique=reed_sol_van "
697 ) // default properties of osd pool create
698 OPTION(osd_erasure_code_plugins
, OPT_STR
,
701 #ifdef HAVE_BETTER_YASM_ELF64
704 ) // list of erasure code plugins
706 // Allows the "peered" state for recovery and backfill below min_size
707 OPTION(osd_allow_recovery_below_min_size
, OPT_BOOL
, true)
709 OPTION(osd_pool_default_flags
, OPT_INT
, 0) // default flags for new pools
710 OPTION(osd_pool_default_flag_hashpspool
, OPT_BOOL
, true) // use new pg hashing to prevent pool/pg overlap
711 OPTION(osd_pool_default_flag_nodelete
, OPT_BOOL
, false) // pool can't be deleted
712 OPTION(osd_pool_default_flag_nopgchange
, OPT_BOOL
, false) // pool's pg and pgp num can't be changed
713 OPTION(osd_pool_default_flag_nosizechange
, OPT_BOOL
, false) // pool's size and min size can't be changed
714 OPTION(osd_pool_default_hit_set_bloom_fpp
, OPT_FLOAT
, .05)
715 OPTION(osd_pool_default_cache_target_dirty_ratio
, OPT_FLOAT
, .4)
716 OPTION(osd_pool_default_cache_target_dirty_high_ratio
, OPT_FLOAT
, .6)
717 OPTION(osd_pool_default_cache_target_full_ratio
, OPT_FLOAT
, .8)
718 OPTION(osd_pool_default_cache_min_flush_age
, OPT_INT
, 0) // seconds
719 OPTION(osd_pool_default_cache_min_evict_age
, OPT_INT
, 0) // seconds
720 OPTION(osd_pool_default_cache_max_evict_check_size
, OPT_INT
, 10) // max size to check for eviction
721 OPTION(osd_hit_set_min_size
, OPT_INT
, 1000) // min target size for a HitSet
722 OPTION(osd_hit_set_max_size
, OPT_INT
, 100000) // max target size for a HitSet
723 OPTION(osd_hit_set_namespace
, OPT_STR
, ".ceph-internal") // rados namespace for hit_set tracking
725 // conservative default throttling values
726 OPTION(osd_tier_promote_max_objects_sec
, OPT_U64
, 25)
727 OPTION(osd_tier_promote_max_bytes_sec
, OPT_U64
, 5 * 1024*1024)
729 OPTION(osd_tier_default_cache_mode
, OPT_STR
, "writeback")
730 OPTION(osd_tier_default_cache_hit_set_count
, OPT_INT
, 4)
731 OPTION(osd_tier_default_cache_hit_set_period
, OPT_INT
, 1200)
732 OPTION(osd_tier_default_cache_hit_set_type
, OPT_STR
, "bloom")
733 OPTION(osd_tier_default_cache_min_read_recency_for_promote
, OPT_INT
, 1) // number of recent HitSets the object must appear in to be promoted (on read)
734 OPTION(osd_tier_default_cache_min_write_recency_for_promote
, OPT_INT
, 1) // number of recent HitSets the object must appear in to be promoted (on write)
735 OPTION(osd_tier_default_cache_hit_set_grade_decay_rate
, OPT_INT
, 20)
736 OPTION(osd_tier_default_cache_hit_set_search_last_n
, OPT_INT
, 1)
738 OPTION(osd_map_dedup
, OPT_BOOL
, true)
739 OPTION(osd_map_max_advance
, OPT_INT
, 150) // make this < cache_size!
740 OPTION(osd_map_cache_size
, OPT_INT
, 200)
741 OPTION(osd_map_message_max
, OPT_INT
, 100) // max maps per MOSDMap message
742 OPTION(osd_map_share_max_epochs
, OPT_INT
, 100) // cap on # of inc maps we send to peers, clients
743 OPTION(osd_inject_bad_map_crc_probability
, OPT_FLOAT
, 0)
744 OPTION(osd_inject_failure_on_pg_removal
, OPT_BOOL
, false)
745 // shutdown the OSD if stuatus flipping more than max_markdown_count times in recent max_markdown_period seconds
746 OPTION(osd_max_markdown_period
, OPT_INT
, 600)
747 OPTION(osd_max_markdown_count
, OPT_INT
, 5)
749 OPTION(osd_op_threads
, OPT_INT
, 2) // 0 == no threading
750 OPTION(osd_peering_wq_batch_size
, OPT_U64
, 20)
751 OPTION(osd_op_pq_max_tokens_per_priority
, OPT_U64
, 4194304)
752 OPTION(osd_op_pq_min_cost
, OPT_U64
, 65536)
753 OPTION(osd_disk_threads
, OPT_INT
, 1)
754 OPTION(osd_disk_thread_ioprio_class
, OPT_STR
, "") // rt realtime be best effort idle
755 OPTION(osd_disk_thread_ioprio_priority
, OPT_INT
, -1) // 0-7
756 OPTION(osd_recover_clone_overlap
, OPT_BOOL
, true) // preserve clone_overlap during recovery/migration
757 OPTION(osd_op_num_threads_per_shard
, OPT_INT
, 2)
758 OPTION(osd_op_num_shards
, OPT_INT
, 5)
759 OPTION(osd_op_queue
, OPT_STR
, "wpq") // PrioritzedQueue (prio), Weighted Priority Queue (wpq), or debug_random
760 OPTION(osd_op_queue_cut_off
, OPT_STR
, "low") // Min priority to go to strict queue. (low, high, debug_random)
762 OPTION(osd_ignore_stale_divergent_priors
, OPT_BOOL
, false) // do not assert on divergent_prior entries which aren't in the log and whose on-disk objects are newer
764 // Set to true for testing. Users should NOT set this.
765 // If set to true even after reading enough shards to
766 // decode the object, any error will be reported.
767 OPTION(osd_read_ec_check_for_errors
, OPT_BOOL
, false) // return error if any ec shard has an error
769 // Only use clone_overlap for recovery if there are fewer than
770 // osd_recover_clone_overlap_limit entries in the overlap set
771 OPTION(osd_recover_clone_overlap_limit
, OPT_INT
, 10)
773 OPTION(osd_backfill_scan_min
, OPT_INT
, 64)
774 OPTION(osd_backfill_scan_max
, OPT_INT
, 512)
775 OPTION(osd_op_thread_timeout
, OPT_INT
, 15)
776 OPTION(osd_op_thread_suicide_timeout
, OPT_INT
, 150)
777 OPTION(osd_recovery_thread_timeout
, OPT_INT
, 30)
778 OPTION(osd_recovery_thread_suicide_timeout
, OPT_INT
, 300)
779 OPTION(osd_recovery_sleep
, OPT_FLOAT
, 0) // seconds to sleep between recovery ops
780 OPTION(osd_snap_trim_sleep
, OPT_DOUBLE
, 0)
781 OPTION(osd_scrub_invalid_stats
, OPT_BOOL
, true)
782 OPTION(osd_remove_thread_timeout
, OPT_INT
, 60*60)
783 OPTION(osd_remove_thread_suicide_timeout
, OPT_INT
, 10*60*60)
784 OPTION(osd_command_thread_timeout
, OPT_INT
, 10*60)
785 OPTION(osd_command_thread_suicide_timeout
, OPT_INT
, 15*60)
786 OPTION(osd_heartbeat_addr
, OPT_ADDR
, entity_addr_t())
787 OPTION(osd_heartbeat_interval
, OPT_INT
, 6) // (seconds) how often we ping peers
789 // (seconds) how long before we decide a peer has failed
790 // This setting is read by the MONs and OSDs and has to be set to a equal value in both settings of the configuration
791 OPTION(osd_heartbeat_grace
, OPT_INT
, 20)
792 OPTION(osd_heartbeat_min_peers
, OPT_INT
, 10) // minimum number of peers
793 OPTION(osd_heartbeat_use_min_delay_socket
, OPT_BOOL
, false) // prio the heartbeat tcp socket and set dscp as CS6 on it if true
795 // max number of parallel snap trims/pg
796 OPTION(osd_pg_max_concurrent_snap_trims
, OPT_U64
, 2)
797 // max number of trimming pgs
798 OPTION(osd_max_trimming_pgs
, OPT_U64
, 2)
800 // minimum number of peers that must be reachable to mark ourselves
801 // back up after being wrongly marked down.
802 OPTION(osd_heartbeat_min_healthy_ratio
, OPT_FLOAT
, .33)
804 OPTION(osd_mon_heartbeat_interval
, OPT_INT
, 30) // (seconds) how often to ping monitor if no peers
805 OPTION(osd_mon_report_interval_max
, OPT_INT
, 600)
806 OPTION(osd_mon_report_interval_min
, OPT_INT
, 5) // pg stats, failures, up_thru, boot.
807 OPTION(osd_mon_report_max_in_flight
, OPT_INT
, 2) // max updates in flight
808 OPTION(osd_beacon_report_interval
, OPT_INT
, 300) // (second) how often to send beacon message to monitor
809 OPTION(osd_pg_stat_report_interval_max
, OPT_INT
, 500) // report pg stats for any given pg at least this often
810 OPTION(osd_mon_ack_timeout
, OPT_DOUBLE
, 30.0) // time out a mon if it doesn't ack stats
811 OPTION(osd_stats_ack_timeout_factor
, OPT_DOUBLE
, 2.0) // multiples of mon_ack_timeout
812 OPTION(osd_stats_ack_timeout_decay
, OPT_DOUBLE
, .9)
813 OPTION(osd_default_data_pool_replay_window
, OPT_INT
, 45)
814 OPTION(osd_preserve_trimmed_log
, OPT_BOOL
, false)
815 OPTION(osd_auto_mark_unfound_lost
, OPT_BOOL
, false)
816 OPTION(osd_recovery_delay_start
, OPT_FLOAT
, 0)
817 OPTION(osd_recovery_max_active
, OPT_U64
, 3)
818 OPTION(osd_recovery_max_single_start
, OPT_U64
, 1)
819 OPTION(osd_recovery_max_chunk
, OPT_U64
, 8<<20) // max size of push chunk
820 OPTION(osd_recovery_max_omap_entries_per_chunk
, OPT_U64
, 64000) // max number of omap entries per chunk; 0 to disable limit
821 OPTION(osd_copyfrom_max_chunk
, OPT_U64
, 8<<20) // max size of a COPYFROM chunk
822 OPTION(osd_push_per_object_cost
, OPT_U64
, 1000) // push cost per object
823 OPTION(osd_max_push_cost
, OPT_U64
, 8<<20) // max size of push message
824 OPTION(osd_max_push_objects
, OPT_U64
, 10) // max objects in single push op
825 OPTION(osd_recovery_forget_lost_objects
, OPT_BOOL
, false) // off for now
826 OPTION(osd_max_scrubs
, OPT_INT
, 1)
827 OPTION(osd_scrub_during_recovery
, OPT_BOOL
, false) // Allow new scrubs to start while recovery is active on the OSD
828 OPTION(osd_scrub_begin_hour
, OPT_INT
, 0)
829 OPTION(osd_scrub_end_hour
, OPT_INT
, 24)
830 OPTION(osd_scrub_load_threshold
, OPT_FLOAT
, 0.5)
831 OPTION(osd_scrub_min_interval
, OPT_FLOAT
, 60*60*24) // if load is low
832 OPTION(osd_scrub_max_interval
, OPT_FLOAT
, 7*60*60*24) // regardless of load
833 OPTION(osd_scrub_interval_randomize_ratio
, OPT_FLOAT
, 0.5) // randomize the scheduled scrub in the span of [min,min*(1+randomize_ratio))
834 OPTION(osd_scrub_backoff_ratio
, OPT_DOUBLE
, .66) // the probability to back off the scheduled scrub
835 OPTION(osd_scrub_chunk_min
, OPT_INT
, 5)
836 OPTION(osd_scrub_chunk_max
, OPT_INT
, 25)
837 OPTION(osd_scrub_sleep
, OPT_FLOAT
, 0) // sleep between [deep]scrub ops
838 OPTION(osd_scrub_auto_repair
, OPT_BOOL
, false) // whether auto-repair inconsistencies upon deep-scrubbing
839 OPTION(osd_scrub_auto_repair_num_errors
, OPT_U32
, 5) // only auto-repair when number of errors is below this threshold
840 OPTION(osd_deep_scrub_interval
, OPT_FLOAT
, 60*60*24*7) // once a week
841 OPTION(osd_deep_scrub_randomize_ratio
, OPT_FLOAT
, 0.15) // scrubs will randomly become deep scrubs at this rate (0.15 -> 15% of scrubs are deep)
842 OPTION(osd_deep_scrub_stride
, OPT_INT
, 524288)
843 OPTION(osd_deep_scrub_update_digest_min_age
, OPT_INT
, 2*60*60) // objects must be this old (seconds) before we update the whole-object digest on scrub
844 OPTION(osd_scan_list_ping_tp_interval
, OPT_U64
, 100)
845 OPTION(osd_class_dir
, OPT_STR
, CEPH_LIBDIR
"/rados-classes") // where rados plugins are stored
846 OPTION(osd_open_classes_on_start
, OPT_BOOL
, true)
847 OPTION(osd_class_load_list
, OPT_STR
, "cephfs hello journal lock log numops "
848 "rbd refcount replica_log rgw statelog timeindex user version") // list of object classes allowed to be loaded (allow all: *)
849 OPTION(osd_class_default_list
, OPT_STR
, "cephfs hello journal lock log numops "
850 "rbd refcount replica_log rgw statelog timeindex user version") // list of object classes with default execute perm (allow all: *)
851 OPTION(osd_check_for_log_corruption
, OPT_BOOL
, false)
852 OPTION(osd_use_stale_snap
, OPT_BOOL
, false)
853 OPTION(osd_rollback_to_cluster_snap
, OPT_STR
, "")
854 OPTION(osd_default_notify_timeout
, OPT_U32
, 30) // default notify timeout in seconds
855 OPTION(osd_kill_backfill_at
, OPT_INT
, 0)
857 // Bounds how infrequently a new map epoch will be persisted for a pg
858 OPTION(osd_pg_epoch_persisted_max_stale
, OPT_U32
, 150) // make this < map_cache_size!
860 OPTION(osd_min_pg_log_entries
, OPT_U32
, 3000) // number of entries to keep in the pg log when trimming it
861 OPTION(osd_max_pg_log_entries
, OPT_U32
, 10000) // max entries, say when degraded, before we trim
862 OPTION(osd_force_recovery_pg_log_entries_factor
, OPT_FLOAT
, 1.3) // max entries factor before force recovery
863 OPTION(osd_pg_log_trim_min
, OPT_U32
, 100)
864 OPTION(osd_op_complaint_time
, OPT_FLOAT
, 30) // how many seconds old makes an op complaint-worthy
865 OPTION(osd_command_max_records
, OPT_INT
, 256)
866 OPTION(osd_max_pg_blocked_by
, OPT_U32
, 16) // max peer osds to report that are blocking our progress
867 OPTION(osd_op_log_threshold
, OPT_INT
, 5) // how many op log messages to show in one go
868 OPTION(osd_verify_sparse_read_holes
, OPT_BOOL
, false) // read fiemap-reported holes and verify they are zeros
869 OPTION(osd_backoff_on_unfound
, OPT_BOOL
, true) // object unfound
870 OPTION(osd_backoff_on_degraded
, OPT_BOOL
, false) // [mainly for debug?] object unreadable/writeable
871 OPTION(osd_backoff_on_down
, OPT_BOOL
, true) // pg in down/incomplete state
872 OPTION(osd_backoff_on_peering
, OPT_BOOL
, false) // [debug] pg peering
873 OPTION(osd_debug_crash_on_ignored_backoff
, OPT_BOOL
, false) // crash osd if client ignores a backoff; useful for debugging
874 OPTION(osd_debug_inject_dispatch_delay_probability
, OPT_DOUBLE
, 0)
875 OPTION(osd_debug_inject_dispatch_delay_duration
, OPT_DOUBLE
, .1)
876 OPTION(osd_debug_drop_ping_probability
, OPT_DOUBLE
, 0)
877 OPTION(osd_debug_drop_ping_duration
, OPT_INT
, 0)
878 OPTION(osd_debug_op_order
, OPT_BOOL
, false)
879 OPTION(osd_debug_verify_missing_on_start
, OPT_BOOL
, false)
880 OPTION(osd_debug_scrub_chance_rewrite_digest
, OPT_U64
, 0)
881 OPTION(osd_debug_verify_snaps_on_info
, OPT_BOOL
, false)
882 OPTION(osd_debug_verify_stray_on_activate
, OPT_BOOL
, false)
883 OPTION(osd_debug_skip_full_check_in_backfill_reservation
, OPT_BOOL
, false)
884 OPTION(osd_debug_reject_backfill_probability
, OPT_DOUBLE
, 0)
885 OPTION(osd_debug_inject_copyfrom_error
, OPT_BOOL
, false) // inject failure during copyfrom completion
886 OPTION(osd_debug_misdirected_ops
, OPT_BOOL
, false)
887 OPTION(osd_debug_skip_full_check_in_recovery
, OPT_BOOL
, false)
888 OPTION(osd_enxio_on_misdirected_op
, OPT_BOOL
, false)
889 OPTION(osd_debug_verify_cached_snaps
, OPT_BOOL
, false)
890 OPTION(osd_enable_op_tracker
, OPT_BOOL
, true) // enable/disable OSD op tracking
891 OPTION(osd_num_op_tracker_shard
, OPT_U32
, 32) // The number of shards for holding the ops
892 OPTION(osd_op_history_size
, OPT_U32
, 20) // Max number of completed ops to track
893 OPTION(osd_op_history_duration
, OPT_U32
, 600) // Oldest completed op to track
894 OPTION(osd_op_history_slow_op_size
, OPT_U32
, 20) // Max number of slow ops to track
895 OPTION(osd_op_history_slow_op_threshold
, OPT_DOUBLE
, 10.0) // track the op if over this threshold
896 OPTION(osd_target_transaction_size
, OPT_INT
, 30) // to adjust various transactions that batch smaller items
897 OPTION(osd_failsafe_full_ratio
, OPT_FLOAT
, .97) // what % full makes an OSD "full" (failsafe)
898 OPTION(osd_fast_fail_on_connection_refused
, OPT_BOOL
, true) // immediately mark OSDs as down once they refuse to accept connections
900 OPTION(osd_pg_object_context_cache_count
, OPT_INT
, 64)
901 OPTION(osd_tracing
, OPT_BOOL
, false) // true if LTTng-UST tracepoints should be enabled
903 OPTION(osd_fast_info
, OPT_BOOL
, true) // use fast info attr, if we can
905 // determines whether PGLog::check() compares written out log to stored log
906 OPTION(osd_debug_pg_log_writeout
, OPT_BOOL
, false)
907 OPTION(osd_loop_before_reset_tphandle
, OPT_U32
, 64) // Max number of loop before we reset thread-pool's handle
908 // default timeout while caling WaitInterval on an empty queue
909 OPTION(threadpool_default_timeout
, OPT_INT
, 60)
910 // default wait time for an empty queue before pinging the hb timeout
911 OPTION(threadpool_empty_queue_max_wait
, OPT_INT
, 2)
913 OPTION(leveldb_log_to_ceph_log
, OPT_BOOL
, true)
914 OPTION(leveldb_write_buffer_size
, OPT_U64
, 8 *1024*1024) // leveldb write buffer size
915 OPTION(leveldb_cache_size
, OPT_U64
, 128 *1024*1024) // leveldb cache size
916 OPTION(leveldb_block_size
, OPT_U64
, 0) // leveldb block size
917 OPTION(leveldb_bloom_size
, OPT_INT
, 0) // leveldb bloom bits per entry
918 OPTION(leveldb_max_open_files
, OPT_INT
, 0) // leveldb max open files
919 OPTION(leveldb_compression
, OPT_BOOL
, true) // leveldb uses compression
920 OPTION(leveldb_paranoid
, OPT_BOOL
, false) // leveldb paranoid flag
921 OPTION(leveldb_log
, OPT_STR
, "/dev/null") // enable leveldb log file
922 OPTION(leveldb_compact_on_mount
, OPT_BOOL
, false)
924 OPTION(kinetic_host
, OPT_STR
, "") // hostname or ip address of a kinetic drive to use
925 OPTION(kinetic_port
, OPT_INT
, 8123) // port number of the kinetic drive
926 OPTION(kinetic_user_id
, OPT_INT
, 1) // kinetic user to authenticate as
927 OPTION(kinetic_hmac_key
, OPT_STR
, "asdfasdf") // kinetic key to authenticate with
928 OPTION(kinetic_use_ssl
, OPT_BOOL
, false) // whether to secure kinetic traffic with TLS
931 OPTION(rocksdb_separate_wal_dir
, OPT_BOOL
, false) // use $path.wal for wal
932 SAFE_OPTION(rocksdb_db_paths
, OPT_STR
, "") // path,size( path,size)*
933 OPTION(rocksdb_log_to_ceph_log
, OPT_BOOL
, true) // log to ceph log
934 OPTION(rocksdb_cache_size
, OPT_U64
, 128*1024*1024) // default rocksdb cache size
935 OPTION(rocksdb_cache_shard_bits
, OPT_INT
, 4) // rocksdb block cache shard bits, 4 bit -> 16 shards
936 OPTION(rocksdb_block_size
, OPT_INT
, 4*1024) // default rocksdb block size
937 OPTION(rocksdb_perf
, OPT_BOOL
, false) // Enabling this will have 5-10% impact on performance for the stats collection
938 OPTION(rocksdb_collect_compaction_stats
, OPT_BOOL
, false) //For rocksdb, this behavior will be an overhead of 5%~10%, collected only rocksdb_perf is enabled.
939 OPTION(rocksdb_collect_extended_stats
, OPT_BOOL
, false) //For rocksdb, this behavior will be an overhead of 5%~10%, collected only rocksdb_perf is enabled.
940 OPTION(rocksdb_collect_memory_stats
, OPT_BOOL
, false) //For rocksdb, this behavior will be an overhead of 5%~10%, collected only rocksdb_perf is enabled.
941 OPTION(rocksdb_enable_rmrange
, OPT_BOOL
, false) // see https://github.com/facebook/rocksdb/blob/master/include/rocksdb/db.h#L253
943 // rocksdb options that will be used for omap(if omap_backend is rocksdb)
944 OPTION(filestore_rocksdb_options
, OPT_STR
, "")
945 // rocksdb options that will be used in monstore
946 OPTION(mon_rocksdb_options
, OPT_STR
, "write_buffer_size=33554432,compression=kNoCompression")
949 * osd_*_priority adjust the relative priority of client io, recovery io,
952 * osd_*_priority determines the ratio of available io between client and
953 * recovery. Each option may be set between
956 OPTION(osd_client_op_priority
, OPT_U32
, 63)
957 OPTION(osd_recovery_op_priority
, OPT_U32
, 3)
959 OPTION(osd_snap_trim_priority
, OPT_U32
, 5)
960 OPTION(osd_snap_trim_cost
, OPT_U32
, 1<<20) // set default cost equal to 1MB io
962 OPTION(osd_scrub_priority
, OPT_U32
, 5)
963 // set default cost equal to 50MB io
964 OPTION(osd_scrub_cost
, OPT_U32
, 50<<20)
965 // set requested scrub priority higher than scrub priority to make the
966 // requested scrubs jump the queue of scheduled scrubs
967 OPTION(osd_requested_scrub_priority
, OPT_U32
, 120)
969 OPTION(osd_recovery_priority
, OPT_U32
, 5)
970 // set default cost equal to 20MB io
971 OPTION(osd_recovery_cost
, OPT_U32
, 20<<20)
974 * osd_recovery_op_warn_multiple scales the normal warning threshhold,
975 * osd_op_complaint_time, so that slow recovery ops won't cause noise
977 OPTION(osd_recovery_op_warn_multiple
, OPT_U32
, 16)
979 // Max time to wait between notifying mon of shutdown and shutting down
980 OPTION(osd_mon_shutdown_timeout
, OPT_DOUBLE
, 5)
982 OPTION(osd_max_object_size
, OPT_U64
, 100*1024L*1024L*1024L) // OSD's maximum object size
983 OPTION(osd_max_object_name_len
, OPT_U32
, 2048) // max rados object name len
984 OPTION(osd_max_object_namespace_len
, OPT_U32
, 256) // max rados object namespace len
985 OPTION(osd_max_attr_name_len
, OPT_U32
, 100) // max rados attr name len; cannot go higher than 100 chars for file system backends
986 OPTION(osd_max_attr_size
, OPT_U64
, 0)
988 OPTION(osd_max_omap_entries_per_request
, OPT_U64
, 131072)
989 OPTION(osd_max_omap_bytes_per_request
, OPT_U64
, 1<<30)
991 OPTION(osd_objectstore
, OPT_STR
, "filestore") // ObjectStore backend type
992 OPTION(osd_objectstore_tracing
, OPT_BOOL
, false) // true if LTTng-UST tracepoints should be enabled
993 // Override maintaining compatibility with older OSDs
994 // Set to true for testing. Users should NOT set this.
995 OPTION(osd_debug_override_acting_compat
, OPT_BOOL
, false)
996 OPTION(osd_objectstore_fuse
, OPT_BOOL
, false)
998 OPTION(osd_bench_small_size_max_iops
, OPT_U32
, 100) // 100 IOPS
999 OPTION(osd_bench_large_size_max_throughput
, OPT_U64
, 100 << 20) // 100 MB/s
1000 OPTION(osd_bench_max_block_size
, OPT_U64
, 64 << 20) // cap the block size at 64MB
1001 OPTION(osd_bench_duration
, OPT_U32
, 30) // duration of 'osd bench', capped at 30s to avoid triggering timeouts
1003 OPTION(osd_blkin_trace_all
, OPT_BOOL
, false) // create a blkin trace for all osd requests
1004 OPTION(osdc_blkin_trace_all
, OPT_BOOL
, false) // create a blkin trace for all objecter requests
1006 OPTION(osd_discard_disconnected_ops
, OPT_BOOL
, true)
1008 OPTION(memstore_device_bytes
, OPT_U64
, 1024*1024*1024)
1009 OPTION(memstore_page_set
, OPT_BOOL
, true)
1010 OPTION(memstore_page_size
, OPT_U64
, 64 << 10)
1012 OPTION(bdev_debug_inflight_ios
, OPT_BOOL
, false)
1013 OPTION(bdev_inject_crash
, OPT_INT
, 0) // if N>0, then ~ 1/N IOs will complete before we crash on flush.
1014 OPTION(bdev_inject_crash_flush_delay
, OPT_INT
, 2) // wait N more seconds on flush
1015 OPTION(bdev_aio
, OPT_BOOL
, true)
1016 OPTION(bdev_aio_poll_ms
, OPT_INT
, 250) // milliseconds
1017 OPTION(bdev_aio_max_queue_depth
, OPT_INT
, 1024)
1018 OPTION(bdev_block_size
, OPT_INT
, 4096)
1019 OPTION(bdev_debug_aio
, OPT_BOOL
, false)
1020 OPTION(bdev_debug_aio_suicide_timeout
, OPT_FLOAT
, 60.0)
1022 // if yes, osd will unbind all NVMe devices from kernel driver and bind them
1023 // to the uio_pci_generic driver. The purpose is to prevent the case where
1024 // NVMe driver is loaded while osd is running.
1025 OPTION(bdev_nvme_unbind_from_kernel
, OPT_BOOL
, false)
1026 OPTION(bdev_nvme_retry_count
, OPT_INT
, -1) // -1 means by default which is 4
1028 OPTION(objectstore_blackhole
, OPT_BOOL
, false)
1030 OPTION(bluefs_alloc_size
, OPT_U64
, 1048576)
1031 OPTION(bluefs_max_prefetch
, OPT_U64
, 1048576)
1032 OPTION(bluefs_min_log_runway
, OPT_U64
, 1048576) // alloc when we get this low
1033 OPTION(bluefs_max_log_runway
, OPT_U64
, 4194304) // alloc this much at a time
1034 OPTION(bluefs_log_compact_min_ratio
, OPT_FLOAT
, 5.0) // before we consider
1035 OPTION(bluefs_log_compact_min_size
, OPT_U64
, 16*1048576) // before we consider
1036 OPTION(bluefs_min_flush_size
, OPT_U64
, 524288) // ignore flush until its this big
1037 OPTION(bluefs_compact_log_sync
, OPT_BOOL
, false) // sync or async log compaction?
1038 OPTION(bluefs_buffered_io
, OPT_BOOL
, false)
1039 OPTION(bluefs_sync_write
, OPT_BOOL
, false)
1040 OPTION(bluefs_allocator
, OPT_STR
, "bitmap") // stupid | bitmap
1041 OPTION(bluefs_preextend_wal_files
, OPT_BOOL
, false) // this *requires* that rocksdb has recycling enabled
1043 OPTION(bluestore_bluefs
, OPT_BOOL
, true)
1044 OPTION(bluestore_bluefs_env_mirror
, OPT_BOOL
, false) // mirror to normal Env for debug
1045 OPTION(bluestore_bluefs_min
, OPT_U64
, 1*1024*1024*1024) // 1gb
1046 OPTION(bluestore_bluefs_min_ratio
, OPT_FLOAT
, .02) // min fs free / total free
1047 OPTION(bluestore_bluefs_max_ratio
, OPT_FLOAT
, .90) // max fs free / total free
1048 OPTION(bluestore_bluefs_gift_ratio
, OPT_FLOAT
, .02) // how much to add at a time
1049 OPTION(bluestore_bluefs_reclaim_ratio
, OPT_FLOAT
, .20) // how much to reclaim at a time
1050 OPTION(bluestore_bluefs_balance_interval
, OPT_FLOAT
, 1) // how often (sec) to balance free space between bluefs and bluestore
1051 // If you want to use spdk driver, you need to specify NVMe serial number here
1052 // with "spdk:" prefix.
1053 // Users can use 'lspci -vvv -d 8086:0953 | grep "Device Serial Number"' to
1054 // get the serial number of Intel(R) Fultondale NVMe controllers.
1056 // bluestore_block_path = spdk:55cd2e404bd73932
1057 // If you want to run multiple SPDK instances per node, you must specify the
1058 // amount of dpdk memory size in MB each instance will use, to make sure each
1059 // instance uses its own dpdk memory
1060 OPTION(bluestore_spdk_mem
, OPT_U32
, 512)
1061 // A hexadecimal bit mask of the cores to run on. Note the core numbering can change between platforms and should be determined beforehand.
1062 OPTION(bluestore_spdk_coremask
, OPT_STR
, "0x3")
1063 // Specify the maximal I/Os to be batched completed while checking queue pair completions.
1064 // Default value 0 means that let SPDK nvme library determine the value.
1065 OPTION(bluestore_spdk_max_io_completion
, OPT_U32
, 0)
1066 OPTION(bluestore_block_path
, OPT_STR
, "")
1067 OPTION(bluestore_block_size
, OPT_U64
, 10 * 1024*1024*1024) // 10gb for testing
1068 OPTION(bluestore_block_create
, OPT_BOOL
, true)
1069 OPTION(bluestore_block_db_path
, OPT_STR
, "")
1070 OPTION(bluestore_block_db_size
, OPT_U64
, 0) // rocksdb ssts (hot/warm)
1071 OPTION(bluestore_block_db_create
, OPT_BOOL
, false)
1072 OPTION(bluestore_block_wal_path
, OPT_STR
, "")
1073 OPTION(bluestore_block_wal_size
, OPT_U64
, 96 * 1024*1024) // rocksdb wal
1074 OPTION(bluestore_block_wal_create
, OPT_BOOL
, false)
1075 OPTION(bluestore_block_preallocate_file
, OPT_BOOL
, false) //whether preallocate space if block/db_path/wal_path is file rather that block device.
1076 OPTION(bluestore_csum_type
, OPT_STR
, "crc32c") // none|xxhash32|xxhash64|crc32c|crc32c_16|crc32c_8
1077 OPTION(bluestore_csum_min_block
, OPT_U32
, 4096)
1078 OPTION(bluestore_csum_max_block
, OPT_U32
, 64*1024)
1079 OPTION(bluestore_min_alloc_size
, OPT_U32
, 0)
1080 OPTION(bluestore_min_alloc_size_hdd
, OPT_U32
, 64*1024)
1081 OPTION(bluestore_min_alloc_size_ssd
, OPT_U32
, 16*1024)
1082 OPTION(bluestore_max_alloc_size
, OPT_U32
, 0)
1083 OPTION(bluestore_prefer_deferred_size
, OPT_U32
, 0)
1084 OPTION(bluestore_prefer_deferred_size_hdd
, OPT_U32
, 32768)
1085 OPTION(bluestore_prefer_deferred_size_ssd
, OPT_U32
, 0)
1086 OPTION(bluestore_compression_mode
, OPT_STR
, "none") // force|aggressive|passive|none
1087 OPTION(bluestore_compression_algorithm
, OPT_STR
, "snappy")
1088 OPTION(bluestore_compression_min_blob_size
, OPT_U32
, 0)
1089 OPTION(bluestore_compression_min_blob_size_hdd
, OPT_U32
, 128*1024)
1090 OPTION(bluestore_compression_min_blob_size_ssd
, OPT_U32
, 8*1024)
1091 OPTION(bluestore_compression_max_blob_size
, OPT_U32
, 0)
1092 OPTION(bluestore_compression_max_blob_size_hdd
, OPT_U32
, 512*1024)
1093 OPTION(bluestore_compression_max_blob_size_ssd
, OPT_U32
, 64*1024)
1095 * Specifies minimum expected amount of saved allocation units
1096 * per single blob to enable compressed blobs garbage collection
1099 OPTION(bluestore_gc_enable_blob_threshold
, OPT_INT
, 0)
1101 * Specifies minimum expected amount of saved allocation units
1102 * per all blobsb to enable compressed blobs garbage collection
1105 OPTION(bluestore_gc_enable_total_threshold
, OPT_INT
, 0)
1107 OPTION(bluestore_max_blob_size
, OPT_U32
, 0)
1108 OPTION(bluestore_max_blob_size_hdd
, OPT_U32
, 512*1024)
1109 OPTION(bluestore_max_blob_size_ssd
, OPT_U32
, 64*1024)
1111 * Require the net gain of compression at least to be at this ratio,
1112 * otherwise we don't compress.
1113 * And ask for compressing at least 12.5%(1/8) off, by default.
1115 OPTION(bluestore_compression_required_ratio
, OPT_DOUBLE
, .875)
1116 OPTION(bluestore_extent_map_shard_max_size
, OPT_U32
, 1200)
1117 OPTION(bluestore_extent_map_shard_target_size
, OPT_U32
, 500)
1118 OPTION(bluestore_extent_map_shard_min_size
, OPT_U32
, 150)
1119 OPTION(bluestore_extent_map_shard_target_size_slop
, OPT_DOUBLE
, .2)
1120 OPTION(bluestore_extent_map_inline_shard_prealloc_size
, OPT_U32
, 256)
1121 OPTION(bluestore_cache_trim_interval
, OPT_DOUBLE
, .1)
1122 OPTION(bluestore_cache_trim_max_skip_pinned
, OPT_U32
, 64) // skip this many onodes pinned in cache before we give up
1123 OPTION(bluestore_cache_type
, OPT_STR
, "2q") // lru, 2q
1124 OPTION(bluestore_2q_cache_kin_ratio
, OPT_DOUBLE
, .5) // kin page slot size / max page slot size
1125 OPTION(bluestore_2q_cache_kout_ratio
, OPT_DOUBLE
, .5) // number of kout page slot / total number of page slot
1126 OPTION(bluestore_cache_size
, OPT_U64
, 1024*1024*1024)
1127 OPTION(bluestore_cache_meta_ratio
, OPT_DOUBLE
, .9)
1128 OPTION(bluestore_kvbackend
, OPT_STR
, "rocksdb")
1129 OPTION(bluestore_allocator
, OPT_STR
, "bitmap") // stupid | bitmap
1130 OPTION(bluestore_freelist_blocks_per_key
, OPT_INT
, 128)
1131 OPTION(bluestore_bitmapallocator_blocks_per_zone
, OPT_INT
, 1024) // must be power of 2 aligned, e.g., 512, 1024, 2048...
1132 OPTION(bluestore_bitmapallocator_span_size
, OPT_INT
, 1024) // must be power of 2 aligned, e.g., 512, 1024, 2048...
1133 OPTION(bluestore_rocksdb_options
, OPT_STR
, "compression=kNoCompression,max_write_buffer_number=4,min_write_buffer_number_to_merge=1,recycle_log_file_num=4,write_buffer_size=268435456,writable_file_max_buffer_size=0,compaction_readahead_size=2097152")
1134 OPTION(bluestore_fsck_on_mount
, OPT_BOOL
, false)
1135 OPTION(bluestore_fsck_on_mount_deep
, OPT_BOOL
, true)
1136 OPTION(bluestore_fsck_on_umount
, OPT_BOOL
, false)
1137 OPTION(bluestore_fsck_on_umount_deep
, OPT_BOOL
, true)
1138 OPTION(bluestore_fsck_on_mkfs
, OPT_BOOL
, true)
1139 OPTION(bluestore_fsck_on_mkfs_deep
, OPT_BOOL
, false)
1140 OPTION(bluestore_sync_submit_transaction
, OPT_BOOL
, false) // submit kv txn in queueing thread (not kv_sync_thread)
1141 OPTION(bluestore_throttle_bytes
, OPT_U64
, 64*1024*1024)
1142 OPTION(bluestore_throttle_deferred_bytes
, OPT_U64
, 128*1024*1024)
1143 OPTION(bluestore_throttle_cost_per_io_hdd
, OPT_U64
, 1500000)
1144 OPTION(bluestore_throttle_cost_per_io_ssd
, OPT_U64
, 4000)
1145 OPTION(bluestore_throttle_cost_per_io
, OPT_U64
, 0)
1146 OPTION(bluestore_deferred_batch_ops
, OPT_U64
, 0)
1147 OPTION(bluestore_deferred_batch_ops_hdd
, OPT_U64
, 64)
1148 OPTION(bluestore_deferred_batch_ops_ssd
, OPT_U64
, 16)
1149 OPTION(bluestore_nid_prealloc
, OPT_INT
, 1024)
1150 OPTION(bluestore_blobid_prealloc
, OPT_U64
, 10240)
1151 OPTION(bluestore_clone_cow
, OPT_BOOL
, true) // do copy-on-write for clones
1152 OPTION(bluestore_default_buffered_read
, OPT_BOOL
, true)
1153 OPTION(bluestore_default_buffered_write
, OPT_BOOL
, false)
1154 OPTION(bluestore_debug_misc
, OPT_BOOL
, false)
1155 OPTION(bluestore_debug_no_reuse_blocks
, OPT_BOOL
, false)
1156 OPTION(bluestore_debug_small_allocations
, OPT_INT
, 0)
1157 OPTION(bluestore_debug_freelist
, OPT_BOOL
, false)
1158 OPTION(bluestore_debug_prefill
, OPT_FLOAT
, 0)
1159 OPTION(bluestore_debug_prefragment_max
, OPT_INT
, 1048576)
1160 OPTION(bluestore_debug_inject_read_err
, OPT_BOOL
, false)
1161 OPTION(bluestore_debug_randomize_serial_transaction
, OPT_INT
, 0)
1162 OPTION(bluestore_debug_omit_block_device_write
, OPT_BOOL
, false)
1163 OPTION(bluestore_shard_finishers
, OPT_BOOL
, false)
1165 OPTION(kstore_max_ops
, OPT_U64
, 512)
1166 OPTION(kstore_max_bytes
, OPT_U64
, 64*1024*1024)
1167 OPTION(kstore_backend
, OPT_STR
, "rocksdb")
1168 OPTION(kstore_rocksdb_options
, OPT_STR
, "compression=kNoCompression")
1169 OPTION(kstore_rocksdb_bloom_bits_per_key
, OPT_INT
, 0)
1170 OPTION(kstore_fsck_on_mount
, OPT_BOOL
, false)
1171 OPTION(kstore_fsck_on_mount_deep
, OPT_BOOL
, true)
1172 OPTION(kstore_nid_prealloc
, OPT_U64
, 1024)
1173 OPTION(kstore_sync_transaction
, OPT_BOOL
, false)
1174 OPTION(kstore_sync_submit_transaction
, OPT_BOOL
, false)
1175 OPTION(kstore_onode_map_size
, OPT_U64
, 1024)
1176 OPTION(kstore_cache_tails
, OPT_BOOL
, true)
1177 OPTION(kstore_default_stripe_size
, OPT_INT
, 65536)
1179 OPTION(filestore_omap_backend
, OPT_STR
, "rocksdb")
1180 OPTION(filestore_omap_backend_path
, OPT_STR
, "")
1182 /// filestore wb throttle limits
1183 OPTION(filestore_wbthrottle_enable
, OPT_BOOL
, true)
1184 OPTION(filestore_wbthrottle_btrfs_bytes_start_flusher
, OPT_U64
, 41943040)
1185 OPTION(filestore_wbthrottle_btrfs_bytes_hard_limit
, OPT_U64
, 419430400)
1186 OPTION(filestore_wbthrottle_btrfs_ios_start_flusher
, OPT_U64
, 500)
1187 OPTION(filestore_wbthrottle_btrfs_ios_hard_limit
, OPT_U64
, 5000)
1188 OPTION(filestore_wbthrottle_btrfs_inodes_start_flusher
, OPT_U64
, 500)
1189 OPTION(filestore_wbthrottle_xfs_bytes_start_flusher
, OPT_U64
, 41943040)
1190 OPTION(filestore_wbthrottle_xfs_bytes_hard_limit
, OPT_U64
, 419430400)
1191 OPTION(filestore_wbthrottle_xfs_ios_start_flusher
, OPT_U64
, 500)
1192 OPTION(filestore_wbthrottle_xfs_ios_hard_limit
, OPT_U64
, 5000)
1193 OPTION(filestore_wbthrottle_xfs_inodes_start_flusher
, OPT_U64
, 500)
1195 /// These must be less than the fd limit
1196 OPTION(filestore_wbthrottle_btrfs_inodes_hard_limit
, OPT_U64
, 5000)
1197 OPTION(filestore_wbthrottle_xfs_inodes_hard_limit
, OPT_U64
, 5000)
1199 //Introduce a O_DSYNC write in the filestore
1200 OPTION(filestore_odsync_write
, OPT_BOOL
, false)
1202 // Tests index failure paths
1203 OPTION(filestore_index_retry_probability
, OPT_DOUBLE
, 0)
1205 // Allow object read error injection
1206 OPTION(filestore_debug_inject_read_err
, OPT_BOOL
, false)
1208 OPTION(filestore_debug_omap_check
, OPT_BOOL
, 0) // Expensive debugging check on sync
1209 OPTION(filestore_omap_header_cache_size
, OPT_INT
, 1024)
1211 // Use omap for xattrs for attrs over
1212 // filestore_max_inline_xattr_size or
1213 OPTION(filestore_max_inline_xattr_size
, OPT_U32
, 0) //Override
1214 OPTION(filestore_max_inline_xattr_size_xfs
, OPT_U32
, 65536)
1215 OPTION(filestore_max_inline_xattr_size_btrfs
, OPT_U32
, 2048)
1216 OPTION(filestore_max_inline_xattr_size_other
, OPT_U32
, 512)
1218 // for more than filestore_max_inline_xattrs attrs
1219 OPTION(filestore_max_inline_xattrs
, OPT_U32
, 0) //Override
1220 OPTION(filestore_max_inline_xattrs_xfs
, OPT_U32
, 10)
1221 OPTION(filestore_max_inline_xattrs_btrfs
, OPT_U32
, 10)
1222 OPTION(filestore_max_inline_xattrs_other
, OPT_U32
, 2)
1224 // max xattr value size
1225 OPTION(filestore_max_xattr_value_size
, OPT_U32
, 0) //Override
1226 OPTION(filestore_max_xattr_value_size_xfs
, OPT_U32
, 64<<10)
1227 OPTION(filestore_max_xattr_value_size_btrfs
, OPT_U32
, 64<<10)
1228 // ext4 allows 4k xattrs total including some smallish extra fields and the
1229 // keys. We're allowing 2 512 inline attrs in addition some some filestore
1230 // replay attrs. After accounting for those, we still need to fit up to
1231 // two attrs of this value. That means we need this value to be around 1k
1232 // to be safe. This is hacky, but it's not worth complicating the code
1233 // to work around ext4's total xattr limit.
1234 OPTION(filestore_max_xattr_value_size_other
, OPT_U32
, 1<<10)
1236 OPTION(filestore_sloppy_crc
, OPT_BOOL
, false) // track sloppy crcs
1237 OPTION(filestore_sloppy_crc_block_size
, OPT_INT
, 65536)
1239 OPTION(filestore_max_alloc_hint_size
, OPT_U64
, 1ULL << 20) // bytes
1241 OPTION(filestore_max_sync_interval
, OPT_DOUBLE
, 5) // seconds
1242 OPTION(filestore_min_sync_interval
, OPT_DOUBLE
, .01) // seconds
1243 OPTION(filestore_btrfs_snap
, OPT_BOOL
, true)
1244 OPTION(filestore_btrfs_clone_range
, OPT_BOOL
, true)
1245 OPTION(filestore_zfs_snap
, OPT_BOOL
, false) // zfsonlinux is still unstable
1246 OPTION(filestore_fsync_flushes_journal_data
, OPT_BOOL
, false)
1247 OPTION(filestore_fiemap
, OPT_BOOL
, false) // (try to) use fiemap
1248 OPTION(filestore_punch_hole
, OPT_BOOL
, false)
1249 OPTION(filestore_seek_data_hole
, OPT_BOOL
, false) // (try to) use seek_data/hole
1250 OPTION(filestore_splice
, OPT_BOOL
, false)
1251 OPTION(filestore_fadvise
, OPT_BOOL
, true)
1252 //collect device partition information for management application to use
1253 OPTION(filestore_collect_device_partition_information
, OPT_BOOL
, true)
1255 // (try to) use extsize for alloc hint NOTE: extsize seems to trigger
1256 // data corruption in xfs prior to kernel 3.5. filestore will
1257 // implicity disable this if it cannot confirm the kernel is newer
1259 // NOTE: This option involves a tradeoff: When disabled, fragmentation is
1260 // worse, but large sequential writes are faster. When enabled, large
1261 // sequential writes are slower, but fragmentation is reduced.
1262 OPTION(filestore_xfs_extsize
, OPT_BOOL
, false)
1264 OPTION(filestore_journal_parallel
, OPT_BOOL
, false)
1265 OPTION(filestore_journal_writeahead
, OPT_BOOL
, false)
1266 OPTION(filestore_journal_trailing
, OPT_BOOL
, false)
1267 OPTION(filestore_queue_max_ops
, OPT_U64
, 50)
1268 OPTION(filestore_queue_max_bytes
, OPT_U64
, 100 << 20)
1270 OPTION(filestore_caller_concurrency
, OPT_INT
, 10)
1272 /// Expected filestore throughput in B/s
1273 OPTION(filestore_expected_throughput_bytes
, OPT_DOUBLE
, 200 << 20)
1274 /// Expected filestore throughput in ops/s
1275 OPTION(filestore_expected_throughput_ops
, OPT_DOUBLE
, 200)
1277 /// Filestore max delay multiple. Defaults to 0 (disabled)
1278 OPTION(filestore_queue_max_delay_multiple
, OPT_DOUBLE
, 0)
1279 /// Filestore high delay multiple. Defaults to 0 (disabled)
1280 OPTION(filestore_queue_high_delay_multiple
, OPT_DOUBLE
, 0)
1282 /// Use above to inject delays intended to keep the op queue between low and high
1283 OPTION(filestore_queue_low_threshhold
, OPT_DOUBLE
, 0.3)
1284 OPTION(filestore_queue_high_threshhold
, OPT_DOUBLE
, 0.9)
1286 OPTION(filestore_op_threads
, OPT_INT
, 2)
1287 OPTION(filestore_op_thread_timeout
, OPT_INT
, 60)
1288 OPTION(filestore_op_thread_suicide_timeout
, OPT_INT
, 180)
1289 OPTION(filestore_commit_timeout
, OPT_FLOAT
, 600)
1290 OPTION(filestore_fiemap_threshold
, OPT_INT
, 4096)
1291 OPTION(filestore_merge_threshold
, OPT_INT
, 10)
1292 OPTION(filestore_split_multiple
, OPT_INT
, 2)
1293 OPTION(filestore_update_to
, OPT_INT
, 1000)
1294 OPTION(filestore_blackhole
, OPT_BOOL
, false) // drop any new transactions on the floor
1295 OPTION(filestore_fd_cache_size
, OPT_INT
, 128) // FD lru size
1296 OPTION(filestore_fd_cache_shards
, OPT_INT
, 16) // FD number of shards
1297 OPTION(filestore_ondisk_finisher_threads
, OPT_INT
, 1)
1298 OPTION(filestore_apply_finisher_threads
, OPT_INT
, 1)
1299 OPTION(filestore_dump_file
, OPT_STR
, "") // file onto which store transaction dumps
1300 OPTION(filestore_kill_at
, OPT_INT
, 0) // inject a failure at the n'th opportunity
1301 OPTION(filestore_inject_stall
, OPT_INT
, 0) // artificially stall for N seconds in op queue thread
1302 OPTION(filestore_fail_eio
, OPT_BOOL
, true) // fail/crash on EIO
1303 OPTION(filestore_debug_verify_split
, OPT_BOOL
, false)
1304 OPTION(journal_dio
, OPT_BOOL
, true)
1305 OPTION(journal_aio
, OPT_BOOL
, true)
1306 OPTION(journal_force_aio
, OPT_BOOL
, false)
1307 OPTION(journal_block_size
, OPT_INT
, 4096)
1309 // max bytes to search ahead in journal searching for corruption
1310 OPTION(journal_max_corrupt_search
, OPT_U64
, 10<<20)
1311 OPTION(journal_block_align
, OPT_BOOL
, true)
1312 OPTION(journal_write_header_frequency
, OPT_U64
, 0)
1313 OPTION(journal_max_write_bytes
, OPT_INT
, 10 << 20)
1314 OPTION(journal_max_write_entries
, OPT_INT
, 100)
1316 /// Target range for journal fullness
1317 OPTION(journal_throttle_low_threshhold
, OPT_DOUBLE
, 0.6)
1318 OPTION(journal_throttle_high_threshhold
, OPT_DOUBLE
, 0.9)
1320 /// Multiple over expected at high_threshhold. Defaults to 0 (disabled).
1321 OPTION(journal_throttle_high_multiple
, OPT_DOUBLE
, 0)
1322 /// Multiple over expected at max. Defaults to 0 (disabled).
1323 OPTION(journal_throttle_max_multiple
, OPT_DOUBLE
, 0)
1325 OPTION(journal_align_min_size
, OPT_INT
, 64 << 10) // align data payloads >= this.
1326 OPTION(journal_replay_from
, OPT_INT
, 0)
1327 OPTION(journal_zero_on_create
, OPT_BOOL
, false)
1328 OPTION(journal_ignore_corruption
, OPT_BOOL
, false) // assume journal is not corrupt
1329 OPTION(journal_discard
, OPT_BOOL
, false) //using ssd disk as journal, whether support discard nouse journal-data.
1331 OPTION(fio_dir
, OPT_STR
, "/tmp/fio") // fio data directory for fio-objectstore
1333 OPTION(rados_mon_op_timeout
, OPT_DOUBLE
, 0) // how many seconds to wait for a response from the monitor before returning an error from a rados operation. 0 means on limit.
1334 OPTION(rados_osd_op_timeout
, OPT_DOUBLE
, 0) // how many seconds to wait for a response from osds before returning an error from a rados operation. 0 means no limit.
1335 OPTION(rados_tracing
, OPT_BOOL
, false) // true if LTTng-UST tracepoints should be enabled
1337 OPTION(rbd_op_threads
, OPT_INT
, 1)
1338 OPTION(rbd_op_thread_timeout
, OPT_INT
, 60)
1339 OPTION(rbd_non_blocking_aio
, OPT_BOOL
, true) // process AIO ops from a worker thread to prevent blocking
1340 OPTION(rbd_cache
, OPT_BOOL
, true) // whether to enable caching (writeback unless rbd_cache_max_dirty is 0)
1341 OPTION(rbd_cache_writethrough_until_flush
, OPT_BOOL
, true) // whether to make writeback caching writethrough until flush is called, to be sure the user of librbd will send flushs so that writeback is safe
1342 OPTION(rbd_cache_size
, OPT_LONGLONG
, 32<<20) // cache size in bytes
1343 OPTION(rbd_cache_max_dirty
, OPT_LONGLONG
, 24<<20) // dirty limit in bytes - set to 0 for write-through caching
1344 OPTION(rbd_cache_target_dirty
, OPT_LONGLONG
, 16<<20) // target dirty limit in bytes
1345 OPTION(rbd_cache_max_dirty_age
, OPT_FLOAT
, 1.0) // seconds in cache before writeback starts
1346 OPTION(rbd_cache_max_dirty_object
, OPT_INT
, 0) // dirty limit for objects - set to 0 for auto calculate from rbd_cache_size
1347 OPTION(rbd_cache_block_writes_upfront
, OPT_BOOL
, false) // whether to block writes to the cache before the aio_write call completes (true), or block before the aio completion is called (false)
1348 OPTION(rbd_concurrent_management_ops
, OPT_INT
, 10) // how many operations can be in flight for a management operation like deleting or resizing an image
1349 OPTION(rbd_balance_snap_reads
, OPT_BOOL
, false)
1350 OPTION(rbd_localize_snap_reads
, OPT_BOOL
, false)
1351 OPTION(rbd_balance_parent_reads
, OPT_BOOL
, false)
1352 OPTION(rbd_localize_parent_reads
, OPT_BOOL
, true)
1353 OPTION(rbd_readahead_trigger_requests
, OPT_INT
, 10) // number of sequential requests necessary to trigger readahead
1354 OPTION(rbd_readahead_max_bytes
, OPT_LONGLONG
, 512 * 1024) // set to 0 to disable readahead
1355 OPTION(rbd_readahead_disable_after_bytes
, OPT_LONGLONG
, 50 * 1024 * 1024) // how many bytes are read in total before readahead is disabled
1356 OPTION(rbd_clone_copy_on_read
, OPT_BOOL
, false)
1357 OPTION(rbd_blacklist_on_break_lock
, OPT_BOOL
, true) // whether to blacklist clients whose lock was broken
1358 OPTION(rbd_blacklist_expire_seconds
, OPT_INT
, 0) // number of seconds to blacklist - set to 0 for OSD default
1359 OPTION(rbd_request_timed_out_seconds
, OPT_INT
, 30) // number of seconds before maint request times out
1360 OPTION(rbd_skip_partial_discard
, OPT_BOOL
, false) // when trying to discard a range inside an object, set to true to skip zeroing the range.
1361 OPTION(rbd_enable_alloc_hint
, OPT_BOOL
, true) // when writing a object, it will issue a hint to osd backend to indicate the expected size object need
1362 OPTION(rbd_tracing
, OPT_BOOL
, false) // true if LTTng-UST tracepoints should be enabled
1363 OPTION(rbd_validate_pool
, OPT_BOOL
, true) // true if empty pools should be validated for RBD compatibility
1364 OPTION(rbd_validate_names
, OPT_BOOL
, true) // true if image specs should be validated
1365 OPTION(rbd_auto_exclusive_lock_until_manual_request
, OPT_BOOL
, true) // whether to automatically acquire/release exclusive lock until it is explicitly requested, i.e. before we know the user of librbd is properly using the lock API
1366 OPTION(rbd_mirroring_resync_after_disconnect
, OPT_BOOL
, false) // automatically start image resync after mirroring is disconnected due to being laggy
1367 OPTION(rbd_mirroring_replay_delay
, OPT_INT
, 0) // time-delay in seconds for rbd-mirror asynchronous replication
1370 * The following options change the behavior for librbd's image creation methods that
1371 * don't require all of the parameters. These are provided so that older programs
1372 * can take advantage of newer features without being rewritten to use new versions
1373 * of the image creation functions.
1375 * rbd_create()/RBD::create() are affected by all of these options.
1377 * rbd_create2()/RBD::create2() and rbd_clone()/RBD::clone() are affected by:
1378 * - rbd_default_order
1379 * - rbd_default_stripe_count
1380 * - rbd_default_stripe_size
1382 * rbd_create3()/RBD::create3() and rbd_clone2/RBD::clone2() are only
1383 * affected by rbd_default_order.
1385 OPTION(rbd_default_format
, OPT_INT
, 2)
1386 OPTION(rbd_default_order
, OPT_INT
, 22)
1387 OPTION(rbd_default_stripe_count
, OPT_U64
, 0) // changing requires stripingv2 feature
1388 OPTION(rbd_default_stripe_unit
, OPT_U64
, 0) // changing to non-object size requires stripingv2 feature
1389 OPTION(rbd_default_data_pool
, OPT_STR
, "") // optional default pool for storing image data blocks
1392 * RBD features are only applicable for v2 images. This setting accepts either
1393 * an integer bitmask value or comma-delimited string of RBD feature names.
1394 * This setting is always internally stored as an integer bitmask value. The
1395 * mapping between feature bitmask value and feature name is as follows:
1399 * +4 -> exclusive-lock
1402 * +32 -> deep-flatten
1406 SAFE_OPTION(rbd_default_features
, OPT_STR
, "layering,exclusive-lock,object-map,fast-diff,deep-flatten")
1407 OPTION_VALIDATOR(rbd_default_features
)
1409 OPTION(rbd_default_map_options
, OPT_STR
, "") // default rbd map -o / --options
1412 * RBD journal options.
1414 OPTION(rbd_journal_order
, OPT_U32
, 24) // bits to shift to compute journal object max size, between 12 and 64
1415 OPTION(rbd_journal_splay_width
, OPT_U32
, 4) // number of active journal objects
1416 OPTION(rbd_journal_commit_age
, OPT_DOUBLE
, 5) // commit time interval, seconds
1417 OPTION(rbd_journal_object_flush_interval
, OPT_INT
, 0) // maximum number of pending commits per journal object
1418 OPTION(rbd_journal_object_flush_bytes
, OPT_INT
, 0) // maximum number of pending bytes per journal object
1419 OPTION(rbd_journal_object_flush_age
, OPT_DOUBLE
, 0) // maximum age (in seconds) for pending commits
1420 OPTION(rbd_journal_pool
, OPT_STR
, "") // pool for journal objects
1421 OPTION(rbd_journal_max_payload_bytes
, OPT_U32
, 16384) // maximum journal payload size before splitting
1422 OPTION(rbd_journal_max_concurrent_object_sets
, OPT_INT
, 0) // maximum number of object sets a journal client can be behind before it is automatically unregistered
1425 * RBD Mirror options
1427 OPTION(rbd_mirror_journal_commit_age
, OPT_DOUBLE
, 5) // commit time interval, seconds
1428 OPTION(rbd_mirror_journal_poll_age
, OPT_DOUBLE
, 5) // maximum age (in seconds) between successive journal polls
1429 OPTION(rbd_mirror_journal_max_fetch_bytes
, OPT_U32
, 32768) // maximum bytes to read from each journal data object per fetch
1430 OPTION(rbd_mirror_sync_point_update_age
, OPT_DOUBLE
, 30) // number of seconds between each update of the image sync point object number
1431 OPTION(rbd_mirror_concurrent_image_syncs
, OPT_U32
, 5) // maximum number of image syncs in parallel
1432 OPTION(rbd_mirror_pool_replayers_refresh_interval
, OPT_INT
, 30) // interval to refresh peers in rbd-mirror daemon
1433 OPTION(rbd_mirror_delete_retry_interval
, OPT_DOUBLE
, 30) // interval to check and retry the failed requests in deleter
1434 OPTION(rbd_mirror_image_state_check_interval
, OPT_INT
, 30) // interval to get images from pool watcher and set sources in replayer
1435 OPTION(rbd_mirror_leader_heartbeat_interval
, OPT_INT
, 5) // interval (in seconds) between mirror leader heartbeats
1436 OPTION(rbd_mirror_leader_max_missed_heartbeats
, OPT_INT
, 2) // number of missed heartbeats for non-lock owner to attempt to acquire lock
1437 OPTION(rbd_mirror_leader_max_acquire_attempts_before_break
, OPT_INT
, 3) // number of failed attempts to acquire lock after missing heartbeats before breaking lock
1439 OPTION(nss_db_path
, OPT_STR
, "") // path to nss db
1442 OPTION(rgw_max_chunk_size
, OPT_INT
, 4 * 1024 * 1024)
1443 OPTION(rgw_put_obj_min_window_size
, OPT_INT
, 16 * 1024 * 1024)
1444 OPTION(rgw_put_obj_max_window_size
, OPT_INT
, 64 * 1024 * 1024)
1445 OPTION(rgw_max_put_size
, OPT_U64
, 5ULL*1024*1024*1024)
1446 OPTION(rgw_max_put_param_size
, OPT_U64
, 1 * 1024 * 1024) // max input size for PUT requests accepting json/xml params
1449 * override max bucket index shards in zone configuration (if not zero)
1451 * Represents the number of shards for the bucket index object, a value of zero
1452 * indicates there is no sharding. By default (no sharding, the name of the object
1453 * is '.dir.{marker}', with sharding, the name is '.dir.{markder}.{sharding_id}',
1454 * sharding_id is zero-based value. It is not recommended to set a too large value
1455 * (e.g. thousand) as it increases the cost for bucket listing.
1457 OPTION(rgw_override_bucket_index_max_shards
, OPT_U32
, 0)
1460 * Represents the maximum AIO pending requests for the bucket index object shards.
1462 OPTION(rgw_bucket_index_max_aio
, OPT_U32
, 8)
1465 * whether or not the quota/gc threads should be started
1467 OPTION(rgw_enable_quota_threads
, OPT_BOOL
, true)
1468 OPTION(rgw_enable_gc_threads
, OPT_BOOL
, true)
1469 OPTION(rgw_enable_lc_threads
, OPT_BOOL
, true)
1472 OPTION(rgw_data
, OPT_STR
, "/var/lib/ceph/radosgw/$cluster-$id")
1473 OPTION(rgw_enable_apis
, OPT_STR
, "s3, s3website, swift, swift_auth, admin")
1474 OPTION(rgw_cache_enabled
, OPT_BOOL
, true) // rgw cache enabled
1475 OPTION(rgw_cache_lru_size
, OPT_INT
, 10000) // num of entries in rgw cache
1476 OPTION(rgw_socket_path
, OPT_STR
, "") // path to unix domain socket, if not specified, rgw will not run as external fcgi
1477 OPTION(rgw_host
, OPT_STR
, "") // host for radosgw, can be an IP, default is 0.0.0.0
1478 OPTION(rgw_port
, OPT_STR
, "") // port to listen, format as "8080" "5000", if not specified, rgw will not run external fcgi
1479 OPTION(rgw_dns_name
, OPT_STR
, "") // hostname suffix on buckets
1480 OPTION(rgw_dns_s3website_name
, OPT_STR
, "") // hostname suffix on buckets for s3-website endpoint
1481 OPTION(rgw_content_length_compat
, OPT_BOOL
, false) // Check both HTTP_CONTENT_LENGTH and CONTENT_LENGTH in fcgi env
1482 OPTION(rgw_lifecycle_work_time
, OPT_STR
, "00:00-06:00") //job process lc at 00:00-06:00s
1483 OPTION(rgw_lc_lock_max_time
, OPT_INT
, 60) // total run time for a single lc processor work
1484 OPTION(rgw_lc_max_objs
, OPT_INT
, 32)
1485 OPTION(rgw_lc_debug_interval
, OPT_INT
, -1) // Debug run interval, in seconds
1486 OPTION(rgw_script_uri
, OPT_STR
, "") // alternative value for SCRIPT_URI if not set in request
1487 OPTION(rgw_request_uri
, OPT_STR
, "") // alternative value for REQUEST_URI if not set in request
1488 OPTION(rgw_swift_url
, OPT_STR
, "") // the swift url, being published by the internal swift auth
1489 OPTION(rgw_swift_url_prefix
, OPT_STR
, "swift") // entry point for which a url is considered a swift url
1490 OPTION(rgw_swift_auth_url
, OPT_STR
, "") // default URL to go and verify tokens for v1 auth (if not using internal swift auth)
1491 OPTION(rgw_swift_auth_entry
, OPT_STR
, "auth") // entry point for which a url is considered a swift auth url
1492 OPTION(rgw_swift_tenant_name
, OPT_STR
, "") // tenant name to use for swift access
1493 OPTION(rgw_swift_account_in_url
, OPT_BOOL
, false) // assume that URL always contain the account (aka tenant) part
1494 OPTION(rgw_swift_enforce_content_length
, OPT_BOOL
, false) // enforce generation of Content-Length even in cost of performance or scalability
1495 OPTION(rgw_keystone_url
, OPT_STR
, "") // url for keystone server
1496 OPTION(rgw_keystone_admin_token
, OPT_STR
, "") // keystone admin token (shared secret)
1497 OPTION(rgw_keystone_admin_user
, OPT_STR
, "") // keystone admin user name
1498 OPTION(rgw_keystone_admin_password
, OPT_STR
, "") // keystone admin user password
1499 OPTION(rgw_keystone_admin_tenant
, OPT_STR
, "") // keystone admin user tenant (for keystone v2.0)
1500 OPTION(rgw_keystone_admin_project
, OPT_STR
, "") // keystone admin user project (for keystone v3)
1501 OPTION(rgw_keystone_admin_domain
, OPT_STR
, "") // keystone admin user domain
1502 OPTION(rgw_keystone_barbican_user
, OPT_STR
, "") // keystone user to access barbican secrets
1503 OPTION(rgw_keystone_barbican_password
, OPT_STR
, "") // keystone password for barbican user
1504 OPTION(rgw_keystone_barbican_tenant
, OPT_STR
, "") // keystone barbican user tenant (for keystone v2.0)
1505 OPTION(rgw_keystone_barbican_project
, OPT_STR
, "") // keystone barbican user project (for keystone v3)
1506 OPTION(rgw_keystone_barbican_domain
, OPT_STR
, "") // keystone barbican user domain
1507 OPTION(rgw_keystone_api_version
, OPT_INT
, 2) // Version of Keystone API to use (2 or 3)
1508 OPTION(rgw_keystone_accepted_roles
, OPT_STR
, "Member, admin") // roles required to serve requests
1509 OPTION(rgw_keystone_accepted_admin_roles
, OPT_STR
, "") // list of roles allowing an user to gain admin privileges
1510 OPTION(rgw_keystone_token_cache_size
, OPT_INT
, 10000) // max number of entries in keystone token cache
1511 OPTION(rgw_keystone_revocation_interval
, OPT_INT
, 15 * 60) // seconds between tokens revocation check
1512 OPTION(rgw_keystone_verify_ssl
, OPT_BOOL
, true) // should we try to verify keystone's ssl
1513 OPTION(rgw_keystone_implicit_tenants
, OPT_BOOL
, false) // create new users in their own tenants of the same name
1514 OPTION(rgw_cross_domain_policy
, OPT_STR
, "<allow-access-from domain=\"*\" secure=\"false\" />")
1515 OPTION(rgw_healthcheck_disabling_path
, OPT_STR
, "") // path that existence causes the healthcheck to respond 503
1516 OPTION(rgw_s3_auth_use_rados
, OPT_BOOL
, true) // should we try to use the internal credentials for s3?
1517 OPTION(rgw_s3_auth_use_keystone
, OPT_BOOL
, false) // should we try to use keystone for s3?
1518 OPTION(rgw_s3_auth_aws4_force_boto2_compat
, OPT_BOOL
, true) // force aws4 auth boto2 compatibility
1519 OPTION(rgw_barbican_url
, OPT_STR
, "") // url for barbican server
1521 /* OpenLDAP-style LDAP parameter strings */
1522 /* rgw_ldap_uri space-separated list of LDAP servers in URI format */
1523 OPTION(rgw_ldap_uri
, OPT_STR
, "ldaps://<ldap.your.domain>")
1524 /* rgw_ldap_binddn LDAP entry RGW will bind with (user match) */
1525 OPTION(rgw_ldap_binddn
, OPT_STR
, "uid=admin,cn=users,dc=example,dc=com")
1526 /* rgw_ldap_searchdn LDAP search base (basedn) */
1527 OPTION(rgw_ldap_searchdn
, OPT_STR
, "cn=users,cn=accounts,dc=example,dc=com")
1528 /* rgw_ldap_dnattr LDAP attribute containing RGW user names (to form binddns)*/
1529 OPTION(rgw_ldap_dnattr
, OPT_STR
, "uid")
1530 /* rgw_ldap_secret file containing credentials for rgw_ldap_binddn */
1531 OPTION(rgw_ldap_secret
, OPT_STR
, "/etc/openldap/secret")
1532 /* rgw_s3_auth_use_ldap use LDAP for RGW auth? */
1533 OPTION(rgw_s3_auth_use_ldap
, OPT_BOOL
, false)
1534 /* rgw_ldap_searchfilter LDAP search filter */
1535 OPTION(rgw_ldap_searchfilter
, OPT_STR
, "")
1537 OPTION(rgw_admin_entry
, OPT_STR
, "admin") // entry point for which a url is considered an admin request
1538 OPTION(rgw_enforce_swift_acls
, OPT_BOOL
, true)
1539 OPTION(rgw_swift_token_expiration
, OPT_INT
, 24 * 3600) // time in seconds for swift token expiration
1540 OPTION(rgw_print_continue
, OPT_BOOL
, true) // enable if 100-Continue works
1541 OPTION(rgw_print_prohibited_content_length
, OPT_BOOL
, false) // violate RFC 7230 and send Content-Length in 204 and 304
1542 OPTION(rgw_remote_addr_param
, OPT_STR
, "REMOTE_ADDR") // e.g. X-Forwarded-For, if you have a reverse proxy
1543 OPTION(rgw_op_thread_timeout
, OPT_INT
, 10*60)
1544 OPTION(rgw_op_thread_suicide_timeout
, OPT_INT
, 0)
1545 OPTION(rgw_thread_pool_size
, OPT_INT
, 100)
1546 OPTION(rgw_num_control_oids
, OPT_INT
, 8)
1547 OPTION(rgw_num_rados_handles
, OPT_U32
, 1)
1549 /* The following are tunables for caches of RGW NFS (and other file
1552 * The file handle cache is a partitioned hash table
1553 * (fhcache_partitions), each with a closed hash part and backing
1554 * b-tree mapping. The number of partions is expected to be a small
1555 * prime, the cache size something larger but less than 5K, the total
1556 * size of the cache is n_part * cache_size.
1558 OPTION(rgw_nfs_lru_lanes
, OPT_INT
, 5)
1559 OPTION(rgw_nfs_lru_lane_hiwat
, OPT_INT
, 911)
1560 OPTION(rgw_nfs_fhcache_partitions
, OPT_INT
, 3)
1561 OPTION(rgw_nfs_fhcache_size
, OPT_INT
, 2017) /* 3*2017=6051 */
1562 OPTION(rgw_nfs_namespace_expire_secs
, OPT_INT
, 300) /* namespace invalidate
1564 OPTION(rgw_nfs_max_gc
, OPT_INT
, 300) /* max gc events per cycle */
1565 OPTION(rgw_nfs_write_completion_interval_s
, OPT_INT
, 10) /* stateless (V3)
1569 OPTION(rgw_zone
, OPT_STR
, "") // zone name
1570 OPTION(rgw_zone_root_pool
, OPT_STR
, ".rgw.root") // pool where zone specific info is stored
1571 OPTION(rgw_default_zone_info_oid
, OPT_STR
, "default.zone") // oid where default zone info is stored
1572 OPTION(rgw_region
, OPT_STR
, "") // region name
1573 OPTION(rgw_region_root_pool
, OPT_STR
, ".rgw.root") // pool where all region info is stored
1574 OPTION(rgw_default_region_info_oid
, OPT_STR
, "default.region") // oid where default region info is stored
1575 OPTION(rgw_zonegroup
, OPT_STR
, "") // zone group name
1576 OPTION(rgw_zonegroup_root_pool
, OPT_STR
, ".rgw.root") // pool where all zone group info is stored
1577 OPTION(rgw_default_zonegroup_info_oid
, OPT_STR
, "default.zonegroup") // oid where default zone group info is stored
1578 OPTION(rgw_realm
, OPT_STR
, "") // realm name
1579 OPTION(rgw_realm_root_pool
, OPT_STR
, ".rgw.root") // pool where all realm info is stored
1580 OPTION(rgw_default_realm_info_oid
, OPT_STR
, "default.realm") // oid where default realm info is stored
1581 OPTION(rgw_period_root_pool
, OPT_STR
, ".rgw.root") // pool where all period info is stored
1582 OPTION(rgw_period_latest_epoch_info_oid
, OPT_STR
, ".latest_epoch") // oid where current period info is stored
1583 OPTION(rgw_log_nonexistent_bucket
, OPT_BOOL
, false)
1584 OPTION(rgw_log_object_name
, OPT_STR
, "%Y-%m-%d-%H-%i-%n") // man date to see codes (a subset are supported)
1585 OPTION(rgw_log_object_name_utc
, OPT_BOOL
, false)
1586 OPTION(rgw_usage_max_shards
, OPT_INT
, 32)
1587 OPTION(rgw_usage_max_user_shards
, OPT_INT
, 1)
1588 OPTION(rgw_enable_ops_log
, OPT_BOOL
, false) // enable logging every rgw operation
1589 OPTION(rgw_enable_usage_log
, OPT_BOOL
, false) // enable logging bandwidth usage
1590 OPTION(rgw_ops_log_rados
, OPT_BOOL
, true) // whether ops log should go to rados
1591 OPTION(rgw_ops_log_socket_path
, OPT_STR
, "") // path to unix domain socket where ops log can go
1592 OPTION(rgw_ops_log_data_backlog
, OPT_INT
, 5 << 20) // max data backlog for ops log
1593 OPTION(rgw_fcgi_socket_backlog
, OPT_INT
, 1024) // socket backlog for fcgi
1594 OPTION(rgw_usage_log_flush_threshold
, OPT_INT
, 1024) // threshold to flush pending log data
1595 OPTION(rgw_usage_log_tick_interval
, OPT_INT
, 30) // flush pending log data every X seconds
1596 OPTION(rgw_intent_log_object_name
, OPT_STR
, "%Y-%m-%d-%i-%n") // man date to see codes (a subset are supported)
1597 OPTION(rgw_intent_log_object_name_utc
, OPT_BOOL
, false)
1598 OPTION(rgw_init_timeout
, OPT_INT
, 300) // time in seconds
1599 OPTION(rgw_mime_types_file
, OPT_STR
, "/etc/mime.types")
1600 OPTION(rgw_gc_max_objs
, OPT_INT
, 32)
1601 OPTION(rgw_gc_obj_min_wait
, OPT_INT
, 2 * 3600) // wait time before object may be handled by gc
1602 OPTION(rgw_gc_processor_max_time
, OPT_INT
, 3600) // total run time for a single gc processor work
1603 OPTION(rgw_gc_processor_period
, OPT_INT
, 3600) // gc processor cycle time
1604 OPTION(rgw_s3_success_create_obj_status
, OPT_INT
, 0) // alternative success status response for create-obj (0 - default)
1605 OPTION(rgw_resolve_cname
, OPT_BOOL
, false) // should rgw try to resolve hostname as a dns cname record
1606 OPTION(rgw_obj_stripe_size
, OPT_INT
, 4 << 20)
1607 OPTION(rgw_extended_http_attrs
, OPT_STR
, "") // list of extended attrs that can be set on objects (beyond the default)
1608 OPTION(rgw_exit_timeout_secs
, OPT_INT
, 120) // how many seconds to wait for process to go down before exiting unconditionally
1609 OPTION(rgw_get_obj_window_size
, OPT_INT
, 16 << 20) // window size in bytes for single get obj request
1610 OPTION(rgw_get_obj_max_req_size
, OPT_INT
, 4 << 20) // max length of a single get obj rados op
1611 OPTION(rgw_relaxed_s3_bucket_names
, OPT_BOOL
, false) // enable relaxed bucket name rules for US region buckets
1612 OPTION(rgw_defer_to_bucket_acls
, OPT_STR
, "") // if the user has bucket perms, use those before key perms (recurse and full_control)
1613 OPTION(rgw_list_buckets_max_chunk
, OPT_INT
, 1000) // max buckets to retrieve in a single op when listing user buckets
1614 OPTION(rgw_md_log_max_shards
, OPT_INT
, 64) // max shards for metadata log
1615 OPTION(rgw_num_zone_opstate_shards
, OPT_INT
, 128) // max shards for keeping inter-region copy progress info
1616 OPTION(rgw_opstate_ratelimit_sec
, OPT_INT
, 30) // min time between opstate updates on a single upload (0 for disabling ratelimit)
1617 OPTION(rgw_curl_wait_timeout_ms
, OPT_INT
, 1000) // timeout for certain curl calls
1618 OPTION(rgw_copy_obj_progress
, OPT_BOOL
, true) // should dump progress during long copy operations?
1619 OPTION(rgw_copy_obj_progress_every_bytes
, OPT_INT
, 1024 * 1024) // min bytes between copy progress output
1620 OPTION(rgw_obj_tombstone_cache_size
, OPT_INT
, 1000) // how many objects in tombstone cache, which is used in multi-zone sync to keep
1621 // track of removed objects' mtime
1623 OPTION(rgw_data_log_window
, OPT_INT
, 30) // data log entries window (in seconds)
1624 OPTION(rgw_data_log_changes_size
, OPT_INT
, 1000) // number of in-memory entries to hold for data changes log
1625 OPTION(rgw_data_log_num_shards
, OPT_INT
, 128) // number of objects to keep data changes log on
1626 OPTION(rgw_data_log_obj_prefix
, OPT_STR
, "data_log") //
1627 OPTION(rgw_replica_log_obj_prefix
, OPT_STR
, "replica_log") //
1629 OPTION(rgw_bucket_quota_ttl
, OPT_INT
, 600) // time for cached bucket stats to be cached within rgw instance
1630 OPTION(rgw_bucket_quota_soft_threshold
, OPT_DOUBLE
, 0.95) // threshold from which we don't rely on cached info for quota decisions
1631 OPTION(rgw_bucket_quota_cache_size
, OPT_INT
, 10000) // number of entries in bucket quota cache
1632 OPTION(rgw_bucket_default_quota_max_objects
, OPT_INT
, -1) // number of objects allowed
1633 OPTION(rgw_bucket_default_quota_max_size
, OPT_LONGLONG
, -1) // Max size of object in bytes
1635 OPTION(rgw_expose_bucket
, OPT_BOOL
, false) // Return the bucket name in the 'Bucket' response header
1637 OPTION(rgw_frontends
, OPT_STR
, "fastcgi, civetweb port=7480") // rgw front ends
1639 OPTION(rgw_user_quota_bucket_sync_interval
, OPT_INT
, 180) // time period for accumulating modified buckets before syncing stats
1640 OPTION(rgw_user_quota_sync_interval
, OPT_INT
, 3600 * 24) // time period for accumulating modified buckets before syncing entire user stats
1641 OPTION(rgw_user_quota_sync_idle_users
, OPT_BOOL
, false) // whether stats for idle users be fully synced
1642 OPTION(rgw_user_quota_sync_wait_time
, OPT_INT
, 3600 * 24) // min time between two full stats sync for non-idle users
1643 OPTION(rgw_user_default_quota_max_objects
, OPT_INT
, -1) // number of objects allowed
1644 OPTION(rgw_user_default_quota_max_size
, OPT_LONGLONG
, -1) // Max size of object in bytes
1646 OPTION(rgw_multipart_min_part_size
, OPT_INT
, 5 * 1024 * 1024) // min size for each part (except for last one) in multipart upload
1647 OPTION(rgw_multipart_part_upload_limit
, OPT_INT
, 10000) // parts limit in multipart upload
1649 OPTION(rgw_max_slo_entries
, OPT_INT
, 1000) // default number of max entries in slo
1651 OPTION(rgw_olh_pending_timeout_sec
, OPT_INT
, 3600) // time until we retire a pending olh change
1652 OPTION(rgw_user_max_buckets
, OPT_INT
, 1000) // global option to set max buckets count for all user
1654 OPTION(rgw_objexp_gc_interval
, OPT_U32
, 60 * 10) // maximum time between round of expired objects garbage collecting
1655 OPTION(rgw_objexp_time_step
, OPT_U32
, 4096) // number of seconds for rounding the timestamps
1656 OPTION(rgw_objexp_hints_num_shards
, OPT_U32
, 127) // maximum number of parts in which the hint index is stored in
1657 OPTION(rgw_objexp_chunk_size
, OPT_U32
, 100) // maximum number of entries in a single operation when processing objexp data
1659 OPTION(rgw_enable_static_website
, OPT_BOOL
, false) // enable static website feature
1660 OPTION(rgw_log_http_headers
, OPT_STR
, "" ) // list of HTTP headers to log when seen, ignores case (e.g., http_x_forwarded_for
1662 OPTION(rgw_num_async_rados_threads
, OPT_INT
, 32) // num of threads to use for async rados operations
1663 OPTION(rgw_md_notify_interval_msec
, OPT_INT
, 200) // metadata changes notification interval to followers
1664 OPTION(rgw_run_sync_thread
, OPT_BOOL
, true) // whether radosgw (not radosgw-admin) spawns the sync thread
1665 OPTION(rgw_sync_lease_period
, OPT_INT
, 120) // time in second for lease that rgw takes on a specific log (or log shard)
1666 OPTION(rgw_sync_log_trim_interval
, OPT_INT
, 1200) // time in seconds between attempts to trim sync logs
1668 OPTION(rgw_sync_data_inject_err_probability
, OPT_DOUBLE
, 0) // range [0, 1]
1669 OPTION(rgw_sync_meta_inject_err_probability
, OPT_DOUBLE
, 0) // range [0, 1]
1672 OPTION(rgw_period_push_interval
, OPT_DOUBLE
, 2) // seconds to wait before retrying "period push"
1673 OPTION(rgw_period_push_interval_max
, OPT_DOUBLE
, 30) // maximum interval after exponential backoff
1675 OPTION(rgw_safe_max_objects_per_shard
, OPT_INT
, 100*1024) // safe max loading
1676 OPTION(rgw_shard_warning_threshold
, OPT_DOUBLE
, 90) // pct of safe max
1679 OPTION(rgw_swift_versioning_enabled
, OPT_BOOL
, false) // whether swift object versioning feature is enabled
1681 OPTION(mgr_module_path
, OPT_STR
, CEPH_PKGLIBDIR
"/mgr") // where to load python modules from
1682 OPTION(mgr_modules
, OPT_STR
, "rest") // Which modules to load
1683 OPTION(mgr_data
, OPT_STR
, "/var/lib/ceph/mgr/$cluster-$id") // where to find keyring etc
1684 OPTION(mgr_beacon_period
, OPT_INT
, 5) // How frequently to send beacon
1685 OPTION(mgr_stats_period
, OPT_INT
, 5) // How frequently to send stats
1686 OPTION(mgr_client_bytes
, OPT_U64
, 128*1048576) // bytes from clients
1687 OPTION(mgr_client_messages
, OPT_U64
, 512) // messages from clients
1688 OPTION(mgr_osd_bytes
, OPT_U64
, 512*1048576) // bytes from osds
1689 OPTION(mgr_osd_messages
, OPT_U64
, 8192) // messages from osds
1690 OPTION(mgr_mds_bytes
, OPT_U64
, 128*1048576) // bytes from mdss
1691 OPTION(mgr_mds_messages
, OPT_U64
, 128) // messages from mdss
1692 OPTION(mgr_mon_bytes
, OPT_U64
, 128*1048576) // bytes from mons
1693 OPTION(mgr_mon_messages
, OPT_U64
, 128) // messages from mons
1695 OPTION(mgr_connect_retry_interval
, OPT_DOUBLE
, 1.0)
1697 OPTION(mon_mgr_digest_period
, OPT_INT
, 5) // How frequently to send digests
1698 OPTION(mon_mgr_beacon_grace
, OPT_INT
, 30) // How long to wait to failover
1699 OPTION(mon_mgr_inactive_grace
, OPT_INT
, 60) // How long before health WARN -> ERR
1700 OPTION(rgw_crypt_require_ssl
, OPT_BOOL
, true) // requests including encryption key headers must be sent over ssl
1701 OPTION(rgw_crypt_default_encryption_key
, OPT_STR
, "") // base64 encoded key for encryption of rgw objects
1702 OPTION(rgw_crypt_s3_kms_encryption_keys
, OPT_STR
, "") // extra keys that may be used for aws:kms
1703 // defined as map "key1=YmluCmJvb3N0CmJvb3N0LQ== key2=b3V0CnNyYwpUZXN0aW5nCg=="
1704 OPTION(rgw_crypt_suppress_logs
, OPT_BOOL
, true) // suppress logs that might print customer key
1705 OPTION(rgw_list_bucket_min_readahead
, OPT_INT
, 1000) // minimum number of entries to read from rados for bucket listing
1707 OPTION(rgw_rest_getusage_op_compat
, OPT_BOOL
, false) // dump description of total stats for s3 GetUsage API
1709 OPTION(mutex_perf_counter
, OPT_BOOL
, false) // enable/disable mutex perf counter
1710 OPTION(throttler_perf_counter
, OPT_BOOL
, true) // enable/disable throttler perf counter
1712 /* The following are tunables for torrent data */
1713 OPTION(rgw_torrent_flag
, OPT_BOOL
, false) // produce torrent function flag
1714 OPTION(rgw_torrent_tracker
, OPT_STR
, "") // torrent field annouce and annouce list
1715 OPTION(rgw_torrent_createby
, OPT_STR
, "") // torrent field created by
1716 OPTION(rgw_torrent_comment
, OPT_STR
, "") // torrent field comment
1717 OPTION(rgw_torrent_encoding
, OPT_STR
, "") // torrent field encoding
1718 OPTION(rgw_torrent_origin
, OPT_STR
, "") // torrent origin
1719 OPTION(rgw_torrent_sha_unit
, OPT_INT
, 512*1024) // torrent field piece length 512K
1721 OPTION(event_tracing
, OPT_BOOL
, false) // true if LTTng-UST tracepoints should be enabled
1723 // This will be set to true when it is safe to start threads.
1724 // Once it is true, it will never change.
1725 OPTION(internal_safe_to_start_threads
, OPT_BOOL
, false)
1727 OPTION(debug_deliberately_leak_memory
, OPT_BOOL
, false)
1729 OPTION(rgw_swift_custom_header
, OPT_STR
, "") // option to enable swift custom headers