Contrail analytics api query response time issue when querying 1d samples

Bug #1757334 reported by Parth Sarupria on 2018-03-21
8
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Juniper Openstack
Status tracked in Trunk
R4.0
Invalid
Medium
Parth Sarupria
R4.1
Invalid
Medium
Parth Sarupria
Trunk
Invalid
Medium
Parth Sarupria

Bug Description

Contrail analytics api query response time increase when the duration is increased, even though the number of samples returned is the same.

In a setup with UVE pushed for 5 days, we have observed that when the query is made for last 7 days ("start_time": "now-7d"), we see time take around 5 secs (returning around 208 samples).
When we do for last 31 days ("start_time": "now-31d"), we see time take around 25 secs (returning same 208 samples).

Query used:
{"select_fields": ["app_metrics_diff_link_1d.value.session_duration", "app_metrics_diff_link_1d.value.session_count", "app_metrics_diff_link_1d.value.total_bytes", "app_metrics_diff_link_1d.value.input_bytes", "app_metrics_diff_link_1d.value.output_bytes", "app_metrics_diff_link_1d.value.sla_violation_duration", "app_metrics_diff_link_1d.value.sla_violation_count", "app_metrics_diff_link_1d.value.session_switch_count", "app_metrics_diff_link_1d.value.jitter_violation_count", "app_metrics_diff_link_1d.value.rtt_violation_count", "app_metrics_diff_link_1d.value.pkt_loss_violation_count", "app_metrics_diff_link_1d.__key", "T", "name"], "start_time": "now-7d", "where": [[{"name": "name", "value": "", "op": 7}]], "end_time": "now", "table": "StatTable.SDWANMetricsRecord.app_metrics_diff_link_1d.value"}

information type: Proprietary → Public
description: updated
Jeba Paulaiyan (jebap) wrote :

Please update contrail version, deployment details

no longer affects: juniperopenstack/r4.1
no longer affects: juniperopenstack/r4.0
no longer affects: juniperopenstack/r3.2
Jeba Paulaiyan (jebap) wrote :

As per Vineet, this was tried in R4.0

Sundaresan Rajangam (srajanga) wrote :

The query is divided into multiple chunks based on the start_time and end_time. So, even if there is no data for some time period, there is a cost involved in RTT for the query/response to/from cassandra. Can you please provide the analytics_statistics_ttl configured in contrail-collector.conf?

Parth Sarupria (psarupria) wrote :

here are the respective ttl values,

analytics_data_ttl = 750
analytics_config_audit_ttl = 2160
analytics_statistics_ttl = 750
analytics_flow_ttl = 2

the analytics_statistics_ttl is 750 hours

To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers