kodo-inner Configuration: Query Concurrency¶

Overview¶

This article describes an optimization method. When Guance has sufficient underlying resources and the Doris cluster shows no bottlenecks in CPU, memory, or disk I/O, but queries for logs and other data still respond slowly, you can increase the query concurrency of the kodo-inner service to improve query speed.

Modify Configuration¶

Method 1¶

On the launcher, go to the top-right menu → Modify Application Configuration → Namespace: forethought-kodo → kodoInner (Kodo Inner), and add the following content:

# Same level as global
dql:
    metric_query_workers: 64 # Number of workers for DQL metric data queries
    log_query_workers: 64 # Number of workers for DQL log/text data queries (logs, traces, RUM, and all other text data)
    general_query_workers: 64 # Number of workers for non-metric or non-log queries

After adding, check Auto-restart related services after configuration modification, and click Confirm configuration modification.

Method 2¶

Add parameters via the command line:

kubectl edit deployment -n forethought-kodo kodo-inner

Add:

# Same level as global
dql:
    metric_query_workers: 64 # Number of workers for DQL metric data queries
    log_query_workers: 64 # Number of workers for DQL log/text data queries (logs, traces, RUM, and all other text data)
    general_query_workers: 64 # Number of workers for non-metric or non-log queries

After adding, restart the service:

kubectl rollout restart deployment -n forethought-kodo kodo-inner

For other configurations, refer to: Application Configuration Guide