Potential DNS errors failing to cancel HTTP requests.


Recently running a gatling load tests to make requests to our service with constantConcurrentUsers(1) during (10 minutes) we expect ~60 users per second. However, we saw our users drop down to zero and then shoot up to ~2400 users per second which resulted in our requests backing up.

Out of ~300k requests, ~250k failed, with the following errors pulled out of the gatling report.

  • i.n.c.AbstractChannel$AnnotatedSocketException: Cannot assign requested address: /:

  • status.find.is(200), but actually found 429

  • i.g.h.c.i.RequestTimeoutException: Request timeout after 60000 ms

  • j.n.UnknownHostException:

  • j.n.s.SSLException: handshake timed out

  • j.n.UnknownHostException: : Temporary failure in name resolution

We believe these may be caused by DNS errors failing to cancel the HTTP requests, is there a known issue within the http client?

Many thanks,

Jade Redworth

See https://github.com/gatling/gatling/issues/3622

Thanks for reporting!