BugBot
Original Text
Context from Zulip thread "Community Help Chat > Traffic pattern after fiber merge":
- Trendal migrated traffic from a 1G DIA connection with its own LibreQoS box onto a 10G DC setup.
- After the migration, the 10G site did not show the expected traffic increase in the posted graphs.
- Packet Pusher asked whether traffic ever kicked up past 2 Gbps.
- Trendal then posted 7-day graphs, a 5-minute iperf graph, Insight live/1-hour/24-hour screenshots, and said the Insight numbers may be wrong.
- On the Insight graph showing the last 24 hours, Trendal says it appears to flat-top at around 2 Gbps regardless of whether more traffic is flowing.
- He says he can clearly drive the circuit to 3.5+ Gbps with iperf, but the Insight graph reports a high of 1.86 Gbps.
- He also says peak time on the 24-hour chart flat-lined for a while, which makes him nervous about a possible 2 Gbps bottleneck for customer traffic.
Expected behavior:
Insight throughput graphs should reflect observed throughput above 2 Gbps when the shaper/interface is actually passing more traffic, and graph highs should align with the underlying measurements.
Observed behavior:
Insight appears to visually or numerically cap/flatten throughput around ~2 Gbps. In Trendal's example, Insight reports a high of 1.86 Gbps while iperf can push the circuit above 3.5 Gbps.
Impact:
Operators may misread available capacity or miss real congestion/bottleneck signals. The report is especially concerning on 10G deployments where traffic above 2 Gbps is expected.
Initial triage notes:
This sounds more like an Insight collection/aggregation/chart scaling/unit issue than confirmed LibreQoS forwarding/shaping behavior, because iperf can exceed 2 Gbps. However, it would be useful to compare Insight's stored samples/API response with local interface counters and LibreQoS metrics during a >2 Gbps test.
BugBot
Original Text
Context from Zulip thread "Community Help Chat > Traffic pattern after fiber merge":
Expected behavior:
Insight throughput graphs should reflect observed throughput above 2 Gbps when the shaper/interface is actually passing more traffic, and graph highs should align with the underlying measurements.
Observed behavior:
Insight appears to visually or numerically cap/flatten throughput around ~2 Gbps. In Trendal's example, Insight reports a high of 1.86 Gbps while iperf can push the circuit above 3.5 Gbps.
Impact:
Operators may misread available capacity or miss real congestion/bottleneck signals. The report is especially concerning on 10G deployments where traffic above 2 Gbps is expected.
Initial triage notes:
This sounds more like an Insight collection/aggregation/chart scaling/unit issue than confirmed LibreQoS forwarding/shaping behavior, because iperf can exceed 2 Gbps. However, it would be useful to compare Insight's stored samples/API response with local interface counters and LibreQoS metrics during a >2 Gbps test.