Closed hejjestemjarek closed 1 year ago
Created test_dp_stress_test.py
script that simulates X thousands CBSDs trying to concurrently send X thousands EnodebdUpdateCbsd
requests. At current state of DP testing with 5k CBSDs results in Radio Controller pod being terminated with OOMKilled
status due to reaching pod memory limit. Increasing pod's memory resolves the issue, but is a temporary fix and underlying issue still exists - with more CBSDs trying to concurrently update it would require more and more memory to handle these requests.
Tests are to be continued on a staging environment in a new task, passed the script and knowledge to @WojSad
Simulate multiple CBSDs sending requests to DP at the same time. Find element of the system that is responsible for possible failure under heavy load.