net/http: TestIssue4191_InfiniteGetToPutTimeout failures with timeouts

[gopherbot]

#!watchflakes
post <- pkg == "net/http" && test == "TestIssue4191_InfiniteGetToPutTimeout" && `increasing timeout` && `Error issuing GET`

Issue created automatically to collect these failures.

Example (log):

--- FAIL: TestIssue4191_InfiniteGetToPutTimeout (0.00s)
    --- FAIL: TestIssue4191_InfiniteGetToPutTimeout/h1 (2.44s)
        transport_test.go:2289: increasing timeout
        transport_test.go:2294: Error issuing GET: Get "http://127.0.0.1:58443/get": read tcp 127.0.0.1:58441->127.0.0.1:58443: i/o timeout

— watchflakes

[gopherbot]

Found new dashboard test flakes for:

#!watchflakes
default <- pkg == "net/http" && test == "TestIssue4191_InfiniteGetToPutTimeout"

2023-07-21 05:26 netbsd-arm-bsiegert go@b104a0ef net/http.TestIssue4191_InfiniteGetToPutTimeout (log)

--- FAIL: TestIssue4191_InfiniteGetToPutTimeout (0.00s)
    --- FAIL: TestIssue4191_InfiniteGetToPutTimeout/h1 (2.44s)
        transport_test.go:2289: increasing timeout
        transport_test.go:2294: Error issuing GET: Get "http://127.0.0.1:58443/get": read tcp 127.0.0.1:58441->127.0.0.1:58443: i/o timeout

— watchflakes

[bcmills]

(attn @neild)

It looks like the test only tries increasing the timeout once, from a hard-coded 100ms to a hard-coded 1s.
It seems to me that if the test machine is slow enough (or loaded heavily enough), even that 1s timeout may be too short.

• See previously net/http: TestIssue4191_InfiniteGetTimeout failures #56274.