{ "__inputs": [{"name":"DS_PROMETHEUS","label":"Prometheus","type":"datasource","pluginId":"prometheus","pluginName":"Prometheus"}], "__requires": [ {"type":"grafana","id":"grafana","name":"Grafana","version":"10.0.0"}, {"type":"datasource","id":"prometheus","name":"Prometheus","version":"1.0.0"}, {"type":"panel","id":"stat","name":"Stat","version":""}, {"type":"panel","id":"bargauge","name":"Bar gauge","version":""}, {"type":"panel","id":"timeseries","name":"Time series","version":""}, {"type":"panel","id":"table","name":"Table","version":""} ], "annotations":{"list":[{"builtIn":1,"datasource":{"type":"grafana","uid":"-- Grafana --"},"enable":true,"hide":true,"iconColor":"rgba(0,211,255,1)","name":"Annotations & Alerts","type":"dashboard"}]}, "description":"MetalLB L2 Mode ā€” controller pool metrics + speaker ARP/NDP health, VIP-to-node ownership, failover severity ranking. Scaled for large (15+ node) clusters.", "editable":true,"fiscalYearStartMonth":0,"graphTooltip":1,"id":null,"links":[], "panels":[ { "collapsed":false, "gridPos":{"h":1,"w":24,"x":0,"y":0}, "id":100, "title":"šŸ”µ Cluster Status ā€” Config, Pool & Speaker Health", "type":"row" }, { "datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"}, "description":"Aggregate count of speakers with config loaded. Must equal total speaker count (= node count in L2 mode). If a broken YAML config is applied, this drops below total ā€” speakers stop updating routing tables.", "fieldConfig":{ "defaults":{ "color":{"mode":"thresholds"}, "thresholds":{"mode":"absolute","steps":[{"color":"red","value":null},{"color":"green","value":1}]}, "unit":"short" },"overrides":[] }, "gridPos":{"h":4,"w":6,"x":0,"y":1}, "id":1, "options":{"colorMode":"background","graphMode":"none","justifyMode":"center","orientation":"auto","reduceOptions":{"calcs":["lastNotNull"],"fields":"","values":false},"textMode":"value_and_name"}, "title":"Speakers Config Loaded (sum)", "type":"stat", "targets":[{"datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"},"expr":"sum(metallb_k8s_client_config_loaded_bool)","legendFormat":"loaded","refId":"A"}] }, { "datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"}, "description":"Aggregate count of speakers running stale config. Must be 0. Any value >0 means at least one speaker failed to reload the latest L2Advertisement/IPAddressPool ā€” investigate that node immediately.", "fieldConfig":{ "defaults":{ "color":{"mode":"thresholds"}, "thresholds":{"mode":"absolute","steps":[{"color":"green","value":null},{"color":"red","value":1}]}, "unit":"short" },"overrides":[] }, "gridPos":{"h":4,"w":6,"x":6,"y":1}, "id":2, "options":{"colorMode":"background","graphMode":"none","justifyMode":"center","orientation":"auto","reduceOptions":{"calcs":["lastNotNull"],"fields":"","values":false},"textMode":"value_and_name"}, "title":"Speakers Config Stale (sum)", "type":"stat", "targets":[{"datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"},"expr":"sum(metallb_k8s_client_config_stale_bool)","legendFormat":"stale","refId":"A"}] }, { "datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"}, "description":"Total IPs in use vs total available. From metallb-controller (allocator metrics).", "fieldConfig":{ "defaults":{ "color":{"mode":"thresholds"}, "thresholds":{"mode":"absolute","steps":[{"color":"green","value":null}]}, "unit":"short" },"overrides":[] }, "gridPos":{"h":4,"w":3,"x":12,"y":1}, "id":3, "options":{"colorMode":"value","graphMode":"area","justifyMode":"center","orientation":"auto","reduceOptions":{"calcs":["lastNotNull"],"fields":"","values":false},"textMode":"auto"}, "title":"IPs In Use", "type":"stat", "targets":[{"datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"},"expr":"sum(metallb_allocator_addresses_in_use_total)","legendFormat":"in use","refId":"A"}] }, { "datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"}, "description":"Total usable IPs across all pools (controller allocator).", "fieldConfig":{ "defaults":{ "color":{"mode":"thresholds"}, "thresholds":{"mode":"absolute","steps":[{"color":"blue","value":null}]}, "unit":"short" },"overrides":[] }, "gridPos":{"h":4,"w":3,"x":15,"y":1}, "id":4, "options":{"colorMode":"value","graphMode":"none","justifyMode":"center","orientation":"auto","reduceOptions":{"calcs":["lastNotNull"],"fields":"","values":false},"textMode":"auto"}, "title":"Total Pool IPs", "type":"stat", "targets":[{"datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"},"expr":"sum(metallb_allocator_addresses_total)","legendFormat":"total","refId":"A"}] }, { "datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"}, "description":"Overall pool utilisation %. Red >90%, Yellow >70%.", "fieldConfig":{ "defaults":{ "color":{"mode":"thresholds"}, "thresholds":{"mode":"absolute","steps":[{"color":"green","value":null},{"color":"yellow","value":70},{"color":"red","value":90}]}, "unit":"percent","min":0,"max":100 },"overrides":[] }, "gridPos":{"h":4,"w":3,"x":18,"y":1}, "id":5, "options":{"colorMode":"background","graphMode":"area","justifyMode":"center","orientation":"auto","reduceOptions":{"calcs":["lastNotNull"],"fields":"","values":false},"textMode":"auto"}, "title":"Pool Utilisation %", "type":"stat", "targets":[{"datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"},"expr":"100 * sum(metallb_allocator_addresses_in_use_total) / sum(metallb_allocator_addresses_total)","legendFormat":"util%","refId":"A"}] }, { "datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"}, "description":"Free IPs = total - in_use. Near zero = pool exhaustion risk.", "fieldConfig":{ "defaults":{ "color":{"mode":"thresholds"}, "thresholds":{"mode":"absolute","steps":[{"color":"red","value":null},{"color":"yellow","value":3},{"color":"green","value":10}]}, "unit":"short" },"overrides":[] }, "gridPos":{"h":4,"w":3,"x":21,"y":1}, "id":6, "options":{"colorMode":"background","graphMode":"none","justifyMode":"center","orientation":"auto","reduceOptions":{"calcs":["lastNotNull"],"fields":"","values":false},"textMode":"auto"}, "title":"Free IPs Remaining", "type":"stat", "targets":[{"datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"},"expr":"sum(metallb_allocator_addresses_total) - sum(metallb_allocator_addresses_in_use_total)","legendFormat":"free","refId":"A"}] }, { "collapsed":false, "gridPos":{"h":1,"w":24,"x":0,"y":5}, "id":106, "title":"šŸŸ£ Node Leadership Balance ā€” L2 VIP Distribution", "type":"row" }, { "datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"}, "description":"Count of VIPs/services actively owned (announced) per node. In a healthy, balanced L2 cluster these bars should be roughly even. A single node with a disproportionately long bar is hogging L2 leadership ā€” a scaling and blast-radius risk, since that node's failure would trigger a mass failover.", "fieldConfig":{ "defaults":{ "color":{"mode":"thresholds"}, "thresholds":{"mode":"absolute","steps":[{"color":"green","value":null}]}, "unit":"short","min":0 },"overrides":[] }, "gridPos":{"h":8,"w":24,"x":0,"y":6}, "id":60, "options":{"displayMode":"gradient","minVizHeight":10,"minVizWidth":0,"namePlacement":"auto","orientation":"horizontal","reduceOptions":{"calcs":["lastNotNull"],"fields":"","values":false},"showUnfilled":true,"valueMode":"color"}, "title":"Active Services Per Node ā€” L2 Leadership Distribution", "type":"bargauge", "targets":[{"datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"},"expr":"sort_desc(sum by (node) (metallb_speaker_announced))","legendFormat":"{{node}}","refId":"A","instant":true}] }, { "collapsed":false, "gridPos":{"h":1,"w":24,"x":0,"y":14}, "id":107, "title":"šŸ“‹ Live Node-to-Service Mapping", "type":"row" }, { "datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"}, "description":"Live mapping of which Node and Speaker Pod currently announces which VIP, for which Service. Trimmed to operationally relevant labels only (container/endpoint/job/namespace/prometheus instance-scrape labels dropped). This answers: 'which VIP is this node serving and for which service'.", "fieldConfig":{ "defaults":{ "color":{"mode":"thresholds"}, "custom":{"align":"left","cellOptions":{"type":"auto"},"inspect":false}, "thresholds":{"mode":"absolute","steps":[{"color":"green","value":null}]}, "unit":"short" },"overrides":[] }, "gridPos":{"h":9,"w":24,"x":0,"y":15}, "id":61, "options":{"cellHeight":"sm","footer":{"countRows":false,"fields":"","reducer":["sum"],"show":false},"showHeader":true,"sortBy":[{"desc":false,"displayName":"Node"}]}, "title":"Node ā†’ Speaker Pod ā†’ VIP ā†’ Service (live)", "type":"table", "targets":[{"datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"},"expr":"metallb_speaker_announced{protocol=\"layer2\"}","refId":"A","instant":true,"format":"table"}], "transformations":[ { "id":"organize", "options":{ "excludeByName":{"Time":true,"__name__":true,"container":true,"endpoint":true,"job":true,"namespace":true,"prometheus":true,"instance":true,"Value":false}, "renameByName":{ "ip":"VIP", "node":"Node", "pod":"Speaker Pod", "protocol":"Protocol", "service":"Service", "Value":"Announced" }, "indexByName":{"Node":0,"Speaker Pod":1,"VIP":2,"Service":3,"Protocol":4,"Announced":5} } } ] }, { "collapsed":false, "gridPos":{"h":1,"w":24,"x":0,"y":24}, "id":102, "title":"šŸ“” L2 ARP/NDP ā€” Requests, Responses & Drop Detection (per VIP)", "type":"row" }, { "datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"}, "description":"Tracks per-second rate of incoming Layer 2 discovery broadcasts (ARP requests for IPv4 / NDP requests for IPv6) targeting MetalLB-managed LoadBalancer IPs.\n\nWhat it means:\nā€¢ Spikes: new client machines joining the network, expired router ARP caches, or potential network loops/storms.\nā€¢ Baseline: healthy, quiet networks show very low fractional decimal values (m req/s = milli-requests/sec), meaning external clients have already cached the node's MAC address and rarely re-query.\n\nHealthy target: sustained baseline under ~100 m req/s (0.1 req/s) per VIP. Sustained values above 1 req/s per VIP warrant investigation for ARP storms or flapping upstream switches.\n\nMetric: metallb_layer2_requests_received", "fieldConfig":{ "defaults":{ "color":{"mode":"palette-classic"}, "custom":{"axisLabel":"req/s","drawStyle":"line","fillOpacity":15,"gradientMode":"opacity","lineInterpolation":"smooth","lineWidth":2,"showPoints":"never","spanNulls":false,"stacking":{"group":"A","mode":"none"},"thresholdsStyle":{"mode":"off"}}, "unit":"reqps", "decimals":1 },"overrides":[] }, "gridPos":{"h":7,"w":8,"x":0,"y":25}, "id":20, "options":{"legend":{"calcs":["mean","max"],"displayMode":"table","placement":"bottom","showLegend":true},"tooltip":{"mode":"multi","sort":"desc"}}, "title":"ARP/NDP Requests Received (per VIP)", "type":"timeseries", "targets":[{"datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"},"expr":"rate(metallb_layer2_requests_received[$__rate_interval])","legendFormat":"{{ip}}","refId":"A"}] }, { "datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"}, "description":"Tracks per-second rate of outgoing Layer 2 answers (ARP replies for IPv4 / NDP neighbor advertisements for IPv6) sent by MetalLB.\n\nWhat it means:\nā€¢ Healthy state: this metric should exactly match Requests Received. A 1-to-1 ratio proves the cluster is acknowledging all network queries.\nā€¢ Drops/discrepancies: if this value drops below the request rate, the speaker pod is failing to answer network queries ā€” this WILL break external ingress traffic for that VIP.\n\nMetric: metallb_layer2_responses_sent", "fieldConfig":{ "defaults":{ "color":{"mode":"palette-classic"}, "custom":{"axisLabel":"resp/s","drawStyle":"line","fillOpacity":15,"gradientMode":"opacity","lineInterpolation":"smooth","lineWidth":2,"showPoints":"never","spanNulls":false,"stacking":{"group":"A","mode":"none"},"thresholdsStyle":{"mode":"off"}}, "unit":"reqps", "decimals":1 },"overrides":[] }, "gridPos":{"h":7,"w":8,"x":8,"y":25}, "id":21, "options":{"legend":{"calcs":["mean","max"],"displayMode":"table","placement":"bottom","showLegend":true},"tooltip":{"mode":"multi","sort":"desc"}}, "title":"ARP/NDP Responses Sent (per VIP)", "type":"timeseries", "targets":[{"datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"},"expr":"rate(metallb_layer2_responses_sent[$__rate_interval])","legendFormat":"{{ip}}","refId":"A"}] }, { "datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"}, "description":"Cluster-wide delta: total requests received minus total responses sent, per second. In a perfectly healthy network this line sits flat at zero.\n\nIf this value rises above 0 for a sustained period, MetalLB is ignoring network clients ā€” external ingress is being silently dropped for at least one VIP.\n\nSuggested alert: trigger if value > 0 for more than 2 minutes.\n\nFormula: sum(rate(metallb_layer2_requests_received[$__rate_interval])) - sum(rate(metallb_layer2_responses_sent[$__rate_interval]))", "fieldConfig":{ "defaults":{ "color":{"mode":"fixed","fixedColor":"red"}, "custom":{"axisCenteredZero":true,"axisLabel":"dropped req/s","drawStyle":"line","fillOpacity":25,"gradientMode":"opacity","lineInterpolation":"smooth","lineWidth":2,"showPoints":"never","spanNulls":false,"stacking":{"group":"A","mode":"none"},"thresholdsStyle":{"mode":"line+area","steps":[{"color":"transparent","value":null},{"color":"red","value":0.001}]}}, "unit":"reqps", "thresholds":{"mode":"absolute","steps":[{"color":"green","value":null},{"color":"red","value":0.001}]} },"overrides":[] }, "gridPos":{"h":7,"w":8,"x":16,"y":25}, "id":25, "options":{"legend":{"calcs":["mean","max"],"displayMode":"list","placement":"bottom","showLegend":true},"tooltip":{"mode":"single","sort":"none"}}, "title":"MetalLB L2 Dropped/Ignored Requests (cluster-wide) ā€” Must stay at 0", "type":"timeseries", "targets":[{"datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"},"expr":"sum(rate(metallb_layer2_requests_received[$__rate_interval])) - sum(rate(metallb_layer2_responses_sent[$__rate_interval]))","legendFormat":"dropped","refId":"A"}] }, { "collapsed":false, "gridPos":{"h":1,"w":24,"x":0,"y":32}, "id":108, "title":"šŸŸ” Gratuitous ARP/NDP ā€” Failover & Stability Signals", "type":"row" }, { "datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"}, "description":"Tracks per-second rate of unsolicited Layer 2 network updates (Gratuitous ARP for IPv4 / unsolicited NDP advertisement for IPv6) sent by MetalLB, per VIP.\n\nWhat it means:\nā€¢ Baseline (m pps): small fractional values = standard background tick keeping physical switch MAC tables fresh.\nā€¢ Spikes: heavy spikes = failover or cluster update, where a service migrates to a new leader node and forces the network to redirect traffic.\n\nReal-world scenarios:\nā€¢ Normal ops: steady ~12 mp/s ā€” cluster stable, just reminder packets.\nā€¢ Active failover: sharp spike to 500+ mp/s then drops back to baseline ā€” node crashed/maintained, new leader elected, switch flooded once to redirect traffic without dropping connections.\nā€¢ Flapping (critical): continuous erratic mountain-shaped spikes that never return to baseline ā€” split-brain, two+ nodes fighting over the same VIP. Severe internal cluster networking issue.\n\nMetric: metallb_layer2_gratuitous_sent", "fieldConfig":{ "defaults":{ "color":{"mode":"palette-classic"}, "custom":{"axisLabel":"pkt/s","drawStyle":"line","fillOpacity":20,"gradientMode":"opacity","lineInterpolation":"smooth","lineWidth":2,"showPoints":"never","spanNulls":false,"stacking":{"group":"A","mode":"none"},"thresholdsStyle":{"mode":"line+area","steps":[{"color":"transparent","value":null},{"color":"orange","value":1}]}}, "unit":"pps", "decimals":2, "thresholds":{"mode":"absolute","steps":[{"color":"green","value":null},{"color":"orange","value":0.5},{"color":"red","value":2}]} },"overrides":[] }, "gridPos":{"h":8,"w":12,"x":0,"y":33}, "id":22, "options":{"legend":{"calcs":["mean","max"],"displayMode":"table","placement":"bottom","showLegend":true},"tooltip":{"mode":"multi","sort":"desc"}}, "title":"Gratuitous ARP/NDP Rate (per VIP)", "type":"timeseries", "targets":[{"datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"},"expr":"rate(metallb_layer2_gratuitous_sent[$__rate_interval])","legendFormat":"{{ip}}","refId":"A"}] }, { "datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"}, "description":"Tracks per-second rate of gratuitous ARP/NDP sent by MetalLB, mapped directly to the hosting physical OpenShift Node (joins metallb_layer2_gratuitous_sent with kube_pod_info on pod label).\n\nWhat it emphasizes:\nā€¢ Node leadership: which physical server is actively managing traffic redirection / routing table freshness for your LoadBalancer IPs.\nā€¢ Failover dynamics: visualizes traffic migration between host machines during node reboots or cluster upgrades.\n\nTroubleshooting scenarios:\nā€¢ Stable: lines hover flat 0.01ā€“0.1 pps ā€” healthy, just keeping switch MAC tables fresh.\nā€¢ Standard pod update / micro-failover: one node briefly spikes to ~1 pps then drops back ā€” app pod relocated, MetalLB updated switch seamlessly.\nā€¢ Complete node outage: one node's line drops to 0 pps while another simultaneously spikes 10+ pps ā€” failed node, successful failover to survivor, which blasted the network to pull traffic.\nā€¢ Split-brain/flapping (critical): multiple node lines jumping up/down simultaneously in a chaotic zig-zag ā€” two nodes competing for the same VIP, severe packet drops for external users.\n\nRequires kube-state-metrics (kube_pod_info) scraped in the same Prometheus.\n\nMetric: metallb_layer2_gratuitous_sent joined with kube_pod_info{namespace=\"metallb-system\"}", "fieldConfig":{ "defaults":{ "color":{"mode":"palette-classic"}, "custom":{"axisLabel":"pkt/s","drawStyle":"line","fillOpacity":20,"gradientMode":"opacity","lineInterpolation":"smooth","lineWidth":2,"showPoints":"never","spanNulls":false,"stacking":{"group":"A","mode":"none"},"thresholdsStyle":{"mode":"off"}}, "unit":"pps", "decimals":2 },"overrides":[] }, "gridPos":{"h":8,"w":12,"x":12,"y":33}, "id":26, "options":{"legend":{"calcs":["mean","max"],"displayMode":"table","placement":"bottom","showLegend":true},"tooltip":{"mode":"multi","sort":"desc"}}, "title":"Gratuitous ARP/NDP Rate (per Node) ā€” Leadership & Failover Mapping", "type":"timeseries", "targets":[{"datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"},"expr":"sum by (node) (rate(metallb_layer2_gratuitous_sent[$__rate_interval]) * on(pod) group_left(node) kube_pod_info{namespace=\"metallb-system\"})","legendFormat":"{{node}}","refId":"A"}] }, { "datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"}, "description":"FIXED v2: gated to avoid divide-by-near-zero blowup. Old formula (100 * grat_rate / (req_rate + 0.001)) produced fake values like 133,000% whenever request_rate was near-zero ā€” pure math artifact, not real instability, because most L2 VIPs naturally get near-zero ARP traffic once clients cache the MAC.\n\nThis version only computes the ratio when request_rate is non-trivial (> 0.01 req/s, i.e. real client traffic present). When traffic is below that floor, the panel returns no data instead of a misleading huge percentage ā€” gaps in the line are expected and mean 'too quiet to judge', not an error.\n\nHow to read it now:\nā€¢ No line / gap: VIP is quiet, normal for L2 ā€” ratio is not meaningful here, ignore.\nā€¢ Line present, <50%: healthy ā€” gratuitous sends are a small fraction of real traffic.\nā€¢ Line present, >50% sustained: real instability ā€” leader is re-announcing far more than client demand justifies. Investigate pod crash-looping or split-brain on that VIP.\n\nFor raw instability regardless of traffic level, trust the 'Total Gratuitous Sends Ranked' panel below instead ā€” that one is not affected by this math issue.\n\nFormula: (req_rate > 0.01) ? 100 * grat_rate / req_rate : no data", "fieldConfig":{ "defaults":{ "color":{"mode":"palette-classic"}, "custom":{"axisLabel":"grat/req %","drawStyle":"line","fillOpacity":20,"gradientMode":"opacity","lineInterpolation":"smooth","lineWidth":2,"showPoints":"never","spanNulls":false,"stacking":{"group":"A","mode":"none"},"thresholdsStyle":{"mode":"line+area","steps":[{"color":"transparent","value":null},{"color":"red","value":50}]}}, "unit":"percent" },"overrides":[] }, "gridPos":{"h":7,"w":24,"x":0,"y":41}, "id":24, "options":{"legend":{"calcs":["mean","max"],"displayMode":"table","placement":"bottom","showLegend":true},"tooltip":{"mode":"multi","sort":"desc"}}, "title":"Gratuitous / Requests Ratio % (traffic-gated) ā€” gaps = too quiet to judge, line >50% = real instability", "type":"timeseries", "targets":[{"datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"},"expr":"(100 * rate(metallb_layer2_gratuitous_sent[$__rate_interval]) / rate(metallb_layer2_requests_received[$__rate_interval])) and (rate(metallb_layer2_requests_received[$__rate_interval]) > 0.01)","legendFormat":"{{ip}}","refId":"A"}] }, { "collapsed":false, "gridPos":{"h":1,"w":24,"x":0,"y":48}, "id":103, "title":"šŸ”“ Failover Severity ā€” Cumulative Gratuitous Sends Ranked", "type":"row" }, { "datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"}, "description":"Descending rank of raw, cumulative gratuitous packets sent by each MetalLB speaker/VIP. Immediately exposes which VIP or host node is generating the highest volume of unsolicited network table updates.\n\nHow to interpret values over 25,000 (25k):\nā€¢ Healthy (slow accumulation): if 25k took weeks of continuous uptime to reach, it's normal background maintenance behavior.\nā€¢ Critical (rapid vertical spike): if the counter reaches 25k within minutes or hours, a critical network failure is occurring.\n\nCommon causes of rapid 25k+ spikes:\n1. Micro-failover storms ā€” an app pod crash-looping, forcing MetalLB to migrate the VIP back and forth endlessly.\n2. Split-brain routing ā€” a physical network partition causing multiple nodes to believe they're the active leader simultaneously, fighting for control of the IP.\n\nFormula: sort_desc(metallb_layer2_gratuitous_sent)", "fieldConfig":{ "defaults":{ "color":{"mode":"thresholds"}, "thresholds":{"mode":"absolute","steps":[{"color":"green","value":null},{"color":"yellow","value":5000},{"color":"red","value":25000}]}, "unit":"short","min":0 },"overrides":[] }, "gridPos":{"h":9,"w":24,"x":0,"y":49}, "id":30, "options":{"displayMode":"gradient","minVizHeight":10,"minVizWidth":0,"namePlacement":"auto","orientation":"horizontal","reduceOptions":{"calcs":["lastNotNull"],"fields":"","values":false},"showUnfilled":true,"valueMode":"color"}, "title":"Total Gratuitous Sends Ranked (per VIP) ā€” bar length = failover severity", "type":"bargauge", "targets":[{"datasource":{"type":"prometheus","uid":"${DS_PROMETHEUS}"},"expr":"sort_desc(metallb_layer2_gratuitous_sent)","legendFormat":"{{ip}}","refId":"A","instant":true}] } ], "refresh":"30s", "schemaVersion":38, "tags":["metallb","l2","openshift","networking","arp"], "templating":{"list":[{"current":{},"hide":0,"includeAll":false,"label":"Datasource","multi":false,"name":"DS_PROMETHEUS","options":[],"query":"prometheus","refresh":1,"regex":"","type":"datasource"}]}, "time":{"from":"now-3h","to":"now"}, "timepicker":{}, "timezone":"browser", "title":"MetalLB ā€” L2 Mode", "uid":"metallb-l2-v4", "version":5, "weekStart":"" }