Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

net/http: TestClientCallsCloseOnlyOnce failures with "host has failed to respond" on Windows #49454

Open
bcmills opened this issue Nov 8, 2021 · 3 comments

Comments

@bcmills
Copy link
Member

@bcmills bcmills commented Nov 8, 2021

--- FAIL: TestClientCallsCloseOnlyOnce (21.04s)
    client_test.go:2063: dial tcp 127.0.0.1:51357: connectex: A connection attempt failed because the connected party did not properly respond after a period of time, or established connection failed because connected host has failed to respond.
FAIL
FAIL	net/http	25.262s

greplogs --dashboard -md -l -e '(?m)FAIL: TestClientCallsCloseOnlyOnce.*\n.*host has failed to respond'

2021-11-05T22:03:24-035963c/windows-arm64-10
2020-12-23T02:10:51-b0b0d98/windows-arm-zx2c4

CC @neild

@neild
Copy link
Contributor

@neild neild commented Nov 8, 2021

cc @bufflig

This test creates 50 HTTP client connections, in sequence, to the same server. The error seems to indicate that one of them failed. I don't know what might cause this, but the fact that the failure has happened once in a year and only on Windows makes this hard to diagnose; I suspect it's an issue on the builder, but I don't know what. Something applying a rate limit to connection attempts?

Given the rarity of the failure, this is probably not a priority.

Loading

@bufflig
Copy link

@bufflig bufflig commented Nov 8, 2021

I wonder if the fact that it happens on both arm 32 and 64 bit is significant? I agree that it seems really hard to diagnose when it happens this rarely. I can attempt to get it to happen with brute force running on my parallels VM, but I can not leave any guarantees of success.

Loading

@bcmills
Copy link
Member Author

@bcmills bcmills commented Nov 9, 2021

If the cause isn't easy to spot, I suggest that we give it a while to look for more occurrences. (I agree with Damien that this doesn't seem worth prioritizing at the moment.)

Loading

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Linked pull requests

Successfully merging a pull request may close this issue.

None yet
3 participants