diff options
Diffstat (limited to 'src/common/options/mon.yaml.in')
| -rw-r--r-- | src/common/options/mon.yaml.in | 1340 |
1 files changed, 1340 insertions, 0 deletions
diff --git a/src/common/options/mon.yaml.in b/src/common/options/mon.yaml.in new file mode 100644 index 000000000..1cd655ad4 --- /dev/null +++ b/src/common/options/mon.yaml.in @@ -0,0 +1,1340 @@ +# -*- mode: YAML -*- +--- + +options: +- name: osd_crush_update_weight_set + type: bool + level: advanced + desc: update CRUSH weight-set weights when updating weights + long_desc: If this setting is true, we will update the weight-set weights when adjusting + an item's weight, effectively making changes take effect immediately, and discarding + any previous optimization in the weight-set value. Setting this value to false + will leave it to the balancer to (slowly, presumably) adjust weights to approach + the new target value. + default: true + with_legacy: true +- name: osd_pool_erasure_code_stripe_unit + type: size + level: advanced + desc: the amount of data (in bytes) in a data chunk, per stripe + fmt_desc: Sets the default size, in bytes, of a chunk of an object + stripe for erasure coded pools. Every object of size S + will be stored as N stripes, with each data chunk + receiving ``stripe unit`` bytes. Each stripe of ``N * + stripe unit`` bytes will be encoded/decoded + individually. This option can is overridden by the + ``stripe_unit`` setting in an erasure code profile. + default: 4_K + services: + - mon +- name: osd_pool_default_crimson + type: bool + level: advanced + desc: Create pools by default with FLAG_CRIMSON + default: false + services : + - mon + flags: + - runtime +- name: mon_max_pool_pg_num + type: uint + level: advanced + default: 64_K + fmt_desc: The maximum number of placement groups per pool. +- name: mon_mgr_digest_period + type: int + level: dev + desc: Period in seconds between monitor-to-manager health/status updates + default: 5 + services: + - mon +- name: mon_down_mkfs_grace + type: secs + level: advanced + desc: Period in seconds that the cluster may have a mon down after cluster creation + default: 1_min + services: + - mon +- name: mon_mgr_beacon_grace + type: secs + level: advanced + desc: Period in seconds from last beacon to monitor marking a manager daemon as + failed + default: 30 + services: + - mon +- name: mon_mgr_inactive_grace + type: int + level: advanced + desc: Period in seconds after cluster creation during which cluster may have no + active manager + long_desc: This grace period enables the cluster to come up cleanly without raising + spurious health check failures about managers that aren't online yet + default: 1_min + services: + - mon +- name: mon_mgr_mkfs_grace + type: int + level: advanced + desc: Period in seconds that the cluster may have no active manager before this + is reported as an ERR rather than a WARN + default: 2_min + services: + - mon +- name: mon_mgr_proxy_client_bytes_ratio + type: float + level: dev + desc: ratio of mon_client_bytes that can be consumed by proxied mgr commands before + we error out to client + default: 0.3 + services: + - mon +- name: mon_cluster_log_to_stderr + type: bool + level: advanced + desc: Make monitor send cluster log messages to stderr (prefixed by channel) + default: false + services: + - mon + see_also: + - log_stderr_prefix + flags: + - runtime + with_legacy: true +- name: mon_cluster_log_to_syslog + type: str + level: advanced + desc: Make monitor send cluster log messages to syslog + fmt_desc: Determines if the cluster log should be output to the syslog. + default: default=false + services: + - mon + flags: + - runtime + with_legacy: true +- name: mon_cluster_log_to_syslog_level + type: str + level: advanced + desc: Syslog level for cluster log messages + default: info + services: + - mon + see_also: + - mon_cluster_log_to_syslog + flags: + - runtime + with_legacy: true +- name: mon_cluster_log_to_syslog_facility + type: str + level: advanced + desc: Syslog facility for cluster log messages + default: daemon + services: + - mon + see_also: + - mon_cluster_log_to_syslog + flags: + - runtime + with_legacy: true +- name: mon_cluster_log_to_file + type: bool + level: advanced + desc: Make monitor send cluster log messages to file + default: true + services: + - mon + see_also: + - mon_cluster_log_file + flags: + - runtime + with_legacy: true +- name: mon_cluster_log_file + type: str + level: advanced + desc: File(s) to write cluster log to + long_desc: This can either be a simple file name to receive all messages, or a list + of key/value pairs where the key is the log channel and the value is the filename, + which may include $cluster and $channel metavariables + fmt_desc: | + The locations of the cluster's log files. There are two channels in + Ceph: ``cluster`` and ``audit``. This option represents a mapping + from channels to log files, where the log entries of that + channel are sent to. The ``default`` entry is a fallback + mapping for channels not explicitly specified. So, the following + default setting will send cluster log to ``$cluster.log``, and + send audit log to ``$cluster.audit.log``, where ``$cluster`` will + be replaced with the actual cluster name. + default: default=/var/log/ceph/$cluster.$channel.log cluster=/var/log/ceph/$cluster.log + services: + - mon + see_also: + - mon_cluster_log_to_file + flags: + - runtime + with_legacy: true +- name: mon_cluster_log_file_level + type: str + level: advanced + desc: Lowest level to include is cluster log file + default: debug + services: + - mon + see_also: + - mon_cluster_log_file + flags: + - runtime + with_legacy: true +- name: mon_cluster_log_to_graylog + type: str + level: advanced + desc: Make monitor send cluster log to graylog + default: 'false' + services: + - mon + flags: + - runtime + with_legacy: true +- name: mon_cluster_log_to_graylog_host + type: str + level: advanced + desc: Graylog host for cluster log messages + default: 127.0.0.1 + services: + - mon + see_also: + - mon_cluster_log_to_graylog + flags: + - runtime + with_legacy: true +- name: mon_cluster_log_to_graylog_port + type: str + level: advanced + desc: Graylog port for cluster log messages + default: '12201' + services: + - mon + see_also: + - mon_cluster_log_to_graylog + flags: + - runtime + with_legacy: true +- name: mon_cluster_log_to_journald + type: str + level: advanced + desc: Make monitor send cluster log to journald + default: 'false' + services: + - mon + flags: + - runtime +- name: mon_log_max + type: uint + level: advanced + desc: number of recent cluster log messages to retain + default: 10000 + services: + - mon + with_legacy: true +- name: mon_log_max_summary + type: uint + level: advanced + desc: number of recent cluster log messages to dedup against + default: 50 + services: + - mon + with_legacy: true +- name: mon_log_full_interval + type: uint + level: advanced + desc: how many epochs before we encode a full copy of recent log keys + default: 50 + services: [mon] + with_legacy: true +- name: mon_max_log_entries_per_event + type: int + level: advanced + desc: max cluster log entries per paxos event + fmt_desc: The maximum number of log entries per event. + default: 4096 + services: + - mon + with_legacy: true +- name: mon_health_to_clog + type: bool + level: advanced + desc: log monitor health to cluster log + fmt_desc: Enable sending a health summary to the cluster log periodically. + default: true + services: + - mon + with_legacy: true +- name: mon_health_to_clog_interval + type: int + level: advanced + desc: frequency to log monitor health to cluster log + fmt_desc: How often (in seconds) the monitor sends a health summary to the cluster + log (a non-positive number disables). Monitors will always + send a summary to the cluster log whether or not it differs from + the previous summary. + default: 10_min + services: + - mon + see_also: + - mon_health_to_clog + with_legacy: true +- name: mon_health_to_clog_tick_interval + type: float + level: dev + fmt_desc: How often (in seconds) the monitor sends a health summary to the cluster + log (a non-positive number disables). If current health summary + is empty or identical to the last time, monitor will not send it + to cluster log. + default: 1_min + services: + - mon + with_legacy: true +- name: mon_health_detail_to_clog + type: bool + level: dev + desc: log health detail to cluster log + default: true + with_legacy: true +- name: mon_warn_on_filestore_osds + type: bool + level: dev + desc: log health warn for filestore OSDs + default: true + with_legacy: true +- name: mon_health_max_detail + type: uint + level: advanced + desc: max detailed pgs to report in health detail + default: 50 + services: + - mon +- name: mon_health_log_update_period + type: int + level: dev + desc: minimum time in seconds between log messages about each health check + default: 5 + services: + - mon + min: 0 +- name: mon_data_avail_crit + type: int + level: advanced + desc: issue MON_DISK_CRIT health error when mon available space below this percentage + fmt_desc: Raise ``HEALTH_ERR`` status when the filesystem that houses a + monitor's data store reports that its available capacity is + less than or equal to this percentage. + default: 5 + services: + - mon + with_legacy: true +- name: mon_data_avail_warn + type: int + level: advanced + desc: issue MON_DISK_LOW health warning when mon available space below this percentage + fmt_desc: Raise ``HEALTH_WARN`` status when the filesystem that houses a + monitor's data store reports that its available capacity is + less than or equal to this percentage . + default: 30 + services: + - mon + with_legacy: true +- name: mon_data_size_warn + type: size + level: advanced + desc: issue MON_DISK_BIG health warning when mon database is above this size + fmt_desc: Raise ``HEALTH_WARN`` status when a monitor's data + store grows to be larger than this size, 15GB by default. + default: 15_G + services: + - mon + with_legacy: true +- name: mon_daemon_bytes + type: size + level: advanced + desc: max bytes of outstanding mon messages mon will read off the network + fmt_desc: The message memory cap for metadata server and OSD messages (in bytes). + default: 400_M + services: + - mon + with_legacy: true +- name: mon_election_timeout + type: float + level: advanced + desc: maximum time for a mon election (seconds) + fmt_desc: On election proposer, maximum waiting time for all ACKs in seconds. + default: 5 + services: + - mon + with_legacy: true +- name: mon_election_default_strategy + type: uint + level: advanced + desc: The election strategy to set when constructing the first monmap. + default: 1 + min: 1 + max: 3 +- name: mon_lease + type: float + level: advanced + desc: lease interval between quorum monitors (seconds) + long_desc: This setting controls how sensitive your mon quorum is to intermittent + network issues or other failures. + fmt_desc: The length (in seconds) of the lease on the monitor's versions. + default: 5 + services: + - mon + with_legacy: true +- name: mon_lease_renew_interval_factor + type: float + level: advanced + desc: multiple of mon_lease for the lease renewal interval + long_desc: Leases must be renewed before they time out. A smaller value means frequent + renewals, while a value close to 1 makes a lease expiration more likely. + fmt_desc: | + ``mon_lease`` \* ``mon_lease_renew_interval_factor`` will be the + interval for the Leader to renew the other monitor's leases. The + factor should be less than ``1.0``. + default: 0.6 + services: + - mon + see_also: + - mon_lease + min: 0 + max: 0.9999999 + with_legacy: true +- name: mon_lease_ack_timeout_factor + type: float + level: advanced + desc: multiple of mon_lease for the lease ack interval before calling new election + fmt_desc: The Leader will wait ``mon_lease`` \* ``mon_lease_ack_timeout_factor`` + for the Providers to acknowledge the lease extension. + default: 2 + services: + - mon + see_also: + - mon_lease + min: 1.0001 + max: 100 + with_legacy: true +- name: mon_accept_timeout_factor + type: float + level: advanced + desc: multiple of mon_lease for follower mons to accept proposed state changes before + calling a new election + fmt_desc: The Leader will wait ``mon_lease`` \* ``mon_accept_timeout_factor`` + for the Requester(s) to accept a Paxos update. It is also used + during the Paxos recovery phase for similar purposes. + default: 2 + services: + - mon + see_also: + - mon_lease + with_legacy: true +- name: mon_elector_ping_timeout + type: float + level: advanced + desc: The time after which a ping 'times out' and a connection is considered down + default: 2 + services: + - mon + see_also: + - mon_elector_ping_divisor +- name: mon_elector_ping_divisor + type: uint + level: advanced + desc: We will send a ping up to this many times per timeout per + default: 2 + services: + - mon + see_also: + - mon_elector_ping_timeout +- name: mon_con_tracker_persist_interval + type: uint + level: advanced + desc: how many updates the ConnectionTracker takes before it persists to disk + default: 10 + services: + - mon + min: 1 + max: 100000 +- name: mon_con_tracker_score_halflife + type: uint + level: advanced + desc: The 'halflife' used when updating/calculating peer connection scores + default: 43200 + services: + - mon + min: 60 +- name: mon_elector_ignore_propose_margin + type: float + level: advanced + desc: The difference in connection score allowed before a peon stops ignoring out-of-quorum + PROPOSEs + default: 0.0005 + services: + - mon +- name: mon_warn_on_cache_pools_without_hit_sets + type: bool + level: advanced + desc: issue CACHE_POOL_NO_HIT_SET health warning for cache pools that do not have + hit sets configured + fmt_desc: Raise ``HEALTH_WARN`` when a cache pool does not have the ``hit_set_type`` + value configured. See :ref:`hit_set_type <hit_set_type>` for more details. + default: true + services: + - mon + with_legacy: true +- name: mon_warn_on_pool_pg_num_not_power_of_two + type: bool + level: dev + desc: issue POOL_PG_NUM_NOT_POWER_OF_TWO warning if pool has a non-power-of-two + pg_num value + default: true + services: + - mon +- name: mon_allow_pool_size_one + type: bool + level: advanced + desc: allow configuring pool with no replicas + default: false + services: + - mon +- name: mon_warn_on_crush_straw_calc_version_zero + type: bool + level: advanced + desc: issue OLD_CRUSH_STRAW_CALC_VERSION health warning if the CRUSH map's straw_calc_version + is zero + fmt_desc: Raise ``HEALTH_WARN`` when the CRUSH ``straw_calc_version`` is zero. See + :ref:`CRUSH map tunables <crush-map-tunables>` for details. + default: true + services: + - mon + with_legacy: true +- name: mon_warn_on_pool_no_redundancy + type: bool + level: advanced + desc: Issue a health warning if any pool is configured with no replicas + fmt_desc: Raise ``HEALTH_WARN`` if any pool is configured with no replicas. + default: true + services: + - mon + see_also: + - osd_pool_default_size + - osd_pool_default_min_size +- name: mon_warn_on_osd_down_out_interval_zero + type: bool + level: advanced + desc: issue OSD_NO_DOWN_OUT_INTERVAL health warning if mon_osd_down_out_interval + is zero + long_desc: Having mon_osd_down_out_interval set to 0 means that down OSDs are not + marked out automatically and the cluster does not heal itself without administrator + intervention. + fmt_desc: Raise ``HEALTH_WARN`` when ``mon_osd_down_out_interval`` is zero. Having this + option set to zero on the leader acts much like the ``noout`` flag. It's hard to figure + out what's going wrong with clusters without the ``noout`` flag set but acting like that + just the same, so we report a warning in this case. + default: true + services: + - mon + see_also: + - mon_osd_down_out_interval + with_legacy: true +- name: mon_warn_on_legacy_crush_tunables + type: bool + level: advanced + desc: issue OLD_CRUSH_TUNABLES health warning if CRUSH tunables are older than mon_crush_min_required_version + fmt_desc: Raise ``HEALTH_WARN`` when CRUSH tunables are too old (older than ``mon_min_crush_required_version``) + default: true + services: + - mon + see_also: + - mon_crush_min_required_version + with_legacy: true +- name: mon_crush_min_required_version + type: str + level: advanced + desc: minimum ceph release to use for mon_warn_on_legacy_crush_tunables + fmt_desc: The minimum tunable profile required by the cluster. See + :ref:`CRUSH map tunables <crush-map-tunables>` for details. + default: hammer + services: + - mon + see_also: + - mon_warn_on_legacy_crush_tunables + with_legacy: true +- name: mon_warn_on_degraded_stretch_mode + type: bool + level: advanced + desc: Issue a health warning if we are in degraded stretch mode + default: true + services: + - mon +- name: mon_stretch_cluster_recovery_ratio + type: float + level: advanced + desc: the ratio of up OSDs at which a degraded stretch cluster enters recovery + default: 0.6 + services: + - mon + min: 0.51 + max: 1 +- name: mon_stretch_recovery_min_wait + type: float + level: advanced + desc: how long the monitors wait before considering fully-healthy PGs as evidence + the stretch mode is repaired + default: 15 + services: + - mon + min: 1 +- name: mon_stretch_pool_size + type: uint + level: dev + default: 4 + services: + - mon + min: 3 + max: 6 +- name: mon_stretch_pool_min_size + type: uint + level: dev + default: 2 + services: + - mon + min: 2 + max: 4 +- name: mon_clock_drift_allowed + type: float + level: advanced + desc: allowed clock drift (in seconds) between mons before issuing a health warning + default: 0.05 + services: + - mon + with_legacy: true +# exponential backoff for clock drift warnings +- name: mon_clock_drift_warn_backoff + type: float + level: advanced + desc: exponential backoff factor for logging clock drift warnings in the cluster + log + default: 5 + services: + - mon + with_legacy: true +# on leader, timecheck (clock drift check) interval (seconds) +- name: mon_timecheck_interval + type: float + level: advanced + desc: frequency of clock synchronization checks between monitors (seconds) + fmt_desc: The time check interval (clock drift check) in seconds + for the Leader. + default: 5_min + services: + - mon + with_legacy: true +# on leader, timecheck (clock drift check) interval when in presence of a skew (seconds) +- name: mon_timecheck_skew_interval + type: float + level: advanced + desc: frequency of clock synchronization (re)checks between monitors while clocks + are believed to be skewed (seconds) + fmt_desc: The time check interval (clock drift check) in seconds when in + presence of a skew in seconds for the Leader. + default: 30 + services: + - mon + see_also: + - mon_timecheck_interval + with_legacy: true +# how often (in commits) to stash a full copy of the PaxosService state +- name: paxos_stash_full_interval + type: int + level: advanced + default: 25 + services: + - mon + fmt_desc: How often (in commits) to stash a full copy of the PaxosService state. + Current this setting only affects ``mds``, ``mon``, ``auth`` and ``mgr`` + PaxosServices. + with_legacy: true +# max paxos iterations before we must first sync the monitor stores +- name: paxos_max_join_drift + type: int + level: advanced + default: 10 + services: + - mon + fmt_desc: The maximum Paxos iterations before we must first sync the + monitor data stores. When a monitor finds that its peer is too + far ahead of it, it will first sync with data stores before moving + on. + with_legacy: true +# gather updates for this long before proposing a map update +- name: paxos_propose_interval + type: float + level: advanced + default: 1 + services: + - mon + fmt_desc: Gather updates for this time interval before proposing + a map update. + with_legacy: true +# min time to gather updates for after period of inactivity +- name: paxos_min_wait + type: float + level: advanced + default: 0.05 + services: + - mon + fmt_desc: The minimum amount of time to gather updates after a period of + inactivity. + with_legacy: true +# minimum number of paxos states to keep around +- name: paxos_min + type: int + level: advanced + default: 500 + services: + - mon + fmt_desc: The minimum number of Paxos states to keep around + with_legacy: true +# number of extra proposals tolerated before trimming +- name: paxos_trim_min + type: int + level: advanced + default: 250 + services: + - mon + fmt_desc: Number of extra proposals tolerated before trimming + with_legacy: true +# maximum amount of versions to trim during a single proposal (0 disables it) +- name: paxos_trim_max + type: int + level: advanced + default: 500 + services: + - mon + fmt_desc: The maximum number of extra proposals to trim at a time + with_legacy: true +# minimum amount of versions to trigger a trim (0 disables it) +- name: paxos_service_trim_min + type: uint + level: advanced + default: 250 + services: + - mon + fmt_desc: The minimum amount of versions to trigger a trim (0 disables it) + with_legacy: true +# maximum amount of versions to trim during a single proposal (0 disables it) +- name: paxos_service_trim_max + type: uint + level: advanced + default: 500 + services: + - mon + fmt_desc: The maximum amount of versions to trim during a single proposal (0 disables it) + with_legacy: true +- name: paxos_service_trim_max_multiplier + type: uint + level: advanced + desc: factor by which paxos_service_trim_max will be multiplied to get a new upper + bound when trim sizes are high (0 disables it) + default: 20 + services: + - mon + min: 0 + flags: + - runtime +- name: paxos_kill_at + type: int + level: dev + default: 0 + services: + - mon + with_legacy: true +- name: mon_auth_validate_all_caps + type: bool + level: advanced + desc: Whether to parse non-monitor capabilities set by the 'ceph auth ...' commands. + Disabling this saves CPU on the monitor, but allows invalid capabilities to be + set, and only be rejected later, when they are used. + default: true + services: + - mon + flags: + - runtime +# force mon to trim mdsmaps to this point (dangerous) +- name: mon_mds_force_trim_to + type: int + level: dev + desc: force mons to trim mdsmaps/fsmaps up to this epoch + fmt_desc: Force monitor to trim mdsmaps up to but not including this FSMap + epoch. A value of 0 disables (the default) this config. This command is + potentially dangerous, use with care. + default: 0 + services: + - mon + with_legacy: true +- name: mds_beacon_mon_down_grace + type: secs + level: advanced + desc: tolerance in seconds for missed MDS beacons to monitors + fmt_desc: The interval without beacons before Ceph declares an MDS laggy + when a monitor is down. + default: 1_min +# skip safety assertions on FSMap (in case of bugs where we want to continue anyway) +- name: mon_mds_skip_sanity + type: bool + level: advanced + desc: skip sanity checks on fsmap/mdsmap + fmt_desc: Skip safety assertions on FSMap (in case of bugs where we want to + continue anyway). Monitor terminates if the FSMap sanity check + fails, but we can disable it by enabling this option. + default: false + services: + - mon + with_legacy: true +- name: mon_mds_blocklist_interval + type: float + level: dev + desc: Duration in seconds that blocklist entries for MDS daemons remain in the OSD + map + fmt_desc: The blocklist duration for failed MDSs in the OSD map. Note, + this controls how long failed MDS daemons will stay in the + OSDMap blocklist. It has no effect on how long something is + blocklisted when the administrator blocklists it manually. For + example, ``ceph osd blocklist add`` will still use the default + blocklist time. + default: 1_day + services: + - mon + min: 1_hr + flags: + - runtime +- name: mon_mgr_blocklist_interval + type: float + level: dev + desc: Duration in seconds that blocklist entries for mgr daemons remain in the OSD + map + default: 1_day + services: + - mon + min: 1_hr + flags: + - runtime +- name: mon_osd_laggy_halflife + type: int + level: advanced + desc: halflife of OSD 'lagginess' factor + fmt_desc: The number of seconds laggy estimates will decay. + default: 1_hr + services: + - mon + with_legacy: true +- name: mon_osd_laggy_weight + type: float + level: advanced + desc: how heavily to weight OSD marking itself back up in overall laggy_probability + long_desc: 1.0 means that an OSD marking itself back up (because it was marked down + but not actually dead) means a 100% laggy_probability; 0.0 effectively disables + tracking of laggy_probability. + fmt_desc: The weight for new samples in laggy estimation decay. + default: 0.3 + services: + - mon + min: 0 + max: 1 + with_legacy: true +- name: mon_osd_laggy_max_interval + type: int + level: advanced + desc: cap value for period for OSD to be marked for laggy_interval calculation + fmt_desc: Maximum value of ``laggy_interval`` in laggy estimations (in seconds). + Monitor uses an adaptive approach to evaluate the ``laggy_interval`` of + a certain OSD. This value will be used to calculate the grace time for + that OSD. + default: 5_min + services: + - mon + with_legacy: true +- name: mon_osd_adjust_heartbeat_grace + type: bool + level: advanced + desc: increase OSD heartbeat grace if peers appear to be laggy + long_desc: If an OSD is marked down but then marks itself back up, it implies it + wasn't actually down but was unable to respond to heartbeats. If this option + is true, we can use the laggy_probability and laggy_interval values calculated + to model this situation to increase the heartbeat grace period for this OSD so + that it isn't marked down again. laggy_probability is an estimated probability + that the given OSD is down because it is laggy (not actually down), and laggy_interval + is an estiate on how long it stays down when it is laggy. + fmt_desc: If set to ``true``, Ceph will scale based on laggy estimations. + default: true + services: + - mon + see_also: + - mon_osd_laggy_halflife + - mon_osd_laggy_weight + - mon_osd_laggy_max_interval + with_legacy: true +- name: mon_osd_adjust_down_out_interval + type: bool + level: advanced + desc: increase the mon_osd_down_out_interval if an OSD appears to be laggy + fmt_desc: If set to ``true``, Ceph will scaled based on laggy estimations. + default: true + services: + - mon + see_also: + - mon_osd_adjust_heartbeat_grace + with_legacy: true +- name: mon_osd_auto_mark_in + type: bool + level: advanced + desc: mark any OSD that comes up 'in' + fmt_desc: Ceph will mark any booting Ceph OSD Daemons as ``in`` + the Ceph Storage Cluster. + default: false + services: + - mon + with_legacy: true +- name: mon_osd_auto_mark_auto_out_in + type: bool + level: advanced + desc: mark any OSD that comes up that was automatically marked 'out' back 'in' + fmt_desc: Ceph will mark booting Ceph OSD Daemons auto marked ``out`` + of the Ceph Storage Cluster as ``in`` the cluster. + default: true + services: + - mon + see_also: + - mon_osd_down_out_interval + with_legacy: true +- name: mon_osd_auto_mark_new_in + type: bool + level: advanced + desc: mark any new OSD that comes up 'in' + fmt_desc: Ceph will mark booting new Ceph OSD Daemons as ``in`` the + Ceph Storage Cluster. + default: true + services: + - mon + with_legacy: true +- name: mon_osd_destroyed_out_interval + type: int + level: advanced + desc: mark any OSD 'out' that has been 'destroy'ed for this long (seconds) + default: 10_min + services: + - mon + with_legacy: true +- name: mon_osd_down_out_interval + type: int + level: advanced + desc: mark any OSD 'out' that has been 'down' for this long (seconds) + fmt_desc: The number of seconds Ceph waits before marking a Ceph OSD Daemon + ``down`` and ``out`` if it doesn't respond. + default: 10_min + services: + - mon + with_legacy: true +- name: mon_osd_down_out_subtree_limit + type: str + level: advanced + desc: do not automatically mark OSDs 'out' if an entire subtree of this size is + down + fmt_desc: The smallest :term:`CRUSH` unit type that Ceph will **not** + automatically mark out. For instance, if set to ``host`` and if + all OSDs of a host are down, Ceph will not automatically mark out + these OSDs. + default: rack + services: + - mon + see_also: + - mon_osd_down_out_interval + flags: + - runtime +- name: mon_osd_min_up_ratio + type: float + level: advanced + desc: do not automatically mark OSDs 'out' if fewer than this many OSDs are 'up' + fmt_desc: The minimum ratio of ``up`` Ceph OSD Daemons before Ceph will + mark Ceph OSD Daemons ``down``. + default: 0.3 + services: + - mon + see_also: + - mon_osd_down_out_interval + with_legacy: true +- name: mon_osd_min_in_ratio + type: float + level: advanced + desc: do not automatically mark OSDs 'out' if fewer than this many OSDs are 'in' + fmt_desc: The minimum ratio of ``in`` Ceph OSD Daemons before Ceph will + mark Ceph OSD Daemons ``out``. + default: 0.75 + services: + - mon + see_also: + - mon_osd_down_out_interval + with_legacy: true +- name: mon_osd_warn_op_age + type: float + level: advanced + desc: issue REQUEST_SLOW health warning if OSD ops are slower than this age (seconds) + default: 32 + services: + - mgr + with_legacy: true +- name: mon_osd_warn_num_repaired + type: uint + level: advanced + desc: issue OSD_TOO_MANY_REPAIRS health warning if an OSD has more than this many + read repairs + default: 10 + services: + - mon +- name: mon_osd_prime_pg_temp + type: bool + level: dev + desc: minimize peering work by priming pg_temp values after a map change + fmt_desc: Enables or disables priming the PGMap with the previous OSDs when an ``out`` + OSD comes back into the cluster. With the ``true`` setting, clients + will continue to use the previous OSDs until the newly ``in`` OSDs for + a PG have peered. + default: true + services: + - mon + with_legacy: true +- name: mon_osd_prime_pg_temp_max_time + type: float + level: dev + desc: maximum time to spend precalculating PG mappings on map change (seconds) + fmt_desc: How much time in seconds the monitor should spend trying to prime the + PGMap when an out OSD comes back into the cluster. + default: 0.5 + services: + - mon + with_legacy: true +- name: mon_osd_prime_pg_temp_max_estimate + type: float + level: advanced + desc: calculate all PG mappings if estimated fraction of PGs that change is above + this amount + fmt_desc: Maximum estimate of time spent on each PG before we prime all PGs + in parallel. + default: 0.25 + services: + - mon + with_legacy: true +- name: mon_osd_blocklist_default_expire + type: float + level: advanced + desc: Duration in seconds that blocklist entries for clients remain in the OSD map + default: 1_hr + services: + - mon + with_legacy: true +- name: mon_osd_crush_smoke_test + type: bool + level: advanced + desc: perform a smoke test on any new CRUSH map before accepting changes + default: true + services: + - mon + with_legacy: true +- name: mon_smart_report_timeout + type: uint + level: advanced + desc: Timeout (in seconds) for smartctl to run, default is set to 5 + default: 5 + services: + - mon +- name: mon_warn_on_older_version + type: bool + level: advanced + desc: issue DAEMON_OLD_VERSION health warning if daemons are not all running the + same version + default: true + services: + - mon +- name: mon_warn_older_version_delay + type: secs + level: advanced + desc: issue DAEMON_OLD_VERSION health warning after this amount of time has elapsed + default: 7_day + services: + - mon +- name: mon_data + type: str + level: advanced + desc: path to mon database + fmt_desc: The monitor's data location. + default: /var/lib/ceph/mon/$cluster-$id + services: + - mon + flags: + - no_mon_update + with_legacy: true +- name: mon_rocksdb_options + type: str + level: advanced + default: write_buffer_size=33554432,compression=kNoCompression,level_compaction_dynamic_level_bytes=true + with_legacy: true +- name: mon_enable_op_tracker + type: bool + level: advanced + desc: enable/disable MON op tracking + default: true + services: + - mon +# compact leveldb on ceph-mon start +- name: mon_compact_on_start + type: bool + level: advanced + default: false + services: + - mon + fmt_desc: Compact the database used as Ceph Monitor store on + ``ceph-mon`` start. A manual compaction helps to shrink the + monitor database and improve the performance of it if the regular + compaction fails to work. + with_legacy: true +# trigger leveldb compaction on bootstrap +- name: mon_compact_on_bootstrap + type: bool + level: advanced + default: false + services: + - mon + fmt_desc: Compact the database used as Ceph Monitor store + on bootstrap. Monitors probe each other to establish + a quorum after bootstrap. If a monitor times out before joining the + quorum, it will start over and bootstrap again. + with_legacy: true +# compact (a prefix) when we trim old states +- name: mon_compact_on_trim + type: bool + level: advanced + default: true + services: + - mon + fmt_desc: Compact a certain prefix (including paxos) when we trim its old states. + with_legacy: true +- name: mon_op_complaint_time + type: secs + level: advanced + desc: time after which to consider a monitor operation blocked after no updates + default: 30 + services: + - mon +- name: mon_op_log_threshold + type: int + level: advanced + desc: max number of slow ops to display + default: 5 + services: + - mon +- name: mon_op_history_size + type: uint + level: advanced + desc: max number of completed ops to track + default: 20 + services: + - mon +- name: mon_op_history_duration + type: secs + level: advanced + desc: expiration time in seconds of historical MON OPS + default: 10_min + services: + - mon +- name: mon_op_history_slow_op_size + type: uint + level: advanced + desc: max number of slow historical MON OPS to keep + default: 20 + services: + - mon +- name: mon_op_history_slow_op_threshold + type: secs + level: advanced + desc: duration of an op to be considered as a historical slow op + default: 10 + services: + - mon +- name: mon_osdmap_full_prune_enabled + type: bool + level: advanced + desc: enables pruning full osdmap versions when we go over a given number of maps + default: true + services: + - mon + see_also: + - mon_osdmap_full_prune_min + - mon_osdmap_full_prune_interval + - mon_osdmap_full_prune_txsize +- name: mon_osdmap_full_prune_min + type: uint + level: advanced + desc: minimum number of versions in the store to trigger full map pruning + default: 10000 + services: + - mon + see_also: + - mon_osdmap_full_prune_enabled + - mon_osdmap_full_prune_interval + - mon_osdmap_full_prune_txsize +- name: mon_osdmap_full_prune_interval + type: uint + level: advanced + desc: interval between maps that will not be pruned; maps in the middle will be + pruned. + default: 10 + services: + - mon + see_also: + - mon_osdmap_full_prune_enabled + - mon_osdmap_full_prune_interval + - mon_osdmap_full_prune_txsize +- name: mon_osdmap_full_prune_txsize + type: uint + level: advanced + desc: number of maps we will prune per iteration + default: 100 + services: + - mon + see_also: + - mon_osdmap_full_prune_enabled + - mon_osdmap_full_prune_interval + - mon_osdmap_full_prune_txsize +- name: mon_osd_cache_size + type: int + level: advanced + desc: maximum number of OSDMaps to cache in memory + fmt_desc: The size of osdmaps cache, not to rely on underlying store's cache + default: 500 + services: + - mon + with_legacy: true +- name: mon_osd_cache_size_min + type: size + level: advanced + desc: The minimum amount of bytes to be kept mapped in memory for osd monitor caches. + fmt_desc: The minimum amount of bytes to be kept mapped in memory for osd + monitor caches. + default: 128_M + services: + - mon + with_legacy: true +- name: mon_osd_mapping_pgs_per_chunk + type: int + level: dev + desc: granularity of PG placement calculation background work + fmt_desc: We calculate the mapping from placement group to OSDs in chunks. + This option specifies the number of placement groups per chunk. + default: 4096 + services: + - mon + with_legacy: true +- name: mon_clean_pg_upmaps_per_chunk + type: uint + level: dev + desc: granularity of PG upmap validation background work + default: 256 + services: + - mon + with_legacy: true +- name: mon_osd_max_creating_pgs + type: int + level: advanced + desc: maximum number of PGs the mon will create at once + default: 1024 + services: + - mon + with_legacy: true +- name: mon_osd_max_initial_pgs + type: int + level: advanced + desc: maximum number of PGs a pool will created with + long_desc: If the user specifies more PGs than this, the cluster will subsequently + split PGs after the pool is created in order to reach the target. + default: 1024 + services: + - mon +- name: mon_memory_target + type: size + level: basic + desc: The amount of bytes pertaining to osd monitor caches and kv cache to be kept + mapped in memory with cache auto-tuning enabled + fmt_desc: The amount of bytes pertaining to OSD monitor caches and KV cache + to be kept mapped in memory with cache auto-tuning enabled. + default: 2_G + services: + - mon + flags: + - runtime + with_legacy: true +- name: mon_memory_autotune + type: bool + level: basic + desc: Autotune the cache memory being used for osd monitors and kv database + fmt_desc: Autotune the cache memory used for OSD monitors and KV + database. + default: true + services: + - mon + flags: + - runtime + with_legacy: true +- name: mon_cpu_threads + type: int + level: advanced + desc: worker threads for CPU intensive background work + fmt_desc: Number of threads for performing CPU intensive work on monitor. + default: 4 + services: + - mon + with_legacy: true +- name: mon_tick_interval + type: int + level: advanced + desc: interval for internal mon background checks + fmt_desc: A monitor's tick interval in seconds. + default: 5 + services: + - mon + with_legacy: true +- name: mon_session_timeout + type: int + level: advanced + desc: close inactive mon client connections after this many seconds + fmt_desc: Monitor will terminate inactive sessions stay idle over this + time limit. + default: 5_min + services: + - mon + with_legacy: true +- name: mon_subscribe_interval + type: float + level: dev + desc: subscribe interval for pre-jewel clients + fmt_desc: The refresh interval (in seconds) for subscriptions. The + subscription mechanism enables obtaining cluster maps + and log information. + default: 1_day + services: + - mon + with_legacy: true +- name: mon_use_min_delay_socket + type: bool + level: advanced + default: false + desc: priority packets between mons + with_legacy: true + see_also: + - osd_heartbeat_use_min_delay_socket |