Bug #3183
closed
Go builder: failures connecting to locally created webservice during build tests
0%
Description
The Dragonfly builder for Go reports frequent, but random failures in the crypto/tls test. The root cause seems to be an inability to connect to the local webservice created during the test. This feels kind of race-y to me and I don't see any of the other builders experiencing the problem. I don't know enough to say whether it's a Go issue or a Dragonfly issue. An example of this is:
https://build.golang.org/log/955337d30c03d9fcda402623cc29df5e24db0f84
Updated by sepherosa over 5 years ago
Can you grab a tcpdump on the 127.0.0.1 during the test? I assume the
unrelated traffic on 127.0.0.1 would be low.
On Tue, Apr 2, 2019 at 1:58 AM <bugtracker-admin@leaf.dragonflybsd.org> wrote:
Issue #3183 has been reported by t_dfbsd.
----------------------------------------
Bug #3183: Go builder: failures connecting to locally created webservice during build tests
http://bugs.dragonflybsd.org/issues/3183
- Author: t_dfbsd
- Status: New
- Priority: Normal
- Assignee:
- Category:
- Target version:
----------------------------------------
The Dragonfly builder for Go reports frequent, but random failures in the crypto/tls test. The root cause seems to be an inability to connect to the local webservice created during the test. This feels kind of race-y to me and I don't see any of the other builders experiencing the problem. I don't know enough to say whether it's a Go issue or a Dragonfly issue. An example of this is:https://build.golang.org/log/955337d30c03d9fcda402623cc29df5e24db0f84
--
You have received this notification because you have either subscribed to it, or are involved in it.
To change your notification preferences, please click here: http://bugs.dragonflybsd.org/my/account
--
Tomorrow Will Never Die
Updated by Anonymous over 5 years ago
Thw Go team has posted an update on this bug: https://github.com/golang/go/issues/29583
Updated by Anonymous over 5 years ago
- Status changed from New to Closed
Closing this ticket. The Go team found a bug in the crypto/tls build test. From comments:
"crypto/tls: deflake localPipe in tests
The localPipe implementation assumes that every successful net.Dial
results in exactly one successful listener.Accept. I don't believe this
is guaranteed by essentially any operating system. For this test, we're
seeing flakes on dragonfly (#29583)."
I don't understand why Dragonfly was hitting this bug so often, whereas I don't recall seeing it on OpenBSD or NetBSD. In any case, it's fixed.