Prometheus Metrics Reference¶
Complete catalog of metrics exposed by SenHub Agent on the /metrics
endpoint, grouped by probe. Names follow the OTel → Prometheus
conversion rules — see the Prometheus guide
for the underlying naming convention.
Reading the tables: - Prometheus name is what you query in PromQL / VictoriaMetrics. - Type is
gauge(snapshot value),counter(monotonically increasing, userate()), orupdowncounter(can go up and down — emitted as a gauge in Prometheus). - Key labels lists labels specific to this metric (in addition to the systematicprobe_name,probe_typeandcustom_tagslabels).
Agent self-observability¶
Emitted when the prometheus endpoint is active, regardless of
configured probes. A counter appears once it has counted something, and
the OTLP receiver counters once that probe has received traffic.
| Prometheus name | Type | Description | Key labels |
|---|---|---|---|
senhub_agent_uptime_seconds |
gauge | Process uptime since start | – |
senhub_agent_cache_entries |
gauge | Distinct time series in shared cache | – |
senhub_agent_probes_active |
gauge | Probes that have emitted ≥1 datapoint in cache window | – |
senhub_agent_probe_count |
gauge | Configured probes currently running | – |
senhub_agent_probes_healthy |
gauge | Probes reporting IsHealthy() == true |
– |
senhub_agent_collect_errors_total |
counter | Probe collection errors since start, per probe type and failure reason | probe, reason |
senhub_agent_transformer_fallback_total |
counter | Datapoints processed without a transformer definition (no unit injection or corrections) | – |
senhub_agent_otlp_receiver_ingested_total |
counter | Items accepted by the OTLP receiver, per signal (metrics: emitted internal datapoints after family expansion — a summary counts as its count/sum/quantile points; logs: records; traces: spans) | signal |
senhub_agent_otlp_receiver_received_total |
counter | Records relayed by the OTLP receiver, by signal, origin and sending service | signal, origin, senhub_otlp_receiver_sender_service_name |
senhub_agent_otlp_receiver_received_without_host_id_total |
counter | Relayed records that still had no host.id afterwards; expected on origin=remote, a defect on uds |
signal, origin, senhub_otlp_receiver_sender_service_name |
senhub_agent_otlp_receiver_dropped_total |
counter | Items the OTLP receiver discarded, by signal and reason (no_sink, unmapped) |
signal, reason |
senhub_agent_http_requests_total |
counter | HTTP requests served per route template | endpoint |
senhub_agent_build_info |
gauge (=1) | Agent build metadata | version, commit |
senhub_agent_collect_errors_total carries two bounded labels so a spike can be
attributed without log correlation:
probe— the probe type (cpu,redfish,mysql, …), not the per-instance name, so cardinality stays bounded by the registry.reason—collect(aCollect()cycle failed),timeout(that failure was a deadline/timeout) orroute(routing the collected data to an output strategy failed).
Each series appears only once its first error occurs; sum across labels for the
lifetime total: sum(senhub_agent_collect_errors_total).
OTLP output¶
Emitted when an OTLP output is configured.
| Prometheus name | Type | Description |
|---|---|---|
senhub_agent_otlp_active_endpoint_index |
gauge | Index of the OTLP endpoint currently serving exports: 0 = primary, >0 = a configured fallback (endpoint failover, #217). Always 0 when no fallback_endpoints are configured. |
senhub_agent_otlp_buffer_fill_ratio |
gauge | Highest fill ratio (0..1) across all active log channel subscriptions at scrape time. Approaches 1 means the consumer is falling behind producers. |
senhub_agent_otlp_checkpoint_last_save_age_seconds |
gauge | Seconds since the most recent successful OTLP checkpoint save. 0 when persistence disabled or no save has succeeded yet. |
senhub_agent_otlp_checkpoint_restored_entries |
gauge | Number of entries restored from the OTLP checkpoint at the last agent boot. 0 means no restore (no file present, persistence disabled, or fresh install). |
senhub_agent_otlp_checkpoint_size_bytes |
gauge | Size in bytes of the most recently written OTLP checkpoint file. 0 when persistence disabled or no save has succeeded yet. |
senhub_agent_otlp_dropped_log_records_total |
counter | Cumulative count of log records dropped due to subscriber backpressure on the agent log channel. |
senhub_agent_otlp_dropped_span_batches_total |
counter | Cumulative count of received span batches dropped due to backpressure on the agent span channel (the trace relay could not keep up). |
senhub_agent_otlp_endpoint_switches_total |
counter | Cumulative endpoint failover switches (the active OTLP endpoint changed). Rising means the primary is flapping. |
senhub_agent_otlp_export_duration_seconds |
gauge | Wall-clock duration of the most recent successful OTLP metrics export. |
senhub_agent_otlp_logs_pushed_total |
counter | Cumulative count of log records emitted via OTLP/gRPC. |
senhub_agent_otlp_logs_queue_bytes |
gauge | Bytes currently held by the on-disk logs dead-letter queue. |
senhub_agent_otlp_logs_queue_records |
gauge | Event-log records currently held in the on-disk dead-letter queue (awaiting replay). 0 when the backend is healthy or persistence is disabled. |
senhub_agent_otlp_logs_queued_total |
counter | Cumulative event-log records written to the dead-letter queue after a failed export. |
senhub_agent_otlp_logs_relayed_total |
counter | Cumulative count of ingested log records forwarded verbatim, with the emitting application's Resource preserved. Distinct from logs.pushed, which counts records the agent itself produced. |
senhub_agent_otlp_logs_replayed_total |
counter | Cumulative event-log records re-emitted from the dead-letter queue at boot or on backend recovery. |
senhub_agent_otlp_metrics_pushed_total |
counter | Cumulative count of metric records successfully pushed via OTLP/gRPC. |
senhub_agent_otlp_metrics_relayed_total |
counter | Cumulative count of ingested metric points forwarded verbatim, with the emitting application's Resource preserved. Distinct from metrics.pushed, which counts points re-encoded from the agent's own store. |
senhub_agent_otlp_parallel_sub_batches |
gauge | Number of sub-batches the most recent OTLP push fanned out across. 1 means the single-batch path (max_concurrent_exports=1 or cycle below threshold); >1 means per-probe parallel export fired. |
senhub_agent_otlp_spans_relayed_total |
counter | Cumulative count of received spans forwarded verbatim by the trace relay. Pairs with the receiver's ingested{signal=traces} counter: equal totals mean every ingested span left the agent. |
senhub_agent_otlp_store_size |
gauge | Number of distinct series held in the OTLP strategy's last-writer-wins metric store at the last push. |
Agent process¶
The agent's own process, always emitted.
| Prometheus name | Type | Description |
|---|---|---|
senhub_agent_process_cpu_time_seconds_total |
counter | Cumulative CPU time consumed by the agent process since startup, in seconds. |
senhub_agent_process_gc_cycles_total |
counter | Cumulative number of Go garbage collection cycles. Rate over time exposes GC pressure. |
senhub_agent_process_goroutines |
gauge | Number of goroutines currently running in the agent process. Monotonic growth typically indicates a goroutine leak. |
senhub_agent_process_memory_heap_bytes |
gauge | Go runtime heap memory currently allocated for objects, in bytes. Useful with process.memory.resident to spot heap-vs-OS memory leaks. |
senhub_agent_process_memory_resident_bytes |
gauge | Resident set size of the agent process, in bytes (OS-reported VmRSS / WorkingSetSize). 0 if the running OS exposes no such counter. |
senhub_agent_process_open_fds |
gauge | Open file descriptors / Windows handles for the agent process. 0 if the running OS exposes no such counter. |
System probes¶
Probes flagged host_level: true — filtered out when
expose_host_metrics: false is set.
CPU (type: cpu)¶
OTel-aligned via system.cpu.*. Time-mode breakdown collapsed under
cpu_mode label.
| Prometheus name | Type | Unit | Key labels |
|---|---|---|---|
senhub_system_cpu_utilization_ratio |
gauge | 1 (0-1) | cpu_mode ∈ {user, system, idle, nice, iowait, interrupt, softirq, steal}; per-core via cpu_logical_number; senhub_cpu_perfmon_instance on Windows |
senhub_system_cpu_load_1m |
gauge | {thread} | – (Unix-family) |
senhub_system_cpu_load_5m |
gauge | {thread} | – (Unix-family) |
senhub_system_cpu_load_15m |
gauge | {thread} | – (Unix-family) |
senhub_system_cpu_dpcs_per_second |
gauge | 1/s | cpu_logical_number, senhub_cpu_perfmon_instance (Windows) |
senhub_system_cpu_dpcs_queued_per_second |
gauge | 1/s | – |
senhub_system_cpu_interrupts_per_second |
gauge | 1/s | cpu_logical_number, senhub_cpu_perfmon_instance (Windows) |
senhub_system_cpu_queue_length |
gauge | {thread} | – |
senhub_system_processes_count |
gauge | {process} | – |
senhub_system_time_seconds |
gauge | s | – (host clock as a Unix timestamp; time() - senhub_system_time_seconds gives the drift) |
Notes (since 0.1.91):
- Per-mode utilization is exposed as senhub_system_cpu_utilization_ratio with a cpu_mode label, value in [0,1]. On Unix the probe diffs cpu.Times() between two collect cycles; on Windows it reads \Processor\% User Time and siblings directly from PDH. The previous senhub_system_cpu_time_seconds_total counter was retired because Windows could not produce a true cumulative seconds value through PDH — the metric is now honest about being a gauge rate.
- Cross-OS harmonization: Windows privileged_time is mapped to cpu_mode="system" so one PromQL query reflects kernel time on both Linux and Windows.
- Load average dropped the linux. infix in 0.1.91. The metric is emitted on every Unix-family host (Linux, BSD, macOS) — Windows still has no load average concept and does not emit these series.
Memory (type: memory)¶
| Prometheus name | Type | Unit | Key labels |
|---|---|---|---|
senhub_system_memory_limit_bytes |
updowncounter | By | – |
senhub_system_memory_usage_bytes |
updowncounter | By | system_memory_state ∈ |
senhub_system_memory_utilization_ratio |
gauge | 1 (0-1) | – |
senhub_system_paging_utilization_ratio |
gauge | 1 (0-1) | system_paging_state="used" (Windows pagefile) |
senhub_system_paging_utilization_peak_ratio |
gauge | 1 (0-1) | – (Windows pagefile peak) |
senhub_system_paging_faults_per_second |
gauge | 1/s | – (Windows page faults) |
senhub_system_paging_operations_per_second |
gauge | 1/s | direction ∈ {in, out} (Windows pages in/out) |
Notes:
- system_memory_state="free" receives both Linux MemFree and
Windows Available (harmonized — Available ≈ free in Windows
semantics).
- Windows-only states (committed, modified, nonpaged_pool,
paged_pool) are extension values for system_memory_state.
Network (type: network)¶
100 % OTel-native — no senhub.* extensions.
| Prometheus name | Type | Unit | Key labels |
|---|---|---|---|
senhub_system_network_io_bytes_per_second |
gauge | By/s | network_interface_name, network_io_direction ∈ {receive, transmit}, ip (primary IP) |
senhub_system_network_packet_count_per_second |
gauge | {packet}/s | network_interface_name, network_io_direction |
senhub_system_network_errors_per_second |
gauge | {error}/s | network_interface_name, network_io_direction |
senhub_system_network_packet_dropped_per_second |
gauge | {packet}/s | network_interface_name, network_io_direction |
Notes (since 0.1.91):
- These four metrics ship as gauges in _per_second, not cumulative counters. The agent already computes the rate (probe-side delta over the collect interval, or \Network Interface\Bytes Sent/sec on Windows), so wrapping the metric in rate() would be wrong — query the gauge directly.
- Only the primary IP is exposed via the ip label. The positional ip_1, ip_2, … ip_N labels of pre-0.1.91 releases produced unstable cardinality on interfaces with multiple IPv6 addresses and have been dropped.
Logical Disk → Filesystem (type: logicaldisk)¶
The probe type identifier remains logicaldisk for backward compatibility,
but the metrics it emits use the OTel system.filesystem.* convention
(plus senhub.system.disk.* for Windows I/O rates and
senhub.system.filesystem.inode.* for Linux inode counters).
| Prometheus name | Type | Unit | Key labels |
|---|---|---|---|
senhub_system_filesystem_limit_bytes |
updowncounter | By | system_device, system_filesystem_mountpoint, system_filesystem_type |
senhub_system_filesystem_usage_bytes |
updowncounter | By | system_filesystem_state ∈ {used, free, available} + identifiers above |
senhub_system_filesystem_utilization_ratio |
gauge | 1 (0-1) | system_filesystem_state ∈ {used, free} + identifiers |
senhub_system_filesystem_inode_limit |
updowncounter | – | (Linux only) |
senhub_system_filesystem_inode_usage |
updowncounter | – | system_filesystem_state ∈ {used, free} (Linux) |
senhub_system_filesystem_inode_utilization_ratio |
gauge | 1 (0-1) | (Linux) |
senhub_system_disk_operations_per_second |
gauge | 1/s | disk_io_direction ∈ {read, write} (Windows) |
senhub_system_disk_io_bytes_per_second |
gauge | By/s | disk_io_direction (Windows) |
senhub_system_disk_queue_length |
gauge | {operation} | (Windows) |
senhub_system_disk_io_bytes_total |
counter | By | disk_io_direction ∈ {read, write}, system_device (Linux, whole block devices) |
senhub_system_disk_operations_total |
counter | {operation} | disk_io_direction ∈ {read, write}, system_device (Linux) |
senhub_system_disk_io_time_seconds_total |
counter | s | system_device (Linux) |
Notes:
- Windows drive letters and Linux mount points both surface as
system_filesystem_mountpoint.
- The system_filesystem_state="available" extension carries the Linux
statfs.f_bavail (free space for non-root) — distinct from free.
Network probes¶
Ping Gateway (type: ping_gateway) and Ping WebApp (type: ping_webapp)¶
ICMP probes share the same metric names; ping_webapp adds a url_full
label to distinguish targets.
| Prometheus name | Type | Unit | Key labels |
|---|---|---|---|
senhub_probe_icmp_duration_seconds |
gauge | s | url_full (ping_webapp only) |
senhub_probe_icmp_packet_loss_ratio |
gauge | 1 (0-1) | url_full (ping_webapp only) |
Load WebApp (type: load_webapp)¶
HTTP phase timing aligned on blackbox_exporter convention.
| Prometheus name | Type | Unit | Key labels |
|---|---|---|---|
senhub_probe_http_duration_seconds |
gauge | s | phase ∈ {resolve, connect, tls, processing, total}, url_full |
WiFi Signal Strength (type: wifi_signal_strength)¶
| Prometheus name | Type | Unit | Key labels |
|---|---|---|---|
senhub_system_network_wifi_signal_strength_dbm |
gauge | dBm | senhub_network_wifi_ssid, senhub_network_wifi_bssid |
senhub_system_network_wifi_quality_ratio |
gauge | 1 (0-1) | senhub_network_wifi_ssid, senhub_network_wifi_bssid |
Active checks¶
All four check probes share the same semantics: a failing target is a
measurement (up = 0), never a missing series. Durations are seconds.
ICMP Check (type: icmp_check)¶
| Prometheus name | Type | Unit | Key labels |
|---|---|---|---|
senhub_icmp_up |
gauge | bool 0/1 | icmp_target, icmp_target_ip |
senhub_icmp_packet_loss_ratio |
gauge | 1 (0-1) | icmp_target |
senhub_icmp_packets_sent / _received |
gauge | {packet} | icmp_target |
senhub_icmp_rtt_min_seconds / _avg_ / _max_ / _stddev_ |
gauge | s | icmp_target |
HTTP Check (type: http_check)¶
| Prometheus name | Type | Unit | Key labels |
|---|---|---|---|
senhub_httpcheck_up |
gauge | bool 0/1 | httpcheck_target |
senhub_httpcheck_status_code |
gauge | {code} | httpcheck_target |
senhub_httpcheck_duration_seconds |
gauge | s | httpcheck_target |
senhub_httpcheck_duration_dns_seconds / _connect_ / _tls_ / _ttfb_ |
gauge | s | httpcheck_target |
senhub_httpcheck_response_size_bytes |
gauge | By | httpcheck_target |
senhub_httpcheck_tls_expiry |
gauge | days (negative once expired) | httpcheck_target (TLS targets only) |
senhub_httpcheck_content_match |
gauge | bool 0/1 | httpcheck_target (only with content_match) |
TCP Dial (type: tcp_dial)¶
| Prometheus name | Type | Unit | Key labels |
|---|---|---|---|
senhub_tcpdial_up |
gauge | bool 0/1 | tcpdial_target |
senhub_tcpdial_duration_seconds |
gauge | s | tcpdial_target |
DNS Latency (type: dns_latency)¶
One series per (name x resolver) pair.
| Prometheus name | Type | Unit | Key labels |
|---|---|---|---|
senhub_dns_up |
gauge | bool 0/1 | dns_question_name, dns_resolver |
senhub_dns_lookup_duration_seconds |
gauge | s | dns_question_name, dns_resolver |
senhub_dns_answers |
gauge | {answer} | dns_question_name, dns_resolver |
SNMP¶
SNMP Poll (type: snmp_poll)¶
One series per device (snmp_target); interface metrics add
network_interface_index. Counters are raw — compute rates in the
backend (rate(...[5m])).
| Prometheus name | Type | Unit | Key labels |
|---|---|---|---|
senhub_snmp_up |
gauge | bool 0/1 | snmp_target |
senhub_snmp_poll_duration_seconds |
gauge | s | snmp_target |
senhub_snmp_sys_uptime |
gauge | centiseconds (native SNMP unit) | snmp_target |
senhub_snmp_interface_in_octets_bytes_total / _out_ |
counter | By | snmp_target, network_interface_index |
senhub_snmp_interface_in_errors_total / _out_ |
counter | {error} | snmp_target, network_interface_index |
senhub_snmp_interface_in_discards_total / _out_ |
counter | {packet} | snmp_target, network_interface_index |
senhub_snmp_interface_speed_bytes_per_second |
gauge | By/s | snmp_target, network_interface_index |
senhub_snmp_interface_admin_status / _oper_status |
gauge | IF-MIB enum (1=up, 2=down, ...) | snmp_target, network_interface_index |
Custom mappings and dynamic OIDs surface under the name configured in
custom_mappings (typed pass-through; counters gain _total).
SNMP Trap (type: snmp_trap)¶
Trap payloads ride the OTLP log rail; only the receiver's self-metrics appear here.
| Prometheus name | Type | Unit | Key labels |
|---|---|---|---|
senhub_snmp_trap_rejected_community_total |
counter | {datagram} | – |
senhub_snmp_trap_decode_panics_total |
counter | {datagram} | – |
Universal ingestion¶
Prometheus Scrape (type: prometheus_scrape)¶
Scraped series are re-exported under the senhub_ prefix with their
original name and labels; counter/gauge types are preserved (counters
gain _total if absent). Self-metrics:
| Prometheus name | Type | Unit | Key labels |
|---|---|---|---|
senhub_promscrape_up |
gauge | bool 0/1 | promscrape_target |
senhub_promscrape_scrape_duration_seconds |
gauge | s | promscrape_target |
senhub_promscrape_samples |
gauge | {sample} | promscrape_target |
senhub_promscrape_dropped |
gauge | {sample} (histogram/summary series) | promscrape_target |
Exec (type: exec)¶
Perfdata / JSON metrics surface as senhub_exec_<label> pass-through
gauges (or counters with _total for the c UOM), time normalized to
seconds and bytes to bytes. Self-metrics:
| Prometheus name | Type | Unit | Key labels |
|---|---|---|---|
senhub_exec_status |
gauge | 0 ok / 1 warning / 2 critical / 3 unknown | – |
senhub_exec_duration_seconds |
gauge | s | – |
senhub_exec_timeout |
gauge | bool 0/1 | – |
senhub_exec_skipped |
gauge | bool 0/1 (overlap guard) | – |
Event-conduit probes (skipped)¶
The syslog, event, filetail, windows_eventlog and linux_logs
probes are log-flow conduits — they relay records, they don't
produce metric time series here. Their records ship through the OTLP
logs transport (see the OTLP guide); snmp_trap is the same, with
the two self-metrics listed above.
Hardware (Redfish) — type: redfish¶
Mostly OTel-native via the hw.* namespace, plus senhub.hardware.*
extensions for RAID pools and Redfish-specific concepts.
Health (universal hw.status pattern)¶
Every health metric uses strict OTel expansion — one series per
possible hw_state, value 1 if active else 0.
| Prometheus name | hw_type values |
hw_state values |
|---|---|---|
senhub_hw_status |
power_supply, physical_disk, logical_disk, disk_controller, enclosure |
ok, degraded, failed, predicted_failure |
Drive failure prediction is encoded as hw_state="predicted_failure" on
the same metric (independent of overall health). A component whose health
the BMC does not report has every state at 0: the convention defines no
unknown state. An IPMI sensor emits the ok series only.
Drives, volumes, pools (capacity & I/O)¶
| Prometheus name | Type | Unit | Notes |
|---|---|---|---|
senhub_hw_physical_disk_size_bytes |
updowncounter | By | per drive |
senhub_hw_logical_disk_limit_bytes |
updowncounter | By | volume capacity |
senhub_hw_logical_disk_usage_bytes |
updowncounter | By | with hw_logical_disk_state ∈ |
senhub_hw_logical_disk_utilization_ratio |
gauge | 1 | with hw_logical_disk_state |
senhub_hardware_logical_disk_io_operations_total |
counter | {operation} | disk_io_direction ∈ |
senhub_hardware_logical_disk_io_bytes_total |
counter | By | disk_io_direction |
senhub_hardware_logical_disk_encrypted |
gauge | 1 | bool 0/1 |
senhub_hardware_storage_pool_usage_bytes |
updowncounter | By | senhub_hardware_storage_pool_state ∈ |
senhub_hardware_storage_pool_utilization_ratio |
gauge | 1 | senhub_hardware_storage_pool_state ∈ |
senhub_hardware_storage_pool_io_operations_total |
counter | {operation} | disk_io_direction |
senhub_hardware_storage_pool_io_bytes_total |
counter | By | disk_io_direction |
senhub_hardware_physical_disk_link_speed_bytes_per_second |
gauge | By/s | per drive (Gbps × 1e9 / 8) |
senhub_hardware_physical_disk_block_size_bytes |
gauge | By | per drive |
senhub_hardware_physical_disk_operation_progress_ratio |
gauge | 1 | per drive (mapper ÷100) |
senhub_hardware_physical_disk_has_active_operations |
gauge | 1 | bool |
senhub_hardware_physical_disk_location_indicator_active |
gauge | 1 | bool |
System / controllers / redundancy¶
| Prometheus name | Type | Notes |
|---|---|---|
senhub_hardware_system_power_state_ratio |
updowncounter | strict-OTel expand: senhub_hardware_system_power_state_value ∈ |
senhub_hardware_eventservice_status |
updowncounter | with senhub_hardware_eventservice_state ∈ |
senhub_hardware_redundancy_status |
updowncounter | with senhub_hardware_redundancy_state, senhub_hardware_redundancy_scope (chassis/storage) |
senhub_hardware_redundancy_controllers_count |
updowncounter | with senhub_hardware_redundancy_bound ∈ |
Veeam Backup & Replication — type: veeam¶
All metrics under senhub.veeam.* (no OTel semconv for backup).
Jobs¶
| Prometheus name | Type | Notes |
|---|---|---|
senhub_veeam_job_count |
gauge | senhub_veeam_job_type |
senhub_veeam_jobs_by_last_result |
gauge | senhub_veeam_job_last_result ∈ |
senhub_veeam_job_status |
updowncounter | strict-OTel expand: senhub_veeam_job_state ∈ |
senhub_veeam_job_seconds_since_last_run |
gauge (s) | senhub_veeam_job_name, senhub_veeam_job_type |
senhub_veeam_job_objects |
gauge | per job |
senhub_veeam_job_bottleneck_status |
updowncounter | expand: senhub_veeam_job_bottleneck ∈ |
senhub_veeam_job_last_run_bytes |
gauge (By) | senhub_veeam_job_data_phase ∈ |
Repository¶
| Prometheus name | Type | Unit |
|---|---|---|
senhub_veeam_repository_limit_bytes |
updowncounter | By |
senhub_veeam_repository_usage_bytes |
updowncounter | By + senhub_veeam_repository_state ∈ |
senhub_veeam_repository_utilization_ratio |
gauge | 1 + senhub_veeam_repository_state="free" |
License¶
| Prometheus name | Type | Notes |
|---|---|---|
senhub_veeam_license_status |
updowncounter | expand: senhub_veeam_license_state ∈ |
senhub_veeam_license_expiry_seconds |
gauge (s) | – |
senhub_veeam_license_instances |
gauge | senhub_veeam_license_instances_state ∈ |
Proxies, objects, infrastructure¶
| Prometheus name | Type | Notes |
|---|---|---|
senhub_veeam_proxy_status |
updowncounter | expand: senhub_veeam_proxy_state ∈ |
senhub_veeam_proxies |
gauge | senhub_veeam_proxies_state ∈ |
senhub_veeam_object_restore_points |
gauge | per object |
senhub_veeam_object_last_run_failed |
gauge | bool |
senhub_veeam_objects |
gauge | senhub_veeam_objects_state ∈ |
senhub_veeam_server_status |
updowncounter | expand: senhub_veeam_server_state ∈ |
senhub_veeam_servers |
gauge | senhub_veeam_servers_state ∈ |
Citrix Virtual Apps and Desktops — type: citrix¶
All metrics under senhub.citrix.* (no Citrix CVAD OTel convention; design from scratch).
Sessions, machines, license¶
| Prometheus name | Type | Notes |
|---|---|---|
senhub_citrix_sessions_count |
gauge | senhub_citrix_session_state ∈ |
senhub_citrix_machine_count |
gauge | – |
senhub_citrix_machines_by_registration_state |
gauge | senhub_citrix_machine_registration_state ∈ |
senhub_citrix_machines_overloaded |
gauge | – |
senhub_citrix_machine_multi_session_fault_count |
gauge | – |
senhub_citrix_machines_by_fault_state |
gauge | senhub_citrix_machine_fault_state ∈ |
senhub_citrix_license_sessions_active |
gauge | – |
senhub_citrix_license_peak_concurrent_users |
gauge | – |
senhub_citrix_license_unique_users |
gauge | – |
senhub_citrix_license_grace_sessions_remaining |
gauge | – |
senhub_citrix_license_grace_active |
gauge | bool |
senhub_citrix_license_grace_time_remaining_seconds |
gauge | – (mapper hours × 3600) |
Logon performance¶
| Prometheus name | Type | Notes |
|---|---|---|
senhub_citrix_logon_duration_1h_average_seconds |
gauge | – |
senhub_citrix_logon_last_session_duration_seconds |
gauge | – |
senhub_citrix_logon_sessions_opened |
gauge | – |
senhub_citrix_logon_phase_duration_seconds |
gauge | senhub_citrix_logon_phase ∈ |
Connection failures, load index¶
| Prometheus name | Type | Notes |
|---|---|---|
senhub_citrix_connection_failure_count |
gauge | – |
senhub_citrix_connection_failures_by_category |
gauge | senhub_citrix_connection_failure_category ∈ |
senhub_citrix_load_index_ratio |
gauge | senhub_citrix_load_index_dimension ∈ {effective, cpu, memory, disk, network, sessions} (mapper ÷100) |
NetScaler / Citrix ADC — type: netscaler¶
100 metrics organized by NITRO entity, all under senhub.netscaler.*. The
table below is an overview — see the source YAML
(internal/agent/services/data_store/transformers/definitions/netscaler.yaml)
for exhaustive details.
State enums (strict-OTel expand)¶
All vServer/service/servicegroup/csvserver/gslbvserver state metrics share the NITRO state enum mapping (1=down, 2=unknown, 3=busy, 4=out_of_service, 5=trofs, 7=up, 8=trofs_down):
| Metric pattern | strict-OTel expand attribute |
|---|---|
senhub_netscaler_lbvserver_status |
senhub_netscaler_lbvserver_state |
senhub_netscaler_service_status |
senhub_netscaler_service_state |
senhub_netscaler_servicegroup_status |
senhub_netscaler_servicegroup_state |
senhub_netscaler_csvserver_status |
senhub_netscaler_csvserver_state |
senhub_netscaler_gslb_vserver_status |
senhub_netscaler_gslb_vserver_state |
senhub_netscaler_gslb_site_status |
senhub_netscaler_gslb_site_state ∈ |
senhub_netscaler_gslb_service_status |
senhub_netscaler_gslb_service_state ∈ |
senhub_netscaler_aaa_vserver_status |
senhub_netscaler_aaa_vserver_state ∈ |
senhub_netscaler_vpn_vserver_status |
senhub_netscaler_vpn_vserver_state ∈ |
senhub_netscaler_interface_status |
senhub_netscaler_interface_state ∈ |
senhub_netscaler_ssl_certificate_status |
senhub_netscaler_ssl_certificate_state ∈ |
senhub_netscaler_ha_role_ratio |
senhub_netscaler_ha_role_state ∈ |
senhub_netscaler_ha_node_status |
senhub_netscaler_ha_node_state ∈ |
senhub_netscaler_ha_sync_status |
senhub_netscaler_ha_sync_state ∈ |
System (CPU, memory, throughput, HTTP, TCP)¶
| Prometheus name | Type | Unit | Notes |
|---|---|---|---|
senhub_netscaler_system_cpu_utilization_ratio |
gauge | 1 | senhub_netscaler_cpu_plane ∈ |
senhub_netscaler_system_memory_utilization_ratio |
gauge | 1 | – |
senhub_netscaler_system_network_throughput_bytes_per_second |
gauge | By/s | network_io_direction (Mbps × 1e6 / 8) |
senhub_netscaler_system_http_messages_rate_per_second |
gauge | 1/s | senhub_netscaler_http_message_type ∈ |
senhub_netscaler_system_tcp_connections_active |
updowncounter | {connection} | senhub_netscaler_tcp_side ∈ |
senhub_netscaler_ns_throughput_bytes_per_second |
gauge | By/s | senhub_netscaler_traffic_type ∈ |
senhub_netscaler_system_network_packets_rate_per_second |
gauge | 1/s | network_io_direction |
senhub_netscaler_system_network_packets_total |
counter | {packet} | network_io_direction |
senhub_netscaler_system_dpcs_per_second etc. |
– | – | (windows specifics not present on appliance) |
Disk (OTel-native filesystem)¶
The appliance's local disk is mapped to OTel system.filesystem.* —
distinguished from host OS filesystem by the probe_type="netscaler" label.
Load balancing / services / GSLB / cache / compression / AAA / VPN / WAF¶
The remaining metrics follow the same rule set: rx/tx collapsed via
network.io.direction; hits/misses, requests/responses, successes/failures
collapsed via dedicated *_state/*_result/*_type attributes; counters
get _total. See the YAML definition for the full list of ~65 metric
names in this domain.
Discovering metrics at runtime¶
Easiest way to enumerate everything the agent emits in your specific config:
# All metric names (one per line)
curl -sf -H "Authorization: Bearer KEY" http://agent:8080/metrics \
| grep -E '^# TYPE ' | awk '{print $3}' | sort -u
# All series (with their labels)
curl -sf -H "Authorization: Bearer KEY" http://agent:8080/metrics \
| grep -v '^#' | grep -v '^$'
Or query VictoriaMetrics / Prometheus once the scraper has ingested at least one cycle: