Manager Node Metrics

Name	Type	Freq	Labels	Version Introduced	Version Deprecated	Description
yb_mgmt_1h_scrape_duration_seconds	gauge	1h	-	7.4.5	-	Wall-clock time the /metrics-1h scrape took last time it ran.
yb_mgmt_1h_scrape_success	gauge	1h	-	7.4.5	-	Whether the last /metrics-1h scrape completed without error (1 = success); see mgmt-exporter's own logs for which collector failed.
yb_mgmt_1m_scrape_duration_seconds	gauge	1m	-	7.4.5	-	Wall-clock time the /metrics-1m scrape took last time it ran.
yb_mgmt_1m_scrape_success	gauge	1m	-	7.4.5	-	Whether the last /metrics-1m scrape completed without error (1 = success); see mgmt-exporter's own logs for which collector failed.
yb_mgmt_cpu_seconds_total	counter	1m	hostname, mode	7.4.5	-	Cumulative CPU time summed across all cores, by mode (user, nice, system, idle, iowait, irq, softirq, steal). Deliberately aggregate, not per-core -- per-core would multiply this group by the core count for a signal rarely needed on a manager node.
yb_mgmt_drbd_connected	gauge	1m	hostname, resource	7.4.5	-	Whether the DRBD resource is connected to its peer (1 = connected).
yb_mgmt_drbd_disk_up_to_date	gauge	1m	hostname, resource	7.4.5	-	Whether the local DRBD disk state is UpToDate (1 = up to date).
yb_mgmt_drbd_out_of_sync_bytes	gauge	1m	hostname, resource	7.4.5	-	Data known to be out of sync between the DRBD peers.
yb_mgmt_drbd_primary	gauge	1m	hostname, resource	7.4.5	-	Whether this node holds the DRBD primary role for the resource.
yb_mgmt_drive_capacity_bytes	gauge	1h	hostname, role, slot, device, serial	7.4.5	-	Usable capacity of the drive.
yb_mgmt_drive_critical_warning	gauge	1h	hostname, role, slot, device, serial	7.4.5	-	Drive critical warning bitmask (0 = none; bit 0 spare below threshold, bit 1 temperature, bit 2 reliability degraded, bit 3 read-only, bit 4 volatile memory backup failed). NVMe only.
yb_mgmt_drive_error_log_entries_total	counter	1h	hostname, role, slot, device, serial	7.4.5	-	Entries written to the drive's internal error information log. NVMe only.
yb_mgmt_drive_healthy	gauge	1h	hostname, role, slot, device, serial	7.4.5	-	Whether the drive reports itself healthy (1 = healthy, 0 = failing or predicted to fail).
yb_mgmt_drive_info	gauge	1h	hostname, role, slot, device, serial, model, firmware	7.4.5	-	Drive inventory information; the value is always 1 and the detail is carried in labels.
yb_mgmt_drive_media_errors_total	counter	1h	hostname, role, slot, device, serial	7.4.5	-	Unrecovered data integrity errors detected by the drive. NVMe only.
yb_mgmt_drive_power_cycles_total	counter	1h	hostname, role, slot, device, serial	7.4.5	-	Power cycles recorded by the drive. NVMe only.
yb_mgmt_drive_power_on_seconds_total	counter	1h	hostname, role, slot, device, serial	7.4.5	-	Time the drive has been powered on since manufacture. NVMe only (hourly resolution from the drive, converted to seconds).
yb_mgmt_drive_read_bytes_total	counter	1h	hostname, role, slot, device, serial	7.4.5	-	Bytes read from the drive since manufacture. NVMe only.
yb_mgmt_drive_spare_percent	gauge	1h	hostname, role, slot, device, serial	7.4.5	-	Remaining spare capacity as a percentage of the factory value.
yb_mgmt_drive_spare_threshold_percent	gauge	1h	hostname, role, slot, device, serial	7.4.5	-	Spare capacity percentage below which the drive reports a critical warning. NVMe only.
yb_mgmt_drive_temperature_celsius	gauge	1h	hostname, role, slot, device, serial	7.4.5	-	Current drive composite temperature.
yb_mgmt_drive_unsafe_shutdowns_total	counter	1h	hostname, role, slot, device, serial	7.4.5	-	Unsafe shutdowns recorded by the drive. NVMe only.
yb_mgmt_drive_wear_percent	gauge	1h	hostname, role, slot, device, serial	7.4.5	-	Percentage of the drive's rated write endurance consumed (0 = new, 100 = rated life used up).
yb_mgmt_drive_written_bytes_total	counter	1h	hostname, role, slot, device, serial	7.4.5	-	Bytes written to the drive since manufacture. NVMe only.
yb_mgmt_ethmonitor_ok	gauge	1m	hostname, interface	7.4.5	-	Whether the HA ethernet monitor considers the interface healthy; a 0 blocks this node from holding the primary role (1 = healthy).
yb_mgmt_ha_is_primary	gauge	1m	hostname	7.4.5	-	Whether this manager node currently holds the primary role (1 = primary), derived from which node the ybdfs DRBD resource's Master role is running on.
yb_mgmt_ha_maintenance_mode	gauge	1m	hostname	7.4.5	-	Whether the HA cluster is in maintenance mode (1 = maintenance).
yb_mgmt_ha_peer_online	gauge	1m	hostname	7.4.5	-	Whether the peer manager node is online in the HA cluster (1 = online).
yb_mgmt_ha_quorate	gauge	1m	hostname	7.4.5	-	Whether the HA cluster has quorum (1 = quorate).
yb_mgmt_ha_resources_failed	gauge	1m	hostname	7.4.5	-	HA cluster resources currently in a failed or blocked state.
yb_mgmt_ha_standby	gauge	1m	hostname	7.4.5	-	Whether this manager node is in HA standby (1 = standby).
yb_mgmt_ib_port_errors_max	gauge	1m	type	7.4.5	-	The value each yb_mgmt_ib_port_errors_total counter type saturates at, from the field width the InfiniBand spec gives that hardware PMA counter (255 for the link counters, 65535 for the error counters, 4294967295 for PortXmitWait). These counters stick at their maximum rather than wrapping and nothing clears them, so once a port reaches this value its total stays constant and rate()/increase() read zero however many further errors arrive -- a saturated counter is indistinguishable from a healthy one on the total alone. Join on the type label to detect that: a port at its max needs the counter read as "at least this many, no longer counting". Absent entirely on the standby manager, same as yb_mgmt_ib_port_errors_total.
yb_mgmt_ib_port_errors_threshold	gauge	1m	type	7.4.5	-	The count ibqueryerrors requires a counter to exceed before it reports a port on yb_mgmt_ib_port_errors_total at all -- from /etc/infiniband-diags/error_thresholds, the same file the fabric scan itself reads. Zero means no threshold is configured for that counter, so any nonzero value is reported; that is the normal case. Absent entirely on the standby manager, same as yb_mgmt_ib_port_errors_total.
yb_mgmt_ib_port_errors_total	counter	1m	type, component, chassis, slot, desc, guid, port	7.4.5	-	InfiniBand fabric-wide port error counters, by counter type, for every switch/blade/manager port visible from whichever manager currently holds primary (a single vantage point sees the entire fabric). Deliberately NOT hostname-labeled -- describes fabric/blade/switch endpoints, not the collecting manager. Absent entirely on the standby manager. These are hardware PMA counters with narrow fields (8, 16 or 32 bits depending on the counter) that saturate at their maximum instead of wrapping, and nothing clears them, so a saturated counter reports a constant total and rate()/increase() over it read zero even while errors keep arriving; compare against yb_mgmt_ib_port_errors_max to tell a quiet port from a maxed-out one.
yb_mgmt_ib_port_rate_bits	gauge	1m	hostname, device, port	7.4.5	-	Negotiated InfiniBand link rate on this manager node's own local HCA port, in bits per second.
yb_mgmt_ib_port_state	gauge	1m	hostname, device, port	7.4.5	-	InfiniBand port state on this manager node's own local HCA (1 = down, 2 = init, 3 = armed, 4 = active).
yb_mgmt_ib_scrape_duration_seconds	gauge	1m	-	7.4.5	-	Wall-clock time the /metrics-ib scrape took last time it ran.
yb_mgmt_ib_scrape_success	gauge	1m	-	7.4.5	-	Whether the last /metrics-ib scrape completed without error (1 = success); see mgmt-exporter's own logs for which collector failed.
yb_mgmt_load1	gauge	1m	hostname	7.4.5	-	1-minute load average.
yb_mgmt_load15	gauge	1m	hostname	7.4.5	-	15-minute load average.
yb_mgmt_load5	gauge	1m	hostname	7.4.5	-	5-minute load average.
yb_mgmt_md_array_degraded	gauge	1m	hostname, role	7.4.5	-	Whether the software RAID array is running degraded (1 = degraded).
yb_mgmt_md_disks	gauge	1m	hostname, role, device, state	7.4.5	-	Member disks of the software RAID array, by state (active, failed, spare). The value is always 1.
yb_mgmt_md_resync_percent	gauge	1m	hostname, role	7.4.5	-	Rebuild or resync progress of the software RAID array (100 = in sync). Absent (not 0) when no resync or recovery is in progress.
yb_mgmt_memory_available_bytes	gauge	1m	hostname	7.4.5	-	Memory available for new workloads without swapping.
yb_mgmt_memory_buffers_bytes	gauge	1m	hostname	7.4.5	-	Memory used by kernel block-device buffers.
yb_mgmt_memory_cached_bytes	gauge	1m	hostname	7.4.5	-	Memory used by the page cache.
yb_mgmt_memory_ecc_correctable_errors_total	counter	1m	hostname, dimm	7.4.5	-	Correctable ECC memory errors reported by the EDAC subsystem for this DIMM.
yb_mgmt_memory_ecc_uncorrectable_errors_total	counter	1m	hostname, dimm	7.4.5	-	Uncorrectable ECC memory errors reported by the EDAC subsystem for this DIMM.
yb_mgmt_memory_free_bytes	gauge	1m	hostname	7.4.5	-	Unused physical memory on the manager node.
yb_mgmt_memory_total_bytes	gauge	1m	hostname	7.4.5	-	Total usable physical memory on the manager node.
yb_mgmt_network_carrier_changes_total	counter	1m	hostname, device	7.4.5	-	Link state changes seen on the interface since boot.
yb_mgmt_network_receive_bytes_total	counter	1m	hostname, device	7.4.5	-	Bytes received on the interface.
yb_mgmt_network_receive_dropped_total	counter	1m	hostname, device	7.4.5	-	Packets dropped on receive by the interface.
yb_mgmt_network_receive_errors_total	counter	1m	hostname, device	7.4.5	-	Receive errors reported by the interface.
yb_mgmt_network_receive_packets_total	counter	1m	hostname, device	7.4.5	-	Packets received on the interface.
yb_mgmt_network_transmit_bytes_total	counter	1m	hostname, device	7.4.5	-	Bytes transmitted on the interface.
yb_mgmt_network_transmit_dropped_total	counter	1m	hostname, device	7.4.5	-	Packets dropped on transmit by the interface.
yb_mgmt_network_transmit_errors_total	counter	1m	hostname, device	7.4.5	-	Transmit errors reported by the interface.
yb_mgmt_network_transmit_packets_total	counter	1m	hostname, device	7.4.5	-	Packets transmitted on the interface.
yb_mgmt_network_up	gauge	1m	hostname, device	7.4.5	-	Interface operational state (1 = up).
yb_mgmt_node_info	gauge	1m	hostname, hw_model, bios_version, kernel_version, serial	7.4.5	-	Manager node inventory information; the value is always 1 and the detail is carried in labels.
yb_mgmt_oom_kills_total	counter	1m	hostname	7.4.5	-	Processes killed by the kernel out-of-memory killer since the current boot.
yb_mgmt_procs_blocked	gauge	1m	hostname	7.4.5	-	Processes blocked waiting on I/O.
yb_mgmt_swap_free_bytes	gauge	1m	hostname	7.4.5	-	Unused swap space on the manager node.
yb_mgmt_swap_total_bytes	gauge	1m	hostname	7.4.5	-	Total swap space configured on the manager node.
yb_mgmt_system_version_info	gauge	1m	version, gen	7.4.5	-	The YBD software version currently running on the cluster, and the appliance's hardware generation (gen1/gen2); the value is always 1 and both are carried in labels. Published only by whichever manager currently holds the DRBD primary role -- unlike yb_mgmt_node_info this describes the cluster/appliance as a whole, not a specific manager node, so it carries no hostname label. gen lives here rather than on yb_mgmt_node_info because it's an appliance-wide constant (both managers are always the same generation), not a genuinely per-node fact like the others on that metric, and -- like version -- it's only resolvable from /mnt/ybdata/sysconfig.ini, the DRBD volume.
yb_mgmt_uptime_seconds	gauge	1m	hostname	7.4.5	-	Time since the manager node booted.