Loading...

Details

Type: Bug
Resolution: Done
Priority: Normal
Fix Version/s: netobserv-1.4
Affects Version/s: netobserv-1.3
Component/s: Console Plugin
Labels:
- pre-merge-tested

Blocked:
False
Blocked Reason:
None
Ready:
False
Release Note Text:

Hide
Previously, the Prometheus metrics exported by Network Observability were computed out of potentially duplicated network flows. In the related dashboards (from menu "Observe" > "Dashboards"; note that dashboards from the "Network Traffic" page were not affected), this could result in potentially doubled rates.

Now, network flows are filtered to eliminate duplicates prior to metrics calculation, which results in correct traffic rates displayed in the dashboards.

Show
Previously, the Prometheus metrics exported by Network Observability were computed out of potentially duplicated network flows. In the related dashboards (from menu "Observe" > "Dashboards"; note that dashboards from the "Network Traffic" page were not affected), this could result in potentially doubled rates. Now, network flows are filtered to eliminate duplicates prior to metrics calculation, which results in correct traffic rates displayed in the dashboards.

Sprint:
NetObserv - Sprint 238

SFDC Cases Links:
SFDC Cases Counter:

Description

Note I haven't tested, just figured out looking at the metric definitions, such as https://github.com/netobserv/network-observability-operator/blob/main/controllers/flowlogspipeline/metrics_definitions/node_egress_bytes_total.yaml

There's currently no filter for Duplicate=false, which I think means the produced values are ~doubling the actual byte rates. We should confirm this by comparing rates between our console plugin (which filters for duplicates) and these dashboards.

Note that the fix would involve improving the FLP API, since it seems only able to apply a single filter.

[Edit: discrepancy confirmed, cf attached screenshots]