jbielick / faktory_worker_node / 273

97%

push

travis-ci

test connection before starting worker pool

starting the worker pool involves each worker (n=concurrency) attempting
to fetch (staggered) very quickly. The connection pool faktory create()
method fails when the connection fails and each worker continuously
retries in a tight loop. This causes 100% CPU usage and excessive log
output. Testing the connection once before starting the worker pool
fetch loop can ensure that a connection can be established before
starting that thundering herd and, if it fails, can reject the
faktory.work() promise immediately and keep things from spinning out of
control.

Thus, one single error will occur and halt the process.

closes #38

89 of 91 branches covered (97.8%)

Branch coverage included in aggregate %.

358 of 363 relevant lines covered (98.62%)

978.18 hits per line