Fix potential race condition in Docker provider test
What does this MR do?
As seen in https://gitlab.com/gitlab-org/gitlab-runner/-/jobs/4717853511, there is a potential data race between reading list
in the test. This can happen even though the test has a WaitGroup
that
ensures all its Goroutines have completed. However, in machineProvider
create()
, coordinator.waitForGrowthCapacity
is launched in a separate Goroutine, which may still be running after the test's Goroutines have completed.
To fix this data race, lock the mutex in the test before reading the value of list
.
Edited by Stan Hu