build/dashboard/builders_test.go

1129 строки
32 KiB
Go
Исходник Обычный вид История

// Copyright 2015 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package dashboard
import (
"bytes"
"fmt"
"os/exec"
"path/filepath"
"regexp"
"runtime"
"sort"
"strings"
"testing"
"time"
"github.com/google/go-cmp/cmp"
)
func TestOSARCHAccessors(t *testing.T) {
valid := func(s string) bool { return s != "" && !strings.Contains(s, "-") }
for _, conf := range Builders {
os := conf.GOOS()
arch := conf.GOARCH()
osArch := os + "-" + arch
if !valid(os) || !valid(arch) || !(conf.Name == osArch || strings.HasPrefix(conf.Name, osArch+"-")) {
t.Errorf("OS+ARCH(%q) = %q, %q; invalid", conf.Name, os, arch)
}
}
}
func TestDistTestsExecTimeout(t *testing.T) {
tests := []struct {
c *BuildConfig
want time.Duration
}{
{
&BuildConfig{
env: []string{},
testHostConf: &HostConfig{},
},
20 * time.Minute,
},
{
&BuildConfig{
env: []string{"GO_TEST_TIMEOUT_SCALE=2"},
testHostConf: &HostConfig{},
},
40 * time.Minute,
},
{
&BuildConfig{
env: []string{},
testHostConf: &HostConfig{
env: []string{"GO_TEST_TIMEOUT_SCALE=3"},
},
},
60 * time.Minute,
},
// BuildConfig's env takes precedence:
{
&BuildConfig{
env: []string{"GO_TEST_TIMEOUT_SCALE=2"},
testHostConf: &HostConfig{
env: []string{"GO_TEST_TIMEOUT_SCALE=3"},
},
},
40 * time.Minute,
},
}
for i, tt := range tests {
got := tt.c.DistTestsExecTimeout(nil)
if got != tt.want {
t.Errorf("%d. got %v; want %v", i, got, tt.want)
}
}
}
// TestTrybots tests that a given repo & its branch yields the provided
// complete set of builders. See also: TestBuilders, which tests both trybots
// and post-submit builders, both at arbitrary branches.
func TestTrybots(t *testing.T) {
tests := []struct {
repo string // "go", "net", etc
branch string // of repo
want []string
}{
{
repo: "go",
branch: "master",
want: []string{
"android-amd64-emu",
"freebsd-amd64-12_2",
"js-wasm",
"linux-386",
"linux-amd64",
"linux-amd64-race",
"linux-arm-aws",
"linux-arm64-aws",
dashboard: include at most 3 ports per misc-compile TryBot Between Go 1.15, 1.16 and tip, the following 29 ports don't have a real TryBot and instead rely on misc-compile TryBots for their pre-submit coverage: • aix/ppc64 • darwin/amd64 • darwin/arm64 • dragonfly/amd64 • freebsd/386 • freebsd/arm • freebsd/arm64 • illumos/amd64 • linux/mips • linux/mips64 • linux/mipsle • linux/mips64le • linux/ppc64 • linux/ppc64le • linux/riscv64 • linux/s390x • netbsd/386 • netbsd/amd64 • netbsd/arm • netbsd/arm64 • openbsd/386 • openbsd/arm • openbsd/arm64 • openbsd/mips64 • plan9/386 • plan9/amd64 • plan9/arm • solaris/amd64 • windows/arm The previous approach for misc-compile target selection was to break them up primarily by GOOS value. However, as new architectures were added over time, some misc-compile TryBots got to a point where they were testing upwards of 5 ports (for example, misc-compile-openbsd was testing 386, amd64, arm, arm64, and mips64 architectures). Since each port is tested sequentially, allocating too many to one misc-compile TryBot can cause it to become the bottleneck of an entire TryBot run, exceeding the 10 minute completion time goal. Arrange it so misc-compile TryBot target selection is done explicitly in x/build, and pick 3 as max number of targets per TryBot for now. Based on recent timing observations, that should strike a decent balance between resource use (spinning up a builder) vs chance of a misc-compile TryBot becoming a bottleneck. It will also give us an opportunity to compare timing of 1, 2 and 3 targets per misc-compile in the future. (When we start tracking timing for TryBot completion time holistically, we'll be in a better position to refine this strategy further.) Making misc-compile target selection explicit in x/build also enables removing unnecessary duplicate misc-compile coverage from ports that already have a real TryBot (for example, openbsd/amd64 was previously tested via both the openbsd-amd64-68 TryBot and misc-compile-openbsd). This shouldn't be needed, so it's no longer done. For golang/go#17104. Fixes golang/go#32632. Change-Id: Iac918377b91af3e48780b38ffdf3153e213eeba2 Reviewed-on: https://go-review.googlesource.com/c/build/+/313210 Trust: Dmitri Shuralyov <dmitshur@golang.org> Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Heschi Kreinick <heschi@google.com> Reviewed-by: Carlos Amedee <carlos@golang.org>
2021-04-23 23:56:37 +03:00
"openbsd-amd64-68",
"windows-386-2008",
"windows-amd64-2016",
"misc-compile-darwinarm64",
"misc-compile-freebsd",
dashboard: include at most 3 ports per misc-compile TryBot Between Go 1.15, 1.16 and tip, the following 29 ports don't have a real TryBot and instead rely on misc-compile TryBots for their pre-submit coverage: • aix/ppc64 • darwin/amd64 • darwin/arm64 • dragonfly/amd64 • freebsd/386 • freebsd/arm • freebsd/arm64 • illumos/amd64 • linux/mips • linux/mips64 • linux/mipsle • linux/mips64le • linux/ppc64 • linux/ppc64le • linux/riscv64 • linux/s390x • netbsd/386 • netbsd/amd64 • netbsd/arm • netbsd/arm64 • openbsd/386 • openbsd/arm • openbsd/arm64 • openbsd/mips64 • plan9/386 • plan9/amd64 • plan9/arm • solaris/amd64 • windows/arm The previous approach for misc-compile target selection was to break them up primarily by GOOS value. However, as new architectures were added over time, some misc-compile TryBots got to a point where they were testing upwards of 5 ports (for example, misc-compile-openbsd was testing 386, amd64, arm, arm64, and mips64 architectures). Since each port is tested sequentially, allocating too many to one misc-compile TryBot can cause it to become the bottleneck of an entire TryBot run, exceeding the 10 minute completion time goal. Arrange it so misc-compile TryBot target selection is done explicitly in x/build, and pick 3 as max number of targets per TryBot for now. Based on recent timing observations, that should strike a decent balance between resource use (spinning up a builder) vs chance of a misc-compile TryBot becoming a bottleneck. It will also give us an opportunity to compare timing of 1, 2 and 3 targets per misc-compile in the future. (When we start tracking timing for TryBot completion time holistically, we'll be in a better position to refine this strategy further.) Making misc-compile target selection explicit in x/build also enables removing unnecessary duplicate misc-compile coverage from ports that already have a real TryBot (for example, openbsd/amd64 was previously tested via both the openbsd-amd64-68 TryBot and misc-compile-openbsd). This shouldn't be needed, so it's no longer done. For golang/go#17104. Fixes golang/go#32632. Change-Id: Iac918377b91af3e48780b38ffdf3153e213eeba2 Reviewed-on: https://go-review.googlesource.com/c/build/+/313210 Trust: Dmitri Shuralyov <dmitshur@golang.org> Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Heschi Kreinick <heschi@google.com> Reviewed-by: Carlos Amedee <carlos@golang.org>
2021-04-23 23:56:37 +03:00
"misc-compile-mac-win",
"misc-compile-mips",
dashboard: include at most 3 ports per misc-compile TryBot Between Go 1.15, 1.16 and tip, the following 29 ports don't have a real TryBot and instead rely on misc-compile TryBots for their pre-submit coverage: • aix/ppc64 • darwin/amd64 • darwin/arm64 • dragonfly/amd64 • freebsd/386 • freebsd/arm • freebsd/arm64 • illumos/amd64 • linux/mips • linux/mips64 • linux/mipsle • linux/mips64le • linux/ppc64 • linux/ppc64le • linux/riscv64 • linux/s390x • netbsd/386 • netbsd/amd64 • netbsd/arm • netbsd/arm64 • openbsd/386 • openbsd/arm • openbsd/arm64 • openbsd/mips64 • plan9/386 • plan9/amd64 • plan9/arm • solaris/amd64 • windows/arm The previous approach for misc-compile target selection was to break them up primarily by GOOS value. However, as new architectures were added over time, some misc-compile TryBots got to a point where they were testing upwards of 5 ports (for example, misc-compile-openbsd was testing 386, amd64, arm, arm64, and mips64 architectures). Since each port is tested sequentially, allocating too many to one misc-compile TryBot can cause it to become the bottleneck of an entire TryBot run, exceeding the 10 minute completion time goal. Arrange it so misc-compile TryBot target selection is done explicitly in x/build, and pick 3 as max number of targets per TryBot for now. Based on recent timing observations, that should strike a decent balance between resource use (spinning up a builder) vs chance of a misc-compile TryBot becoming a bottleneck. It will also give us an opportunity to compare timing of 1, 2 and 3 targets per misc-compile in the future. (When we start tracking timing for TryBot completion time holistically, we'll be in a better position to refine this strategy further.) Making misc-compile target selection explicit in x/build also enables removing unnecessary duplicate misc-compile coverage from ports that already have a real TryBot (for example, openbsd/amd64 was previously tested via both the openbsd-amd64-68 TryBot and misc-compile-openbsd). This shouldn't be needed, so it's no longer done. For golang/go#17104. Fixes golang/go#32632. Change-Id: Iac918377b91af3e48780b38ffdf3153e213eeba2 Reviewed-on: https://go-review.googlesource.com/c/build/+/313210 Trust: Dmitri Shuralyov <dmitshur@golang.org> Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Heschi Kreinick <heschi@google.com> Reviewed-by: Carlos Amedee <carlos@golang.org>
2021-04-23 23:56:37 +03:00
"misc-compile-mipsle",
"misc-compile-netbsd",
dashboard: include at most 3 ports per misc-compile TryBot Between Go 1.15, 1.16 and tip, the following 29 ports don't have a real TryBot and instead rely on misc-compile TryBots for their pre-submit coverage: • aix/ppc64 • darwin/amd64 • darwin/arm64 • dragonfly/amd64 • freebsd/386 • freebsd/arm • freebsd/arm64 • illumos/amd64 • linux/mips • linux/mips64 • linux/mipsle • linux/mips64le • linux/ppc64 • linux/ppc64le • linux/riscv64 • linux/s390x • netbsd/386 • netbsd/amd64 • netbsd/arm • netbsd/arm64 • openbsd/386 • openbsd/arm • openbsd/arm64 • openbsd/mips64 • plan9/386 • plan9/amd64 • plan9/arm • solaris/amd64 • windows/arm The previous approach for misc-compile target selection was to break them up primarily by GOOS value. However, as new architectures were added over time, some misc-compile TryBots got to a point where they were testing upwards of 5 ports (for example, misc-compile-openbsd was testing 386, amd64, arm, arm64, and mips64 architectures). Since each port is tested sequentially, allocating too many to one misc-compile TryBot can cause it to become the bottleneck of an entire TryBot run, exceeding the 10 minute completion time goal. Arrange it so misc-compile TryBot target selection is done explicitly in x/build, and pick 3 as max number of targets per TryBot for now. Based on recent timing observations, that should strike a decent balance between resource use (spinning up a builder) vs chance of a misc-compile TryBot becoming a bottleneck. It will also give us an opportunity to compare timing of 1, 2 and 3 targets per misc-compile in the future. (When we start tracking timing for TryBot completion time holistically, we'll be in a better position to refine this strategy further.) Making misc-compile target selection explicit in x/build also enables removing unnecessary duplicate misc-compile coverage from ports that already have a real TryBot (for example, openbsd/amd64 was previously tested via both the openbsd-amd64-68 TryBot and misc-compile-openbsd). This shouldn't be needed, so it's no longer done. For golang/go#17104. Fixes golang/go#32632. Change-Id: Iac918377b91af3e48780b38ffdf3153e213eeba2 Reviewed-on: https://go-review.googlesource.com/c/build/+/313210 Trust: Dmitri Shuralyov <dmitshur@golang.org> Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Heschi Kreinick <heschi@google.com> Reviewed-by: Carlos Amedee <carlos@golang.org>
2021-04-23 23:56:37 +03:00
"misc-compile-netbsd-arm",
"misc-compile-openbsd",
dashboard: include at most 3 ports per misc-compile TryBot Between Go 1.15, 1.16 and tip, the following 29 ports don't have a real TryBot and instead rely on misc-compile TryBots for their pre-submit coverage: • aix/ppc64 • darwin/amd64 • darwin/arm64 • dragonfly/amd64 • freebsd/386 • freebsd/arm • freebsd/arm64 • illumos/amd64 • linux/mips • linux/mips64 • linux/mipsle • linux/mips64le • linux/ppc64 • linux/ppc64le • linux/riscv64 • linux/s390x • netbsd/386 • netbsd/amd64 • netbsd/arm • netbsd/arm64 • openbsd/386 • openbsd/arm • openbsd/arm64 • openbsd/mips64 • plan9/386 • plan9/amd64 • plan9/arm • solaris/amd64 • windows/arm The previous approach for misc-compile target selection was to break them up primarily by GOOS value. However, as new architectures were added over time, some misc-compile TryBots got to a point where they were testing upwards of 5 ports (for example, misc-compile-openbsd was testing 386, amd64, arm, arm64, and mips64 architectures). Since each port is tested sequentially, allocating too many to one misc-compile TryBot can cause it to become the bottleneck of an entire TryBot run, exceeding the 10 minute completion time goal. Arrange it so misc-compile TryBot target selection is done explicitly in x/build, and pick 3 as max number of targets per TryBot for now. Based on recent timing observations, that should strike a decent balance between resource use (spinning up a builder) vs chance of a misc-compile TryBot becoming a bottleneck. It will also give us an opportunity to compare timing of 1, 2 and 3 targets per misc-compile in the future. (When we start tracking timing for TryBot completion time holistically, we'll be in a better position to refine this strategy further.) Making misc-compile target selection explicit in x/build also enables removing unnecessary duplicate misc-compile coverage from ports that already have a real TryBot (for example, openbsd/amd64 was previously tested via both the openbsd-amd64-68 TryBot and misc-compile-openbsd). This shouldn't be needed, so it's no longer done. For golang/go#17104. Fixes golang/go#32632. Change-Id: Iac918377b91af3e48780b38ffdf3153e213eeba2 Reviewed-on: https://go-review.googlesource.com/c/build/+/313210 Trust: Dmitri Shuralyov <dmitshur@golang.org> Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Heschi Kreinick <heschi@google.com> Reviewed-by: Carlos Amedee <carlos@golang.org>
2021-04-23 23:56:37 +03:00
"misc-compile-openbsd-arm",
"misc-compile-plan9",
"misc-compile-ppc",
dashboard: include at most 3 ports per misc-compile TryBot Between Go 1.15, 1.16 and tip, the following 29 ports don't have a real TryBot and instead rely on misc-compile TryBots for their pre-submit coverage: • aix/ppc64 • darwin/amd64 • darwin/arm64 • dragonfly/amd64 • freebsd/386 • freebsd/arm • freebsd/arm64 • illumos/amd64 • linux/mips • linux/mips64 • linux/mipsle • linux/mips64le • linux/ppc64 • linux/ppc64le • linux/riscv64 • linux/s390x • netbsd/386 • netbsd/amd64 • netbsd/arm • netbsd/arm64 • openbsd/386 • openbsd/arm • openbsd/arm64 • openbsd/mips64 • plan9/386 • plan9/amd64 • plan9/arm • solaris/amd64 • windows/arm The previous approach for misc-compile target selection was to break them up primarily by GOOS value. However, as new architectures were added over time, some misc-compile TryBots got to a point where they were testing upwards of 5 ports (for example, misc-compile-openbsd was testing 386, amd64, arm, arm64, and mips64 architectures). Since each port is tested sequentially, allocating too many to one misc-compile TryBot can cause it to become the bottleneck of an entire TryBot run, exceeding the 10 minute completion time goal. Arrange it so misc-compile TryBot target selection is done explicitly in x/build, and pick 3 as max number of targets per TryBot for now. Based on recent timing observations, that should strike a decent balance between resource use (spinning up a builder) vs chance of a misc-compile TryBot becoming a bottleneck. It will also give us an opportunity to compare timing of 1, 2 and 3 targets per misc-compile in the future. (When we start tracking timing for TryBot completion time holistically, we'll be in a better position to refine this strategy further.) Making misc-compile target selection explicit in x/build also enables removing unnecessary duplicate misc-compile coverage from ports that already have a real TryBot (for example, openbsd/amd64 was previously tested via both the openbsd-amd64-68 TryBot and misc-compile-openbsd). This shouldn't be needed, so it's no longer done. For golang/go#17104. Fixes golang/go#32632. Change-Id: Iac918377b91af3e48780b38ffdf3153e213eeba2 Reviewed-on: https://go-review.googlesource.com/c/build/+/313210 Trust: Dmitri Shuralyov <dmitshur@golang.org> Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Heschi Kreinick <heschi@google.com> Reviewed-by: Carlos Amedee <carlos@golang.org>
2021-04-23 23:56:37 +03:00
"misc-compile-other-1",
"misc-compile-other-2",
},
},
{
repo: "go",
branch: "dev.link",
want: []string{
"android-amd64-emu",
"freebsd-amd64-12_2",
"js-wasm",
"linux-386",
"linux-amd64",
"linux-amd64-race",
"linux-arm-aws",
"linux-arm64-aws",
dashboard: include at most 3 ports per misc-compile TryBot Between Go 1.15, 1.16 and tip, the following 29 ports don't have a real TryBot and instead rely on misc-compile TryBots for their pre-submit coverage: • aix/ppc64 • darwin/amd64 • darwin/arm64 • dragonfly/amd64 • freebsd/386 • freebsd/arm • freebsd/arm64 • illumos/amd64 • linux/mips • linux/mips64 • linux/mipsle • linux/mips64le • linux/ppc64 • linux/ppc64le • linux/riscv64 • linux/s390x • netbsd/386 • netbsd/amd64 • netbsd/arm • netbsd/arm64 • openbsd/386 • openbsd/arm • openbsd/arm64 • openbsd/mips64 • plan9/386 • plan9/amd64 • plan9/arm • solaris/amd64 • windows/arm The previous approach for misc-compile target selection was to break them up primarily by GOOS value. However, as new architectures were added over time, some misc-compile TryBots got to a point where they were testing upwards of 5 ports (for example, misc-compile-openbsd was testing 386, amd64, arm, arm64, and mips64 architectures). Since each port is tested sequentially, allocating too many to one misc-compile TryBot can cause it to become the bottleneck of an entire TryBot run, exceeding the 10 minute completion time goal. Arrange it so misc-compile TryBot target selection is done explicitly in x/build, and pick 3 as max number of targets per TryBot for now. Based on recent timing observations, that should strike a decent balance between resource use (spinning up a builder) vs chance of a misc-compile TryBot becoming a bottleneck. It will also give us an opportunity to compare timing of 1, 2 and 3 targets per misc-compile in the future. (When we start tracking timing for TryBot completion time holistically, we'll be in a better position to refine this strategy further.) Making misc-compile target selection explicit in x/build also enables removing unnecessary duplicate misc-compile coverage from ports that already have a real TryBot (for example, openbsd/amd64 was previously tested via both the openbsd-amd64-68 TryBot and misc-compile-openbsd). This shouldn't be needed, so it's no longer done. For golang/go#17104. Fixes golang/go#32632. Change-Id: Iac918377b91af3e48780b38ffdf3153e213eeba2 Reviewed-on: https://go-review.googlesource.com/c/build/+/313210 Trust: Dmitri Shuralyov <dmitshur@golang.org> Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Heschi Kreinick <heschi@google.com> Reviewed-by: Carlos Amedee <carlos@golang.org>
2021-04-23 23:56:37 +03:00
"openbsd-amd64-68",
"windows-386-2008",
"windows-amd64-2016",
"misc-compile-darwinarm64",
"misc-compile-freebsd",
dashboard: include at most 3 ports per misc-compile TryBot Between Go 1.15, 1.16 and tip, the following 29 ports don't have a real TryBot and instead rely on misc-compile TryBots for their pre-submit coverage: • aix/ppc64 • darwin/amd64 • darwin/arm64 • dragonfly/amd64 • freebsd/386 • freebsd/arm • freebsd/arm64 • illumos/amd64 • linux/mips • linux/mips64 • linux/mipsle • linux/mips64le • linux/ppc64 • linux/ppc64le • linux/riscv64 • linux/s390x • netbsd/386 • netbsd/amd64 • netbsd/arm • netbsd/arm64 • openbsd/386 • openbsd/arm • openbsd/arm64 • openbsd/mips64 • plan9/386 • plan9/amd64 • plan9/arm • solaris/amd64 • windows/arm The previous approach for misc-compile target selection was to break them up primarily by GOOS value. However, as new architectures were added over time, some misc-compile TryBots got to a point where they were testing upwards of 5 ports (for example, misc-compile-openbsd was testing 386, amd64, arm, arm64, and mips64 architectures). Since each port is tested sequentially, allocating too many to one misc-compile TryBot can cause it to become the bottleneck of an entire TryBot run, exceeding the 10 minute completion time goal. Arrange it so misc-compile TryBot target selection is done explicitly in x/build, and pick 3 as max number of targets per TryBot for now. Based on recent timing observations, that should strike a decent balance between resource use (spinning up a builder) vs chance of a misc-compile TryBot becoming a bottleneck. It will also give us an opportunity to compare timing of 1, 2 and 3 targets per misc-compile in the future. (When we start tracking timing for TryBot completion time holistically, we'll be in a better position to refine this strategy further.) Making misc-compile target selection explicit in x/build also enables removing unnecessary duplicate misc-compile coverage from ports that already have a real TryBot (for example, openbsd/amd64 was previously tested via both the openbsd-amd64-68 TryBot and misc-compile-openbsd). This shouldn't be needed, so it's no longer done. For golang/go#17104. Fixes golang/go#32632. Change-Id: Iac918377b91af3e48780b38ffdf3153e213eeba2 Reviewed-on: https://go-review.googlesource.com/c/build/+/313210 Trust: Dmitri Shuralyov <dmitshur@golang.org> Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Heschi Kreinick <heschi@google.com> Reviewed-by: Carlos Amedee <carlos@golang.org>
2021-04-23 23:56:37 +03:00
"misc-compile-mac-win",
"misc-compile-mips",
dashboard: include at most 3 ports per misc-compile TryBot Between Go 1.15, 1.16 and tip, the following 29 ports don't have a real TryBot and instead rely on misc-compile TryBots for their pre-submit coverage: • aix/ppc64 • darwin/amd64 • darwin/arm64 • dragonfly/amd64 • freebsd/386 • freebsd/arm • freebsd/arm64 • illumos/amd64 • linux/mips • linux/mips64 • linux/mipsle • linux/mips64le • linux/ppc64 • linux/ppc64le • linux/riscv64 • linux/s390x • netbsd/386 • netbsd/amd64 • netbsd/arm • netbsd/arm64 • openbsd/386 • openbsd/arm • openbsd/arm64 • openbsd/mips64 • plan9/386 • plan9/amd64 • plan9/arm • solaris/amd64 • windows/arm The previous approach for misc-compile target selection was to break them up primarily by GOOS value. However, as new architectures were added over time, some misc-compile TryBots got to a point where they were testing upwards of 5 ports (for example, misc-compile-openbsd was testing 386, amd64, arm, arm64, and mips64 architectures). Since each port is tested sequentially, allocating too many to one misc-compile TryBot can cause it to become the bottleneck of an entire TryBot run, exceeding the 10 minute completion time goal. Arrange it so misc-compile TryBot target selection is done explicitly in x/build, and pick 3 as max number of targets per TryBot for now. Based on recent timing observations, that should strike a decent balance between resource use (spinning up a builder) vs chance of a misc-compile TryBot becoming a bottleneck. It will also give us an opportunity to compare timing of 1, 2 and 3 targets per misc-compile in the future. (When we start tracking timing for TryBot completion time holistically, we'll be in a better position to refine this strategy further.) Making misc-compile target selection explicit in x/build also enables removing unnecessary duplicate misc-compile coverage from ports that already have a real TryBot (for example, openbsd/amd64 was previously tested via both the openbsd-amd64-68 TryBot and misc-compile-openbsd). This shouldn't be needed, so it's no longer done. For golang/go#17104. Fixes golang/go#32632. Change-Id: Iac918377b91af3e48780b38ffdf3153e213eeba2 Reviewed-on: https://go-review.googlesource.com/c/build/+/313210 Trust: Dmitri Shuralyov <dmitshur@golang.org> Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Heschi Kreinick <heschi@google.com> Reviewed-by: Carlos Amedee <carlos@golang.org>
2021-04-23 23:56:37 +03:00
"misc-compile-mipsle",
"misc-compile-netbsd",
dashboard: include at most 3 ports per misc-compile TryBot Between Go 1.15, 1.16 and tip, the following 29 ports don't have a real TryBot and instead rely on misc-compile TryBots for their pre-submit coverage: • aix/ppc64 • darwin/amd64 • darwin/arm64 • dragonfly/amd64 • freebsd/386 • freebsd/arm • freebsd/arm64 • illumos/amd64 • linux/mips • linux/mips64 • linux/mipsle • linux/mips64le • linux/ppc64 • linux/ppc64le • linux/riscv64 • linux/s390x • netbsd/386 • netbsd/amd64 • netbsd/arm • netbsd/arm64 • openbsd/386 • openbsd/arm • openbsd/arm64 • openbsd/mips64 • plan9/386 • plan9/amd64 • plan9/arm • solaris/amd64 • windows/arm The previous approach for misc-compile target selection was to break them up primarily by GOOS value. However, as new architectures were added over time, some misc-compile TryBots got to a point where they were testing upwards of 5 ports (for example, misc-compile-openbsd was testing 386, amd64, arm, arm64, and mips64 architectures). Since each port is tested sequentially, allocating too many to one misc-compile TryBot can cause it to become the bottleneck of an entire TryBot run, exceeding the 10 minute completion time goal. Arrange it so misc-compile TryBot target selection is done explicitly in x/build, and pick 3 as max number of targets per TryBot for now. Based on recent timing observations, that should strike a decent balance between resource use (spinning up a builder) vs chance of a misc-compile TryBot becoming a bottleneck. It will also give us an opportunity to compare timing of 1, 2 and 3 targets per misc-compile in the future. (When we start tracking timing for TryBot completion time holistically, we'll be in a better position to refine this strategy further.) Making misc-compile target selection explicit in x/build also enables removing unnecessary duplicate misc-compile coverage from ports that already have a real TryBot (for example, openbsd/amd64 was previously tested via both the openbsd-amd64-68 TryBot and misc-compile-openbsd). This shouldn't be needed, so it's no longer done. For golang/go#17104. Fixes golang/go#32632. Change-Id: Iac918377b91af3e48780b38ffdf3153e213eeba2 Reviewed-on: https://go-review.googlesource.com/c/build/+/313210 Trust: Dmitri Shuralyov <dmitshur@golang.org> Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Heschi Kreinick <heschi@google.com> Reviewed-by: Carlos Amedee <carlos@golang.org>
2021-04-23 23:56:37 +03:00
"misc-compile-netbsd-arm",
"misc-compile-openbsd",
dashboard: include at most 3 ports per misc-compile TryBot Between Go 1.15, 1.16 and tip, the following 29 ports don't have a real TryBot and instead rely on misc-compile TryBots for their pre-submit coverage: • aix/ppc64 • darwin/amd64 • darwin/arm64 • dragonfly/amd64 • freebsd/386 • freebsd/arm • freebsd/arm64 • illumos/amd64 • linux/mips • linux/mips64 • linux/mipsle • linux/mips64le • linux/ppc64 • linux/ppc64le • linux/riscv64 • linux/s390x • netbsd/386 • netbsd/amd64 • netbsd/arm • netbsd/arm64 • openbsd/386 • openbsd/arm • openbsd/arm64 • openbsd/mips64 • plan9/386 • plan9/amd64 • plan9/arm • solaris/amd64 • windows/arm The previous approach for misc-compile target selection was to break them up primarily by GOOS value. However, as new architectures were added over time, some misc-compile TryBots got to a point where they were testing upwards of 5 ports (for example, misc-compile-openbsd was testing 386, amd64, arm, arm64, and mips64 architectures). Since each port is tested sequentially, allocating too many to one misc-compile TryBot can cause it to become the bottleneck of an entire TryBot run, exceeding the 10 minute completion time goal. Arrange it so misc-compile TryBot target selection is done explicitly in x/build, and pick 3 as max number of targets per TryBot for now. Based on recent timing observations, that should strike a decent balance between resource use (spinning up a builder) vs chance of a misc-compile TryBot becoming a bottleneck. It will also give us an opportunity to compare timing of 1, 2 and 3 targets per misc-compile in the future. (When we start tracking timing for TryBot completion time holistically, we'll be in a better position to refine this strategy further.) Making misc-compile target selection explicit in x/build also enables removing unnecessary duplicate misc-compile coverage from ports that already have a real TryBot (for example, openbsd/amd64 was previously tested via both the openbsd-amd64-68 TryBot and misc-compile-openbsd). This shouldn't be needed, so it's no longer done. For golang/go#17104. Fixes golang/go#32632. Change-Id: Iac918377b91af3e48780b38ffdf3153e213eeba2 Reviewed-on: https://go-review.googlesource.com/c/build/+/313210 Trust: Dmitri Shuralyov <dmitshur@golang.org> Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Heschi Kreinick <heschi@google.com> Reviewed-by: Carlos Amedee <carlos@golang.org>
2021-04-23 23:56:37 +03:00
"misc-compile-openbsd-arm",
"misc-compile-plan9",
"misc-compile-ppc",
dashboard: include at most 3 ports per misc-compile TryBot Between Go 1.15, 1.16 and tip, the following 29 ports don't have a real TryBot and instead rely on misc-compile TryBots for their pre-submit coverage: • aix/ppc64 • darwin/amd64 • darwin/arm64 • dragonfly/amd64 • freebsd/386 • freebsd/arm • freebsd/arm64 • illumos/amd64 • linux/mips • linux/mips64 • linux/mipsle • linux/mips64le • linux/ppc64 • linux/ppc64le • linux/riscv64 • linux/s390x • netbsd/386 • netbsd/amd64 • netbsd/arm • netbsd/arm64 • openbsd/386 • openbsd/arm • openbsd/arm64 • openbsd/mips64 • plan9/386 • plan9/amd64 • plan9/arm • solaris/amd64 • windows/arm The previous approach for misc-compile target selection was to break them up primarily by GOOS value. However, as new architectures were added over time, some misc-compile TryBots got to a point where they were testing upwards of 5 ports (for example, misc-compile-openbsd was testing 386, amd64, arm, arm64, and mips64 architectures). Since each port is tested sequentially, allocating too many to one misc-compile TryBot can cause it to become the bottleneck of an entire TryBot run, exceeding the 10 minute completion time goal. Arrange it so misc-compile TryBot target selection is done explicitly in x/build, and pick 3 as max number of targets per TryBot for now. Based on recent timing observations, that should strike a decent balance between resource use (spinning up a builder) vs chance of a misc-compile TryBot becoming a bottleneck. It will also give us an opportunity to compare timing of 1, 2 and 3 targets per misc-compile in the future. (When we start tracking timing for TryBot completion time holistically, we'll be in a better position to refine this strategy further.) Making misc-compile target selection explicit in x/build also enables removing unnecessary duplicate misc-compile coverage from ports that already have a real TryBot (for example, openbsd/amd64 was previously tested via both the openbsd-amd64-68 TryBot and misc-compile-openbsd). This shouldn't be needed, so it's no longer done. For golang/go#17104. Fixes golang/go#32632. Change-Id: Iac918377b91af3e48780b38ffdf3153e213eeba2 Reviewed-on: https://go-review.googlesource.com/c/build/+/313210 Trust: Dmitri Shuralyov <dmitshur@golang.org> Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Heschi Kreinick <heschi@google.com> Reviewed-by: Carlos Amedee <carlos@golang.org>
2021-04-23 23:56:37 +03:00
"misc-compile-other-1",
"misc-compile-other-2",
},
},
{
repo: "go",
branch: "release-branch.go1.16",
want: []string{
"android-amd64-emu",
"freebsd-amd64-12_2",
"js-wasm",
"linux-386",
"linux-amd64",
"linux-amd64-race",
"linux-arm-aws",
"linux-arm64-aws",
dashboard: include at most 3 ports per misc-compile TryBot Between Go 1.15, 1.16 and tip, the following 29 ports don't have a real TryBot and instead rely on misc-compile TryBots for their pre-submit coverage: • aix/ppc64 • darwin/amd64 • darwin/arm64 • dragonfly/amd64 • freebsd/386 • freebsd/arm • freebsd/arm64 • illumos/amd64 • linux/mips • linux/mips64 • linux/mipsle • linux/mips64le • linux/ppc64 • linux/ppc64le • linux/riscv64 • linux/s390x • netbsd/386 • netbsd/amd64 • netbsd/arm • netbsd/arm64 • openbsd/386 • openbsd/arm • openbsd/arm64 • openbsd/mips64 • plan9/386 • plan9/amd64 • plan9/arm • solaris/amd64 • windows/arm The previous approach for misc-compile target selection was to break them up primarily by GOOS value. However, as new architectures were added over time, some misc-compile TryBots got to a point where they were testing upwards of 5 ports (for example, misc-compile-openbsd was testing 386, amd64, arm, arm64, and mips64 architectures). Since each port is tested sequentially, allocating too many to one misc-compile TryBot can cause it to become the bottleneck of an entire TryBot run, exceeding the 10 minute completion time goal. Arrange it so misc-compile TryBot target selection is done explicitly in x/build, and pick 3 as max number of targets per TryBot for now. Based on recent timing observations, that should strike a decent balance between resource use (spinning up a builder) vs chance of a misc-compile TryBot becoming a bottleneck. It will also give us an opportunity to compare timing of 1, 2 and 3 targets per misc-compile in the future. (When we start tracking timing for TryBot completion time holistically, we'll be in a better position to refine this strategy further.) Making misc-compile target selection explicit in x/build also enables removing unnecessary duplicate misc-compile coverage from ports that already have a real TryBot (for example, openbsd/amd64 was previously tested via both the openbsd-amd64-68 TryBot and misc-compile-openbsd). This shouldn't be needed, so it's no longer done. For golang/go#17104. Fixes golang/go#32632. Change-Id: Iac918377b91af3e48780b38ffdf3153e213eeba2 Reviewed-on: https://go-review.googlesource.com/c/build/+/313210 Trust: Dmitri Shuralyov <dmitshur@golang.org> Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Heschi Kreinick <heschi@google.com> Reviewed-by: Carlos Amedee <carlos@golang.org>
2021-04-23 23:56:37 +03:00
"openbsd-amd64-68",
"windows-386-2008",
"windows-amd64-2016",
"misc-compile-darwinarm64", // Starts with Go 1.16.
"misc-compile-freebsd",
dashboard: include at most 3 ports per misc-compile TryBot Between Go 1.15, 1.16 and tip, the following 29 ports don't have a real TryBot and instead rely on misc-compile TryBots for their pre-submit coverage: • aix/ppc64 • darwin/amd64 • darwin/arm64 • dragonfly/amd64 • freebsd/386 • freebsd/arm • freebsd/arm64 • illumos/amd64 • linux/mips • linux/mips64 • linux/mipsle • linux/mips64le • linux/ppc64 • linux/ppc64le • linux/riscv64 • linux/s390x • netbsd/386 • netbsd/amd64 • netbsd/arm • netbsd/arm64 • openbsd/386 • openbsd/arm • openbsd/arm64 • openbsd/mips64 • plan9/386 • plan9/amd64 • plan9/arm • solaris/amd64 • windows/arm The previous approach for misc-compile target selection was to break them up primarily by GOOS value. However, as new architectures were added over time, some misc-compile TryBots got to a point where they were testing upwards of 5 ports (for example, misc-compile-openbsd was testing 386, amd64, arm, arm64, and mips64 architectures). Since each port is tested sequentially, allocating too many to one misc-compile TryBot can cause it to become the bottleneck of an entire TryBot run, exceeding the 10 minute completion time goal. Arrange it so misc-compile TryBot target selection is done explicitly in x/build, and pick 3 as max number of targets per TryBot for now. Based on recent timing observations, that should strike a decent balance between resource use (spinning up a builder) vs chance of a misc-compile TryBot becoming a bottleneck. It will also give us an opportunity to compare timing of 1, 2 and 3 targets per misc-compile in the future. (When we start tracking timing for TryBot completion time holistically, we'll be in a better position to refine this strategy further.) Making misc-compile target selection explicit in x/build also enables removing unnecessary duplicate misc-compile coverage from ports that already have a real TryBot (for example, openbsd/amd64 was previously tested via both the openbsd-amd64-68 TryBot and misc-compile-openbsd). This shouldn't be needed, so it's no longer done. For golang/go#17104. Fixes golang/go#32632. Change-Id: Iac918377b91af3e48780b38ffdf3153e213eeba2 Reviewed-on: https://go-review.googlesource.com/c/build/+/313210 Trust: Dmitri Shuralyov <dmitshur@golang.org> Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Heschi Kreinick <heschi@google.com> Reviewed-by: Carlos Amedee <carlos@golang.org>
2021-04-23 23:56:37 +03:00
"misc-compile-mac-win",
"misc-compile-mips",
dashboard: include at most 3 ports per misc-compile TryBot Between Go 1.15, 1.16 and tip, the following 29 ports don't have a real TryBot and instead rely on misc-compile TryBots for their pre-submit coverage: • aix/ppc64 • darwin/amd64 • darwin/arm64 • dragonfly/amd64 • freebsd/386 • freebsd/arm • freebsd/arm64 • illumos/amd64 • linux/mips • linux/mips64 • linux/mipsle • linux/mips64le • linux/ppc64 • linux/ppc64le • linux/riscv64 • linux/s390x • netbsd/386 • netbsd/amd64 • netbsd/arm • netbsd/arm64 • openbsd/386 • openbsd/arm • openbsd/arm64 • openbsd/mips64 • plan9/386 • plan9/amd64 • plan9/arm • solaris/amd64 • windows/arm The previous approach for misc-compile target selection was to break them up primarily by GOOS value. However, as new architectures were added over time, some misc-compile TryBots got to a point where they were testing upwards of 5 ports (for example, misc-compile-openbsd was testing 386, amd64, arm, arm64, and mips64 architectures). Since each port is tested sequentially, allocating too many to one misc-compile TryBot can cause it to become the bottleneck of an entire TryBot run, exceeding the 10 minute completion time goal. Arrange it so misc-compile TryBot target selection is done explicitly in x/build, and pick 3 as max number of targets per TryBot for now. Based on recent timing observations, that should strike a decent balance between resource use (spinning up a builder) vs chance of a misc-compile TryBot becoming a bottleneck. It will also give us an opportunity to compare timing of 1, 2 and 3 targets per misc-compile in the future. (When we start tracking timing for TryBot completion time holistically, we'll be in a better position to refine this strategy further.) Making misc-compile target selection explicit in x/build also enables removing unnecessary duplicate misc-compile coverage from ports that already have a real TryBot (for example, openbsd/amd64 was previously tested via both the openbsd-amd64-68 TryBot and misc-compile-openbsd). This shouldn't be needed, so it's no longer done. For golang/go#17104. Fixes golang/go#32632. Change-Id: Iac918377b91af3e48780b38ffdf3153e213eeba2 Reviewed-on: https://go-review.googlesource.com/c/build/+/313210 Trust: Dmitri Shuralyov <dmitshur@golang.org> Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Heschi Kreinick <heschi@google.com> Reviewed-by: Carlos Amedee <carlos@golang.org>
2021-04-23 23:56:37 +03:00
"misc-compile-mipsle",
"misc-compile-netbsd",
dashboard: include at most 3 ports per misc-compile TryBot Between Go 1.15, 1.16 and tip, the following 29 ports don't have a real TryBot and instead rely on misc-compile TryBots for their pre-submit coverage: • aix/ppc64 • darwin/amd64 • darwin/arm64 • dragonfly/amd64 • freebsd/386 • freebsd/arm • freebsd/arm64 • illumos/amd64 • linux/mips • linux/mips64 • linux/mipsle • linux/mips64le • linux/ppc64 • linux/ppc64le • linux/riscv64 • linux/s390x • netbsd/386 • netbsd/amd64 • netbsd/arm • netbsd/arm64 • openbsd/386 • openbsd/arm • openbsd/arm64 • openbsd/mips64 • plan9/386 • plan9/amd64 • plan9/arm • solaris/amd64 • windows/arm The previous approach for misc-compile target selection was to break them up primarily by GOOS value. However, as new architectures were added over time, some misc-compile TryBots got to a point where they were testing upwards of 5 ports (for example, misc-compile-openbsd was testing 386, amd64, arm, arm64, and mips64 architectures). Since each port is tested sequentially, allocating too many to one misc-compile TryBot can cause it to become the bottleneck of an entire TryBot run, exceeding the 10 minute completion time goal. Arrange it so misc-compile TryBot target selection is done explicitly in x/build, and pick 3 as max number of targets per TryBot for now. Based on recent timing observations, that should strike a decent balance between resource use (spinning up a builder) vs chance of a misc-compile TryBot becoming a bottleneck. It will also give us an opportunity to compare timing of 1, 2 and 3 targets per misc-compile in the future. (When we start tracking timing for TryBot completion time holistically, we'll be in a better position to refine this strategy further.) Making misc-compile target selection explicit in x/build also enables removing unnecessary duplicate misc-compile coverage from ports that already have a real TryBot (for example, openbsd/amd64 was previously tested via both the openbsd-amd64-68 TryBot and misc-compile-openbsd). This shouldn't be needed, so it's no longer done. For golang/go#17104. Fixes golang/go#32632. Change-Id: Iac918377b91af3e48780b38ffdf3153e213eeba2 Reviewed-on: https://go-review.googlesource.com/c/build/+/313210 Trust: Dmitri Shuralyov <dmitshur@golang.org> Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Heschi Kreinick <heschi@google.com> Reviewed-by: Carlos Amedee <carlos@golang.org>
2021-04-23 23:56:37 +03:00
"misc-compile-netbsd-arm",
"misc-compile-openbsd",
dashboard: include at most 3 ports per misc-compile TryBot Between Go 1.15, 1.16 and tip, the following 29 ports don't have a real TryBot and instead rely on misc-compile TryBots for their pre-submit coverage: • aix/ppc64 • darwin/amd64 • darwin/arm64 • dragonfly/amd64 • freebsd/386 • freebsd/arm • freebsd/arm64 • illumos/amd64 • linux/mips • linux/mips64 • linux/mipsle • linux/mips64le • linux/ppc64 • linux/ppc64le • linux/riscv64 • linux/s390x • netbsd/386 • netbsd/amd64 • netbsd/arm • netbsd/arm64 • openbsd/386 • openbsd/arm • openbsd/arm64 • openbsd/mips64 • plan9/386 • plan9/amd64 • plan9/arm • solaris/amd64 • windows/arm The previous approach for misc-compile target selection was to break them up primarily by GOOS value. However, as new architectures were added over time, some misc-compile TryBots got to a point where they were testing upwards of 5 ports (for example, misc-compile-openbsd was testing 386, amd64, arm, arm64, and mips64 architectures). Since each port is tested sequentially, allocating too many to one misc-compile TryBot can cause it to become the bottleneck of an entire TryBot run, exceeding the 10 minute completion time goal. Arrange it so misc-compile TryBot target selection is done explicitly in x/build, and pick 3 as max number of targets per TryBot for now. Based on recent timing observations, that should strike a decent balance between resource use (spinning up a builder) vs chance of a misc-compile TryBot becoming a bottleneck. It will also give us an opportunity to compare timing of 1, 2 and 3 targets per misc-compile in the future. (When we start tracking timing for TryBot completion time holistically, we'll be in a better position to refine this strategy further.) Making misc-compile target selection explicit in x/build also enables removing unnecessary duplicate misc-compile coverage from ports that already have a real TryBot (for example, openbsd/amd64 was previously tested via both the openbsd-amd64-68 TryBot and misc-compile-openbsd). This shouldn't be needed, so it's no longer done. For golang/go#17104. Fixes golang/go#32632. Change-Id: Iac918377b91af3e48780b38ffdf3153e213eeba2 Reviewed-on: https://go-review.googlesource.com/c/build/+/313210 Trust: Dmitri Shuralyov <dmitshur@golang.org> Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Heschi Kreinick <heschi@google.com> Reviewed-by: Carlos Amedee <carlos@golang.org>
2021-04-23 23:56:37 +03:00
"misc-compile-openbsd-arm",
"misc-compile-plan9",
"misc-compile-ppc",
dashboard: include at most 3 ports per misc-compile TryBot Between Go 1.15, 1.16 and tip, the following 29 ports don't have a real TryBot and instead rely on misc-compile TryBots for their pre-submit coverage: • aix/ppc64 • darwin/amd64 • darwin/arm64 • dragonfly/amd64 • freebsd/386 • freebsd/arm • freebsd/arm64 • illumos/amd64 • linux/mips • linux/mips64 • linux/mipsle • linux/mips64le • linux/ppc64 • linux/ppc64le • linux/riscv64 • linux/s390x • netbsd/386 • netbsd/amd64 • netbsd/arm • netbsd/arm64 • openbsd/386 • openbsd/arm • openbsd/arm64 • openbsd/mips64 • plan9/386 • plan9/amd64 • plan9/arm • solaris/amd64 • windows/arm The previous approach for misc-compile target selection was to break them up primarily by GOOS value. However, as new architectures were added over time, some misc-compile TryBots got to a point where they were testing upwards of 5 ports (for example, misc-compile-openbsd was testing 386, amd64, arm, arm64, and mips64 architectures). Since each port is tested sequentially, allocating too many to one misc-compile TryBot can cause it to become the bottleneck of an entire TryBot run, exceeding the 10 minute completion time goal. Arrange it so misc-compile TryBot target selection is done explicitly in x/build, and pick 3 as max number of targets per TryBot for now. Based on recent timing observations, that should strike a decent balance between resource use (spinning up a builder) vs chance of a misc-compile TryBot becoming a bottleneck. It will also give us an opportunity to compare timing of 1, 2 and 3 targets per misc-compile in the future. (When we start tracking timing for TryBot completion time holistically, we'll be in a better position to refine this strategy further.) Making misc-compile target selection explicit in x/build also enables removing unnecessary duplicate misc-compile coverage from ports that already have a real TryBot (for example, openbsd/amd64 was previously tested via both the openbsd-amd64-68 TryBot and misc-compile-openbsd). This shouldn't be needed, so it's no longer done. For golang/go#17104. Fixes golang/go#32632. Change-Id: Iac918377b91af3e48780b38ffdf3153e213eeba2 Reviewed-on: https://go-review.googlesource.com/c/build/+/313210 Trust: Dmitri Shuralyov <dmitshur@golang.org> Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Heschi Kreinick <heschi@google.com> Reviewed-by: Carlos Amedee <carlos@golang.org>
2021-04-23 23:56:37 +03:00
"misc-compile-other-1",
"misc-compile-other-2",
// Include longtest builders on Go repo release branches. See issue 37827.
"linux-386-longtest",
"linux-amd64-longtest",
"windows-amd64-longtest",
},
},
{
repo: "go",
branch: "release-branch.go1.15",
want: []string{
"android-amd64-emu",
"freebsd-amd64-12_2",
"js-wasm",
"linux-386",
"linux-amd64",
"linux-amd64-race",
"linux-arm-aws",
"linux-arm64-aws",
dashboard: include at most 3 ports per misc-compile TryBot Between Go 1.15, 1.16 and tip, the following 29 ports don't have a real TryBot and instead rely on misc-compile TryBots for their pre-submit coverage: • aix/ppc64 • darwin/amd64 • darwin/arm64 • dragonfly/amd64 • freebsd/386 • freebsd/arm • freebsd/arm64 • illumos/amd64 • linux/mips • linux/mips64 • linux/mipsle • linux/mips64le • linux/ppc64 • linux/ppc64le • linux/riscv64 • linux/s390x • netbsd/386 • netbsd/amd64 • netbsd/arm • netbsd/arm64 • openbsd/386 • openbsd/arm • openbsd/arm64 • openbsd/mips64 • plan9/386 • plan9/amd64 • plan9/arm • solaris/amd64 • windows/arm The previous approach for misc-compile target selection was to break them up primarily by GOOS value. However, as new architectures were added over time, some misc-compile TryBots got to a point where they were testing upwards of 5 ports (for example, misc-compile-openbsd was testing 386, amd64, arm, arm64, and mips64 architectures). Since each port is tested sequentially, allocating too many to one misc-compile TryBot can cause it to become the bottleneck of an entire TryBot run, exceeding the 10 minute completion time goal. Arrange it so misc-compile TryBot target selection is done explicitly in x/build, and pick 3 as max number of targets per TryBot for now. Based on recent timing observations, that should strike a decent balance between resource use (spinning up a builder) vs chance of a misc-compile TryBot becoming a bottleneck. It will also give us an opportunity to compare timing of 1, 2 and 3 targets per misc-compile in the future. (When we start tracking timing for TryBot completion time holistically, we'll be in a better position to refine this strategy further.) Making misc-compile target selection explicit in x/build also enables removing unnecessary duplicate misc-compile coverage from ports that already have a real TryBot (for example, openbsd/amd64 was previously tested via both the openbsd-amd64-68 TryBot and misc-compile-openbsd). This shouldn't be needed, so it's no longer done. For golang/go#17104. Fixes golang/go#32632. Change-Id: Iac918377b91af3e48780b38ffdf3153e213eeba2 Reviewed-on: https://go-review.googlesource.com/c/build/+/313210 Trust: Dmitri Shuralyov <dmitshur@golang.org> Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Heschi Kreinick <heschi@google.com> Reviewed-by: Carlos Amedee <carlos@golang.org>
2021-04-23 23:56:37 +03:00
"openbsd-amd64-68",
"windows-386-2008",
"windows-amd64-2016",
"misc-compile-freebsd",
dashboard: include at most 3 ports per misc-compile TryBot Between Go 1.15, 1.16 and tip, the following 29 ports don't have a real TryBot and instead rely on misc-compile TryBots for their pre-submit coverage: • aix/ppc64 • darwin/amd64 • darwin/arm64 • dragonfly/amd64 • freebsd/386 • freebsd/arm • freebsd/arm64 • illumos/amd64 • linux/mips • linux/mips64 • linux/mipsle • linux/mips64le • linux/ppc64 • linux/ppc64le • linux/riscv64 • linux/s390x • netbsd/386 • netbsd/amd64 • netbsd/arm • netbsd/arm64 • openbsd/386 • openbsd/arm • openbsd/arm64 • openbsd/mips64 • plan9/386 • plan9/amd64 • plan9/arm • solaris/amd64 • windows/arm The previous approach for misc-compile target selection was to break them up primarily by GOOS value. However, as new architectures were added over time, some misc-compile TryBots got to a point where they were testing upwards of 5 ports (for example, misc-compile-openbsd was testing 386, amd64, arm, arm64, and mips64 architectures). Since each port is tested sequentially, allocating too many to one misc-compile TryBot can cause it to become the bottleneck of an entire TryBot run, exceeding the 10 minute completion time goal. Arrange it so misc-compile TryBot target selection is done explicitly in x/build, and pick 3 as max number of targets per TryBot for now. Based on recent timing observations, that should strike a decent balance between resource use (spinning up a builder) vs chance of a misc-compile TryBot becoming a bottleneck. It will also give us an opportunity to compare timing of 1, 2 and 3 targets per misc-compile in the future. (When we start tracking timing for TryBot completion time holistically, we'll be in a better position to refine this strategy further.) Making misc-compile target selection explicit in x/build also enables removing unnecessary duplicate misc-compile coverage from ports that already have a real TryBot (for example, openbsd/amd64 was previously tested via both the openbsd-amd64-68 TryBot and misc-compile-openbsd). This shouldn't be needed, so it's no longer done. For golang/go#17104. Fixes golang/go#32632. Change-Id: Iac918377b91af3e48780b38ffdf3153e213eeba2 Reviewed-on: https://go-review.googlesource.com/c/build/+/313210 Trust: Dmitri Shuralyov <dmitshur@golang.org> Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Heschi Kreinick <heschi@google.com> Reviewed-by: Carlos Amedee <carlos@golang.org>
2021-04-23 23:56:37 +03:00
"misc-compile-mac-win",
"misc-compile-mips",
dashboard: include at most 3 ports per misc-compile TryBot Between Go 1.15, 1.16 and tip, the following 29 ports don't have a real TryBot and instead rely on misc-compile TryBots for their pre-submit coverage: • aix/ppc64 • darwin/amd64 • darwin/arm64 • dragonfly/amd64 • freebsd/386 • freebsd/arm • freebsd/arm64 • illumos/amd64 • linux/mips • linux/mips64 • linux/mipsle • linux/mips64le • linux/ppc64 • linux/ppc64le • linux/riscv64 • linux/s390x • netbsd/386 • netbsd/amd64 • netbsd/arm • netbsd/arm64 • openbsd/386 • openbsd/arm • openbsd/arm64 • openbsd/mips64 • plan9/386 • plan9/amd64 • plan9/arm • solaris/amd64 • windows/arm The previous approach for misc-compile target selection was to break them up primarily by GOOS value. However, as new architectures were added over time, some misc-compile TryBots got to a point where they were testing upwards of 5 ports (for example, misc-compile-openbsd was testing 386, amd64, arm, arm64, and mips64 architectures). Since each port is tested sequentially, allocating too many to one misc-compile TryBot can cause it to become the bottleneck of an entire TryBot run, exceeding the 10 minute completion time goal. Arrange it so misc-compile TryBot target selection is done explicitly in x/build, and pick 3 as max number of targets per TryBot for now. Based on recent timing observations, that should strike a decent balance between resource use (spinning up a builder) vs chance of a misc-compile TryBot becoming a bottleneck. It will also give us an opportunity to compare timing of 1, 2 and 3 targets per misc-compile in the future. (When we start tracking timing for TryBot completion time holistically, we'll be in a better position to refine this strategy further.) Making misc-compile target selection explicit in x/build also enables removing unnecessary duplicate misc-compile coverage from ports that already have a real TryBot (for example, openbsd/amd64 was previously tested via both the openbsd-amd64-68 TryBot and misc-compile-openbsd). This shouldn't be needed, so it's no longer done. For golang/go#17104. Fixes golang/go#32632. Change-Id: Iac918377b91af3e48780b38ffdf3153e213eeba2 Reviewed-on: https://go-review.googlesource.com/c/build/+/313210 Trust: Dmitri Shuralyov <dmitshur@golang.org> Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Heschi Kreinick <heschi@google.com> Reviewed-by: Carlos Amedee <carlos@golang.org>
2021-04-23 23:56:37 +03:00
"misc-compile-mipsle",
"misc-compile-netbsd",
dashboard: include at most 3 ports per misc-compile TryBot Between Go 1.15, 1.16 and tip, the following 29 ports don't have a real TryBot and instead rely on misc-compile TryBots for their pre-submit coverage: • aix/ppc64 • darwin/amd64 • darwin/arm64 • dragonfly/amd64 • freebsd/386 • freebsd/arm • freebsd/arm64 • illumos/amd64 • linux/mips • linux/mips64 • linux/mipsle • linux/mips64le • linux/ppc64 • linux/ppc64le • linux/riscv64 • linux/s390x • netbsd/386 • netbsd/amd64 • netbsd/arm • netbsd/arm64 • openbsd/386 • openbsd/arm • openbsd/arm64 • openbsd/mips64 • plan9/386 • plan9/amd64 • plan9/arm • solaris/amd64 • windows/arm The previous approach for misc-compile target selection was to break them up primarily by GOOS value. However, as new architectures were added over time, some misc-compile TryBots got to a point where they were testing upwards of 5 ports (for example, misc-compile-openbsd was testing 386, amd64, arm, arm64, and mips64 architectures). Since each port is tested sequentially, allocating too many to one misc-compile TryBot can cause it to become the bottleneck of an entire TryBot run, exceeding the 10 minute completion time goal. Arrange it so misc-compile TryBot target selection is done explicitly in x/build, and pick 3 as max number of targets per TryBot for now. Based on recent timing observations, that should strike a decent balance between resource use (spinning up a builder) vs chance of a misc-compile TryBot becoming a bottleneck. It will also give us an opportunity to compare timing of 1, 2 and 3 targets per misc-compile in the future. (When we start tracking timing for TryBot completion time holistically, we'll be in a better position to refine this strategy further.) Making misc-compile target selection explicit in x/build also enables removing unnecessary duplicate misc-compile coverage from ports that already have a real TryBot (for example, openbsd/amd64 was previously tested via both the openbsd-amd64-68 TryBot and misc-compile-openbsd). This shouldn't be needed, so it's no longer done. For golang/go#17104. Fixes golang/go#32632. Change-Id: Iac918377b91af3e48780b38ffdf3153e213eeba2 Reviewed-on: https://go-review.googlesource.com/c/build/+/313210 Trust: Dmitri Shuralyov <dmitshur@golang.org> Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Heschi Kreinick <heschi@google.com> Reviewed-by: Carlos Amedee <carlos@golang.org>
2021-04-23 23:56:37 +03:00
"misc-compile-netbsd-arm",
"misc-compile-openbsd",
dashboard: include at most 3 ports per misc-compile TryBot Between Go 1.15, 1.16 and tip, the following 29 ports don't have a real TryBot and instead rely on misc-compile TryBots for their pre-submit coverage: • aix/ppc64 • darwin/amd64 • darwin/arm64 • dragonfly/amd64 • freebsd/386 • freebsd/arm • freebsd/arm64 • illumos/amd64 • linux/mips • linux/mips64 • linux/mipsle • linux/mips64le • linux/ppc64 • linux/ppc64le • linux/riscv64 • linux/s390x • netbsd/386 • netbsd/amd64 • netbsd/arm • netbsd/arm64 • openbsd/386 • openbsd/arm • openbsd/arm64 • openbsd/mips64 • plan9/386 • plan9/amd64 • plan9/arm • solaris/amd64 • windows/arm The previous approach for misc-compile target selection was to break them up primarily by GOOS value. However, as new architectures were added over time, some misc-compile TryBots got to a point where they were testing upwards of 5 ports (for example, misc-compile-openbsd was testing 386, amd64, arm, arm64, and mips64 architectures). Since each port is tested sequentially, allocating too many to one misc-compile TryBot can cause it to become the bottleneck of an entire TryBot run, exceeding the 10 minute completion time goal. Arrange it so misc-compile TryBot target selection is done explicitly in x/build, and pick 3 as max number of targets per TryBot for now. Based on recent timing observations, that should strike a decent balance between resource use (spinning up a builder) vs chance of a misc-compile TryBot becoming a bottleneck. It will also give us an opportunity to compare timing of 1, 2 and 3 targets per misc-compile in the future. (When we start tracking timing for TryBot completion time holistically, we'll be in a better position to refine this strategy further.) Making misc-compile target selection explicit in x/build also enables removing unnecessary duplicate misc-compile coverage from ports that already have a real TryBot (for example, openbsd/amd64 was previously tested via both the openbsd-amd64-68 TryBot and misc-compile-openbsd). This shouldn't be needed, so it's no longer done. For golang/go#17104. Fixes golang/go#32632. Change-Id: Iac918377b91af3e48780b38ffdf3153e213eeba2 Reviewed-on: https://go-review.googlesource.com/c/build/+/313210 Trust: Dmitri Shuralyov <dmitshur@golang.org> Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Heschi Kreinick <heschi@google.com> Reviewed-by: Carlos Amedee <carlos@golang.org>
2021-04-23 23:56:37 +03:00
"misc-compile-openbsd-arm",
"misc-compile-plan9",
"misc-compile-ppc",
dashboard: include at most 3 ports per misc-compile TryBot Between Go 1.15, 1.16 and tip, the following 29 ports don't have a real TryBot and instead rely on misc-compile TryBots for their pre-submit coverage: • aix/ppc64 • darwin/amd64 • darwin/arm64 • dragonfly/amd64 • freebsd/386 • freebsd/arm • freebsd/arm64 • illumos/amd64 • linux/mips • linux/mips64 • linux/mipsle • linux/mips64le • linux/ppc64 • linux/ppc64le • linux/riscv64 • linux/s390x • netbsd/386 • netbsd/amd64 • netbsd/arm • netbsd/arm64 • openbsd/386 • openbsd/arm • openbsd/arm64 • openbsd/mips64 • plan9/386 • plan9/amd64 • plan9/arm • solaris/amd64 • windows/arm The previous approach for misc-compile target selection was to break them up primarily by GOOS value. However, as new architectures were added over time, some misc-compile TryBots got to a point where they were testing upwards of 5 ports (for example, misc-compile-openbsd was testing 386, amd64, arm, arm64, and mips64 architectures). Since each port is tested sequentially, allocating too many to one misc-compile TryBot can cause it to become the bottleneck of an entire TryBot run, exceeding the 10 minute completion time goal. Arrange it so misc-compile TryBot target selection is done explicitly in x/build, and pick 3 as max number of targets per TryBot for now. Based on recent timing observations, that should strike a decent balance between resource use (spinning up a builder) vs chance of a misc-compile TryBot becoming a bottleneck. It will also give us an opportunity to compare timing of 1, 2 and 3 targets per misc-compile in the future. (When we start tracking timing for TryBot completion time holistically, we'll be in a better position to refine this strategy further.) Making misc-compile target selection explicit in x/build also enables removing unnecessary duplicate misc-compile coverage from ports that already have a real TryBot (for example, openbsd/amd64 was previously tested via both the openbsd-amd64-68 TryBot and misc-compile-openbsd). This shouldn't be needed, so it's no longer done. For golang/go#17104. Fixes golang/go#32632. Change-Id: Iac918377b91af3e48780b38ffdf3153e213eeba2 Reviewed-on: https://go-review.googlesource.com/c/build/+/313210 Trust: Dmitri Shuralyov <dmitshur@golang.org> Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Heschi Kreinick <heschi@google.com> Reviewed-by: Carlos Amedee <carlos@golang.org>
2021-04-23 23:56:37 +03:00
"misc-compile-other-1",
"misc-compile-other-2",
// Include longtest builders on Go repo release branches. See issue 37827.
"linux-386-longtest",
"linux-amd64-longtest",
"windows-amd64-longtest",
},
},
{
repo: "mobile",
branch: "master",
want: []string{
"android-amd64-emu",
"linux-amd64-androidemu",
},
},
{
repo: "sys",
branch: "master",
want: []string{
"android-amd64-emu",
"freebsd-386-11_4",
"freebsd-amd64-11_4",
"freebsd-amd64-12_2",
"linux-386",
"linux-amd64",
"linux-amd64-race",
"linux-arm-aws",
"linux-arm64-aws",
"netbsd-amd64-9_0",
"openbsd-386-68",
"openbsd-amd64-68",
"windows-386-2008",
"windows-amd64-2016",
},
},
{
repo: "exp",
branch: "master",
want: []string{
"linux-amd64",
"linux-amd64-race",
"windows-386-2008",
"windows-amd64-2016",
},
},
}
for i, tt := range tests {
if tt.branch == "" || tt.repo == "" {
t.Errorf("incomplete test entry %d", i)
return
}
t.Run(fmt.Sprintf("%s/%s", tt.repo, tt.branch), func(t *testing.T) {
var got []string
goBranch := tt.branch // hard-code the common case for now
for _, bc := range TryBuildersForProject(tt.repo, tt.branch, goBranch) {
got = append(got, bc.Name)
}
m := map[string]bool{}
for _, b := range tt.want {
m[b] = true
}
for _, b := range got {
if _, ok := m[b]; !ok {
t.Errorf("got unexpected %q", b)
}
delete(m, b)
}
for b := range m {
t.Errorf("missing expected %q", b)
}
})
}
}
// TestBuilderConfig whether a given builder and repo at different
// branches is either a post-submit builder, trybot, neither, or both.
func TestBuilderConfig(t *testing.T) {
// builderConfigWant is bitmask of 4 different things to assert are wanted:
// - being a post-submit builder
// - NOT being a post-submit builder
// - being a trybot builder
// - NOT being a post-submit builder
type want uint8
const (
isTrybot want = 1 << iota
notTrybot
isBuilder // post-submit
notBuilder // not post-submit
none = notTrybot + notBuilder
both = isTrybot + isBuilder
onlyPost = notTrybot + isBuilder
)
type builderAndRepo struct {
testName string
builder string
repo string
branch string
goBranch string
}
// builder may end in "@go1.N" or "@1.N" (as alias for "@release-branch.go1.N") or "@branch-name".
// repo (other than "go") may end in "@go1.N" or "@1.N" (as alias for "@release-branch.go1.N").
b := func(builder, repo string) builderAndRepo {
br := builderAndRepo{
testName: builder + "," + repo,
builder: builder,
goBranch: "master",
repo: repo,
branch: "master",
}
if strings.Contains(builder, "@") {
f := strings.SplitN(builder, "@", 2)
br.builder = f[0]
br.goBranch = f[1]
}
if strings.Contains(repo, "@") {
f := strings.SplitN(repo, "@", 2)
br.repo = f[0]
br.branch = f[1]
if br.repo == "go" {
panic(fmt.Errorf(`b(%q, %q): for "go" repo, must use the @%s suffix on the builder, not on the repo`, builder, repo, br.branch))
}
}
expandBranch := func(s *string) {
if strings.HasPrefix(*s, "go1.") {
*s = "release-branch." + *s
} else if strings.HasPrefix(*s, "1.") {
*s = "release-branch.go" + *s
}
}
expandBranch(&br.branch)
expandBranch(&br.goBranch)
if br.repo == "go" {
br.branch = br.goBranch
}
return br
}
tests := []struct {
br builderAndRepo
want want
}{
{b("linux-amd64", "go"), both},
{b("linux-amd64", "net"), both},
{b("linux-amd64", "sys"), both},
{b("linux-amd64", "website"), both},
// Don't test all subrepos on all the builders.
{b("linux-amd64-ssacheck", "net"), none},
{b("linux-amd64-ssacheck@go1.15", "net"), none},
{b("linux-386-softfloat", "crypto"), onlyPost},
{b("linux-386-softfloat@go1.16", "crypto"), onlyPost},
{b("linux-386-softfloat@go1.15", "crypto"), none},
{b("linux-386-387", "crypto"), none},
{b("linux-386-387@go1.16", "crypto"), none},
{b("linux-386-387@go1.15", "crypto"), onlyPost},
{b("linux-386-387@go1.15", "crypto@1.15"), onlyPost},
{b("linux-386-387@go1.15", "build"), none},
{b("linux-386-387@go1.14", "crypto"), onlyPost},
{b("linux-arm-arm5spacemonkey@go1.15", "net"), none},
{b("linux-arm-arm5spacemonkey", "exp"), none},
{b("linux-arm-arm5spacemonkey", "mobile"), none},
// The mobile repo requires Go 1.13+.
{b("android-amd64-emu", "mobile"), both},
{b("android-amd64-emu", "mobile@1.10"), none},
{b("android-amd64-emu", "mobile@1.11"), none},
{b("android-amd64-emu@go1.10", "mobile"), none},
{b("android-amd64-emu@go1.12", "mobile"), none},
{b("android-amd64-emu@go1.13", "mobile"), both},
{b("android-amd64-emu", "mobile@1.13"), both},
{b("android-amd64-emu", "go"), both},
{b("android-amd64-emu", "crypto"), both},
{b("android-amd64-emu", "net"), both},
{b("android-amd64-emu", "sync"), both},
{b("android-amd64-emu", "sys"), both},
{b("android-amd64-emu", "text"), both},
{b("android-amd64-emu", "time"), both},
{b("android-amd64-emu", "tools"), both},
{b("android-amd64-emu", "website"), none},
{b("android-386-emu", "go"), onlyPost},
{b("android-386-emu", "mobile"), onlyPost},
{b("android-386-emu", "mobile@1.10"), none},
{b("android-386-emu", "mobile@1.11"), none},
{b("android-386-emu@go1.10", "mobile"), none},
{b("android-386-emu@go1.12", "mobile"), none},
{b("android-386-emu@go1.13", "mobile"), onlyPost},
{b("android-386-emu", "mobile@1.13"), onlyPost},
{b("linux-amd64", "net"), both},
{b("linux-amd64", "net@1.12"), both},
{b("linux-amd64@go1.12", "net@1.12"), both},
{b("linux-amd64", "net@1.11"), both},
{b("linux-amd64", "net@1.11"), both},
{b("linux-amd64", "net@1.10"), none}, // too old
{b("linux-amd64@go1.10", "net"), none}, // too old
{b("linux-amd64@go1.12", "net@1.12"), both},
{b("linux-mips64le-mengzhuo", "go"), onlyPost},
{b("linux-mips64le-mengzhuo", "sys"), onlyPost},
{b("linux-mips64le-mengzhuo", "net"), onlyPost},
// OpenBSD.
{b("openbsd-amd64-68@go1.16", "go"), both},
{b("openbsd-amd64-68@go1.15", "go"), both},
{b("openbsd-amd64-68@go1.14", "go"), both},
{b("openbsd-amd64-64@go1.16", "go"), onlyPost},
{b("openbsd-amd64-64@go1.15", "go"), onlyPost},
{b("openbsd-amd64-64@go1.14", "go"), onlyPost},
{b("openbsd-amd64-62@go1.16", "go"), none}, // golang.org/issue/42426.
{b("openbsd-amd64-62@go1.15", "go"), onlyPost},
{b("openbsd-amd64-62@go1.14", "go"), onlyPost},
// FreeBSD 12.2
{b("freebsd-amd64-12_2", "go"), both},
{b("freebsd-amd64-12_2", "net"), both},
{b("freebsd-amd64-12_2", "mobile"), none},
{b("freebsd-386-12_2", "go"), onlyPost},
{b("freebsd-386-12_2", "net"), onlyPost},
{b("freebsd-386-12_2", "mobile"), none},
// FreeBSD 11.2
// See golang.org/issue/45727
{b("freebsd-amd64-11_2@go1.15", "go"), isBuilder},
{b("freebsd-amd64-11_2@go1.15", "net"), onlyPost},
{b("freebsd-amd64-11_2@go1.15", "sys"), both},
{b("freebsd-amd64-11_2", "go"), none},
{b("freebsd-386-11_2@go1.16", "go"), isBuilder},
{b("freebsd-386-11_2@go1.16", "net"), onlyPost},
{b("freebsd-386-11_2@go1.16", "sys"), both},
{b("freebsd-386-11_2", "go"), none},
// NetBSD
{b("netbsd-amd64-9_0", "go"), onlyPost},
{b("netbsd-amd64-9_0", "net"), onlyPost},
{b("netbsd-amd64-9_0", "sys"), both},
{b("netbsd-386-9_0", "go"), onlyPost},
{b("netbsd-386-9_0", "net"), onlyPost},
// AIX starts at Go 1.12
{b("aix-ppc64", "go"), onlyPost},
{b("aix-ppc64", "net"), onlyPost},
{b("aix-ppc64", "mobile"), none},
{b("aix-ppc64", "exp"), none},
{b("aix-ppc64", "term"), onlyPost},
{b("aix-ppc64@go1.15", "go"), onlyPost},
{b("aix-ppc64@go1.15", "net"), onlyPost},
{b("aix-ppc64@go1.15", "mobile"), none},
{b("aix-ppc64@go1.16", "net"), onlyPost},
{b("aix-ppc64@go1.16", "mobile"), none},
{b("aix-ppc64@dev.link", "go"), onlyPost},
{b("linux-amd64-nocgo", "mobile"), none},
// Virtual mobiledevices
{b("ios-arm64-corellium", "go"), isBuilder},
{b("android-arm64-corellium", "go"), isBuilder},
{b("android-arm-corellium", "go"), isBuilder},
// Mobile builders that run with GOOS=linux/ios and have
// a device attached.
{b("linux-amd64-androidemu", "mobile"), both},
// But the emulators run all:
{b("android-amd64-emu", "mobile"), isBuilder},
{b("android-386-emu", "mobile"), isBuilder},
{b("android-amd64-emu", "net"), isBuilder},
{b("android-386-emu", "net"), isBuilder},
{b("android-amd64-emu", "go"), isBuilder},
{b("android-386-emu", "go"), isBuilder},
// Only test tip for js/wasm, and only for some repos:
{b("js-wasm", "go"), both},
{b("js-wasm", "arch"), onlyPost},
{b("js-wasm", "crypto"), onlyPost},
{b("js-wasm", "sys"), onlyPost},
{b("js-wasm", "net"), onlyPost},
{b("js-wasm@go1.12", "net"), none},
{b("js-wasm", "benchmarks"), none},
{b("js-wasm", "debug"), none},
{b("js-wasm", "mobile"), none},
{b("js-wasm", "perf"), none},
{b("js-wasm", "talks"), none},
{b("js-wasm", "tools"), none},
{b("js-wasm", "tour"), none},
{b("js-wasm", "website"), none},
// Race builders. Linux for all, GCE buidlers for
// post-submit, and only post-submit for "go" for
// Darwin (limited resources).
{b("linux-amd64-race", "go"), both},
{b("linux-amd64-race", "net"), both},
{b("windows-amd64-race", "go"), onlyPost},
{b("windows-amd64-race", "net"), onlyPost},
{b("freebsd-amd64-race", "go"), onlyPost},
{b("freebsd-amd64-race", "net"), onlyPost},
{b("darwin-amd64-race", "go"), onlyPost},
{b("darwin-amd64-race", "net"), none},
// Long test.
{b("linux-amd64-longtest", "go"), onlyPost},
{b("linux-amd64-longtest", "net"), onlyPost},
{b("linux-amd64-longtest@go1.14", "go"), both},
{b("linux-amd64-longtest@go1.14", "net"), none},
{b("windows-amd64-longtest", "go"), onlyPost},
{b("windows-amd64-longtest@go1.14", "go"), both},
{b("windows-amd64-longtest", "net"), onlyPost},
{b("windows-amd64-longtest", "exp"), onlyPost},
{b("windows-amd64-longtest", "mobile"), none},
{b("linux-386-longtest", "go"), onlyPost},
{b("linux-386-longtest", "net"), onlyPost},
{b("linux-386-longtest", "exp"), none},
{b("linux-386-longtest", "mobile"), none},
// Experimental exp repo runs in very few places.
{b("linux-amd64", "exp"), both},
{b("linux-amd64-race", "exp"), both},
{b("linux-amd64-longtest", "exp"), onlyPost},
{b("windows-386-2008", "exp"), both},
{b("windows-amd64-2016", "exp"), both},
{b("darwin-amd64-10_14", "exp"), onlyPost},
{b("darwin-amd64-10_15", "exp"), onlyPost},
// ... but not on most others:
{b("darwin-amd64-10_12", "exp"), none},
{b("freebsd-386-11_4", "exp"), none},
{b("freebsd-386-12_2", "exp"), none},
{b("freebsd-amd64-11_4", "exp"), none},
{b("freebsd-amd64-12_2", "exp"), none},
{b("openbsd-amd64-64", "exp"), none},
{b("openbsd-amd64-68", "exp"), none},
{b("js-wasm", "exp"), none},
// exp is experimental; it doesn't test against release branches.
{b("linux-amd64@go1.12", "exp"), none},
// the build repo is only really useful for linux-amd64 (where we run it),
// and darwin-amd64 and perhaps windows-amd64 (for stuff like gomote).
// No need for any other operating systems to use it.
{b("linux-amd64", "build"), both},
{b("linux-amd64-longtest", "build"), onlyPost},
{b("windows-amd64-2016", "build"), both},
{b("darwin-amd64-10_12", "build"), none},
{b("darwin-amd64-10_14", "build"), none},
{b("darwin-amd64-10_15", "build"), onlyPost},
{b("openbsd-amd64-68", "build"), none},
{b("linux-amd64-fedora", "build"), none},
{b("linux-amd64-clang", "build"), none},
{b("linux-amd64-sid", "build"), none},
{b("linux-amd64-nocgo", "build"), none},
{b("linux-386-longtest", "build"), none},
{b("js-wasm", "build"), none},
{b("android-386-emu", "build"), none},
{b("android-amd64-emu", "build"), none},
// Only use latest macOS for subrepos, and only amd64:
{b("darwin-amd64-10_12@go1.16", "net"), onlyPost},
{b("darwin-amd64-10_12", "net"), none},
{b("darwin-amd64-10_14", "net"), onlyPost},
{b("darwin-amd64-10_15", "go"), onlyPost},
{b("darwin-amd64-10_14", "go"), onlyPost},
{b("darwin-amd64-10_12", "go"), none},
{b("darwin-amd64-10_12@go1.16", "go"), onlyPost},
// plan9 only lived at master. We didn't support any past releases.
// But it's off for now as it's always failing.
{b("plan9-386", "go"), none}, // temporarily disabled
{b("plan9-386", "net"), none}, // temporarily disabled
{b("plan9-386", "exp"), none},
{b("plan9-386", "mobile"), none},
{b("plan9-386@go1.12", "go"), none},
{b("plan9-386@go1.12", "net"), none},
{b("plan9-amd64-0intro", "go"), onlyPost},
{b("plan9-amd64-0intro", "exp"), none},
{b("plan9-amd64-0intro", "mobile"), none},
{b("plan9-amd64-0intro@go1.12", "go"), none},
{b("plan9-amd64-0intro", "net"), onlyPost},
{b("plan9-amd64-0intro@go1.12", "net"), none},
{b("plan9-arm", "go"), onlyPost},
{b("plan9-arm", "exp"), none},
{b("plan9-arm", "mobile"), none},
{b("plan9-arm@go1.12", "go"), none},
{b("plan9-arm", "net"), onlyPost},
{b("plan9-arm@go1.12", "net"), none},
{b("dragonfly-amd64", "go"), onlyPost},
{b("dragonfly-amd64", "net"), onlyPost},
{b("dragonfly-amd64@go1.13", "net"), none}, // Dragonfly ABI changes only supported by Go 1.14+
{b("dragonfly-amd64@go1.13", "go"), none}, // Dragonfly ABI changes only supported by Go 1.14+
{b("dragonfly-amd64-5_8", "go"), onlyPost},
{b("dragonfly-amd64-5_8", "net"), onlyPost},
{b("dragonfly-amd64-5_8@go1.13", "net"), onlyPost},
{b("linux-amd64-staticlockranking", "go"), onlyPost},
{b("linux-amd64-staticlockranking@go1.15", "go"), onlyPost},
{b("linux-amd64-staticlockranking@go1.14", "go"), none},
{b("linux-amd64-staticlockranking", "net"), none},
{b("linux-amd64-noregabi", "go"), onlyPost},
{b("linux-amd64-noregabi", "net"), none},
}
for _, tt := range tests {
t.Run(tt.br.testName, func(t *testing.T) {
bc, ok := Builders[tt.br.builder]
if !ok {
t.Fatalf("unknown builder %q", tt.br.builder)
}
gotPost := bc.BuildsRepoPostSubmit(tt.br.repo, tt.br.branch, tt.br.goBranch)
if tt.want&isBuilder != 0 && !gotPost {
t.Errorf("not a post-submit builder, but expected")
}
if tt.want&notBuilder != 0 && gotPost {
t.Errorf("unexpectedly a post-submit builder")
}
gotTry := bc.BuildsRepoTryBot(tt.br.repo, tt.br.branch, tt.br.goBranch)
if tt.want&isTrybot != 0 && !gotTry {
t.Errorf("not trybot, but expected")
}
if tt.want&notTrybot != 0 && gotTry {
t.Errorf("unexpectedly a trybot")
}
if t.Failed() {
t.Logf("For: %+v", tt.br)
}
})
}
}
func TestHostConfigsAllUsed(t *testing.T) {
knownUnused := map[string]bool{
// Currently host-linux-armhf-cross and host-linux-armel-cross aren't
// referenced, but the coordinator hard-codes them, so don't make
// these two an error for now.
"host-linux-armhf-cross": true,
"host-linux-armel-cross": true,
"host-linux-x86-alpine": true, // TODO(golang.org/issue/19938): Fix the Alpine builder, or remove it.
}
used := make(map[string]bool)
for _, conf := range Builders {
used[conf.HostType] = true
}
for hostType := range Hosts {
if !used[hostType] && !knownUnused[hostType] {
t.Errorf("host type %q is not referenced from any build config", hostType)
}
if used[hostType] && knownUnused[hostType] {
t.Errorf("host type %q should not be listed in knownUnused since it's in use", hostType)
}
}
}
// tests that goBranch is optional for repo == "go"
func TestBuildsRepoAtAllImplicitGoBranch(t *testing.T) {
builder := Builders["android-amd64-emu"]
got := builder.buildsRepoAtAll("go", "master", "")
if !got {
t.Error("got = false; want true")
}
}
func TestShouldRunDistTest(t *testing.T) {
type buildMode int
const (
tryMode buildMode = 0
postSubmit buildMode = 1
)
tests := []struct {
builder string
test string
mode buildMode
want bool
}{
{"linux-amd64", "api", postSubmit, true},
{"linux-amd64", "api", tryMode, true},
{"freebsd-amd64-12_2", "api", postSubmit, true}, // freebsd-amd64-12_2 uses fasterTrybots policy, should still build.
{"freebsd-amd64-12_2", "api", tryMode, false}, // freebsd-amd64-12_2 uses fasterTrybots policy, should skip in try mode.
{"linux-amd64", "reboot", tryMode, true},
{"linux-amd64-race", "reboot", tryMode, false},
{"darwin-amd64-10_12", "test:foo", postSubmit, false},
{"darwin-amd64-10_14", "test:foo", postSubmit, false},
{"darwin-amd64-10_14", "reboot", postSubmit, false},
{"darwin-amd64-10_14", "api", postSubmit, false},
{"darwin-amd64-10_14", "codewalk", postSubmit, false},
{"darwin-amd64-10_15", "test:foo", postSubmit, false},
}
for _, tt := range tests {
bc, ok := Builders[tt.builder]
if !ok {
t.Errorf("unknown builder %q", tt.builder)
continue
}
isTry := tt.mode == tryMode
if isTry && !bc.BuildsRepoTryBot("go", "master", "master") {
t.Errorf("builder %q is not a trybot, so can't run test %q in try mode", tt.builder, tt.test)
continue
}
got := bc.ShouldRunDistTest(tt.test, isTry)
if got != tt.want {
t.Errorf("%q.ShouldRunDistTest(%q, try %v) = %v; want %v", tt.builder, tt.test, isTry, got, tt.want)
}
}
}
func TestShouldTestPackageInGOPATHMode(t *testing.T) {
// This function doesn't change behavior depending on the builder
// at this time, so just use a common one.
bc, ok := Builders["linux-amd64"]
if !ok {
t.Fatal("unknown builder")
}
tests := []struct {
importPath string
want bool
}{
{"golang.org/x/image/bmp", true},
{"golang.org/x/tools/go/ast/astutil", true},
{"golang.org/x/tools/go/packages", true},
{"golang.org/x/tools", true}, // Three isn't a package there, but if there was, it should be tested.
{"golang.org/x/tools/gopls", false},
{"golang.org/x/tools/gopls/internal/foobar", false},
}
for _, tt := range tests {
got := bc.ShouldTestPackageInGOPATHMode(tt.importPath)
if got != tt.want {
t.Errorf("ShouldTestPackageInGOPATHMode(%q) = %v; want %v", tt.importPath, got, tt.want)
}
}
}
func TestSlowBotAliases(t *testing.T) {
for term, name := range slowBotAliases {
if name == "" {
// Empty string means known missing builder.
continue
}
if _, ok := Builders[name]; !ok {
t.Errorf("slowbot term %q references unknown builder %q", term, name)
}
}
out, err := exec.Command(filepath.Join(runtime.GOROOT(), "bin", "go"), "tool", "dist", "list").Output()
if err != nil {
t.Errorf("dist list: %v", err)
}
ports := strings.Fields(string(out))
done := map[string]bool{}
var add bytes.Buffer
check := func(term string, isArch bool) {
if done[term] {
return
}
done[term] = true
_, isBuilderName := Builders[term]
_, hasAlias := slowBotAliases[term]
if !isBuilderName && !hasAlias {
prefix := term
if isArch {
prefix = "linux-" + term
}
var matches []string
for name := range Builders {
if strings.HasPrefix(name, prefix) {
matches = append(matches, name)
}
}
sort.Strings(matches)
t.Errorf("term %q has no match in slowBotAliases", term)
if len(matches) == 1 {
fmt.Fprintf(&add, "%q: %q,\n", term, matches[0])
} else if len(matches) > 1 {
t.Errorf("maybe add: %q: %q, (matches=%q)", term, matches[len(matches)-1], matches)
}
}
}
for _, port := range ports {
slash := strings.IndexByte(port, '/')
if slash == -1 {
t.Fatalf("unexpected port %q", port)
}
goos, goarch := port[:slash], port[slash+1:]
check(goos+"-"+goarch, false)
check(goos, false)
check(goarch, true)
}
if add.Len() > 0 {
t.Errorf("Missing items from slowBotAliases:\n%s", add.String())
}
}
func TestCrossCompileConfigs(t *testing.T) {
// Verify that Builders.CrossCompileConfig have valid host types.
for name, bc := range Builders {
cc := bc.CrossCompileConfig
if cc == nil {
continue
}
if _, ok := Hosts[cc.CompileHostType]; !ok {
t.Errorf("unknown host type %q for builder %q", cc.CompileHostType, name)
}
}
}
dashboard: include at most 3 ports per misc-compile TryBot Between Go 1.15, 1.16 and tip, the following 29 ports don't have a real TryBot and instead rely on misc-compile TryBots for their pre-submit coverage: • aix/ppc64 • darwin/amd64 • darwin/arm64 • dragonfly/amd64 • freebsd/386 • freebsd/arm • freebsd/arm64 • illumos/amd64 • linux/mips • linux/mips64 • linux/mipsle • linux/mips64le • linux/ppc64 • linux/ppc64le • linux/riscv64 • linux/s390x • netbsd/386 • netbsd/amd64 • netbsd/arm • netbsd/arm64 • openbsd/386 • openbsd/arm • openbsd/arm64 • openbsd/mips64 • plan9/386 • plan9/amd64 • plan9/arm • solaris/amd64 • windows/arm The previous approach for misc-compile target selection was to break them up primarily by GOOS value. However, as new architectures were added over time, some misc-compile TryBots got to a point where they were testing upwards of 5 ports (for example, misc-compile-openbsd was testing 386, amd64, arm, arm64, and mips64 architectures). Since each port is tested sequentially, allocating too many to one misc-compile TryBot can cause it to become the bottleneck of an entire TryBot run, exceeding the 10 minute completion time goal. Arrange it so misc-compile TryBot target selection is done explicitly in x/build, and pick 3 as max number of targets per TryBot for now. Based on recent timing observations, that should strike a decent balance between resource use (spinning up a builder) vs chance of a misc-compile TryBot becoming a bottleneck. It will also give us an opportunity to compare timing of 1, 2 and 3 targets per misc-compile in the future. (When we start tracking timing for TryBot completion time holistically, we'll be in a better position to refine this strategy further.) Making misc-compile target selection explicit in x/build also enables removing unnecessary duplicate misc-compile coverage from ports that already have a real TryBot (for example, openbsd/amd64 was previously tested via both the openbsd-amd64-68 TryBot and misc-compile-openbsd). This shouldn't be needed, so it's no longer done. For golang/go#17104. Fixes golang/go#32632. Change-Id: Iac918377b91af3e48780b38ffdf3153e213eeba2 Reviewed-on: https://go-review.googlesource.com/c/build/+/313210 Trust: Dmitri Shuralyov <dmitshur@golang.org> Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Heschi Kreinick <heschi@google.com> Reviewed-by: Carlos Amedee <carlos@golang.org>
2021-04-23 23:56:37 +03:00
// TestTryBotsCompileAllPorts verifies that each port (go tool dist list)
// is covered by either a real TryBot or a misc-compile TryBot.
//
// The special pseudo-port 'linux-arm-arm5' is tested in TestMiscCompileLinuxGOARM5.
func TestTryBotsCompileAllPorts(t *testing.T) {
out, err := exec.Command(filepath.Join(runtime.GOROOT(), "bin", "go"), "tool", "dist", "list").Output()
if err != nil {
t.Errorf("dist list: %v", err)
}
ports := strings.Fields(string(out))
// knownMissing tracks Go ports that that are known to be
// completely missing TryBot (pre-submit) test coverage.
//
// All completed ports should have either a real TryBot or at least a misc-compile TryBot,
// so this map is meant to be used to temporarily fix tests
// when the work of adding a new port is actively underway.
knownMissing := map[string]bool{}
var done = make(map[string]bool)
check := func(goos, goarch string) {
if goos == "android" || goos == "ios" {
// TODO(golang.org/issue/25963): support
// compilation-only Android and iOS trybots.
// buildall.bash doesn't set the environment
// up enough for e.g. compiling android-386
// from linux-amd64. (Issue #35596 too)
// iOS likely needs to be built on macOS
// with Xcode available.
return
}
goosArch := goos + "-" + goarch
if done[goosArch] {
return
}
for _, conf := range Builders {
if conf.GOOS() == goos && conf.GOARCH() == goarch &&
conf.BuildsRepoTryBot("go", "master", "master") {
// There's a real TryBot for this GOOS/GOARCH pair.
done[goosArch] = true
break
}
if strings.HasPrefix(conf.Name, "misc-compile-") {
re, err := regexp.Compile(conf.allScriptArgs[0])
if err != nil {
t.Fatalf("invalid misc-compile filtering pattern for builder %q: %q",
conf.Name, conf.allScriptArgs[0])
}
if re.MatchString(goosArch) {
// There's a misc-compile TryBot for this GOOS/GOARCH pair.
done[goosArch] = true
break
}
}
}
if knownMissing[goosArch] && done[goosArch] {
// Make it visible when a builder is added but the old
// knownMissing entry isn't removed by failing the test.
t.Errorf("knownMissing[%q] is true, but a corresponding TryBot (real or misc-compile) exists", goosArch)
} else if _, ok := done[goosArch]; !ok && !knownMissing[goosArch] {
t.Errorf("missing real TryBot or misc-compile TryBot for %q", goosArch)
}
}
for _, port := range ports {
slash := strings.IndexByte(port, '/')
if slash == -1 {
t.Fatalf("unexpected port %q", port)
}
check(port[:slash], port[slash+1:])
}
}
dashboard: include at most 3 ports per misc-compile TryBot Between Go 1.15, 1.16 and tip, the following 29 ports don't have a real TryBot and instead rely on misc-compile TryBots for their pre-submit coverage: • aix/ppc64 • darwin/amd64 • darwin/arm64 • dragonfly/amd64 • freebsd/386 • freebsd/arm • freebsd/arm64 • illumos/amd64 • linux/mips • linux/mips64 • linux/mipsle • linux/mips64le • linux/ppc64 • linux/ppc64le • linux/riscv64 • linux/s390x • netbsd/386 • netbsd/amd64 • netbsd/arm • netbsd/arm64 • openbsd/386 • openbsd/arm • openbsd/arm64 • openbsd/mips64 • plan9/386 • plan9/amd64 • plan9/arm • solaris/amd64 • windows/arm The previous approach for misc-compile target selection was to break them up primarily by GOOS value. However, as new architectures were added over time, some misc-compile TryBots got to a point where they were testing upwards of 5 ports (for example, misc-compile-openbsd was testing 386, amd64, arm, arm64, and mips64 architectures). Since each port is tested sequentially, allocating too many to one misc-compile TryBot can cause it to become the bottleneck of an entire TryBot run, exceeding the 10 minute completion time goal. Arrange it so misc-compile TryBot target selection is done explicitly in x/build, and pick 3 as max number of targets per TryBot for now. Based on recent timing observations, that should strike a decent balance between resource use (spinning up a builder) vs chance of a misc-compile TryBot becoming a bottleneck. It will also give us an opportunity to compare timing of 1, 2 and 3 targets per misc-compile in the future. (When we start tracking timing for TryBot completion time holistically, we'll be in a better position to refine this strategy further.) Making misc-compile target selection explicit in x/build also enables removing unnecessary duplicate misc-compile coverage from ports that already have a real TryBot (for example, openbsd/amd64 was previously tested via both the openbsd-amd64-68 TryBot and misc-compile-openbsd). This shouldn't be needed, so it's no longer done. For golang/go#17104. Fixes golang/go#32632. Change-Id: Iac918377b91af3e48780b38ffdf3153e213eeba2 Reviewed-on: https://go-review.googlesource.com/c/build/+/313210 Trust: Dmitri Shuralyov <dmitshur@golang.org> Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Heschi Kreinick <heschi@google.com> Reviewed-by: Carlos Amedee <carlos@golang.org>
2021-04-23 23:56:37 +03:00
// The 'linux-arm-arm5' pseduo-port is supported by src/buildall.bash
// and tests linux/arm with GOARM=5 set. Since it's not a normal port,
// the TestTryBotsCompileAllPorts wouldn't report if the misc-compile
// TryBot that covers is is accidentally removed. Check it explicitly.
func TestMiscCompileLinuxGOARM5(t *testing.T) {
var ok bool
for _, b := range Builders {
if !strings.HasPrefix(b.Name, "misc-compile-") {
continue
}
re, err := regexp.Compile(b.allScriptArgs[0])
if err != nil {
t.Fatalf("invalid misc-compile filtering pattern for builder %q: %q",
b.Name, b.allScriptArgs[0])
}
if re.MatchString("linux-arm-arm5") {
ok = true
break
}
}
if !ok {
// We get here if the linux-arm-arm5 port is no longer checked by
// a misc-compile TryBot. Report it as a failure in case the coverage
// was removed accidentally (e.g., as part of a refactor).
t.Errorf("no misc-compile TryBot coverage for the special 'linux-arm-arm5' pseudo-port")
}
}
// TestExpectedMacstadiumVMCount ensures that only 20 instances of macOS virtual machines
// are expected at MacStadium.
// TODO: remove once the scheduler allocates VMs based on demand https://golang.org/issue/35698
func TestExpectedMacstadiumVMCount(t *testing.T) {
got := 0
for host, config := range Hosts {
if strings.HasPrefix(host, "host-darwin-10_") || strings.HasPrefix(host, "host-darwin-amd64-") {
got += config.ExpectNum
}
}
if got != 16 {
t.Fatalf("macstadium host count: got %d; want 16", got)
}
}
// Test that we have a longtest builder and
// that its environment configuration is okay.
func TestLongTestBuilder(t *testing.T) {
long, ok := Builders["linux-amd64-longtest"]
if !ok {
t.Fatal("we don't have a linux-amd64-longtest builder anymore, is that intentional?")
}
if !long.IsLongTest() {
t.Error("the linux-amd64-longtest builder isn't a longtest builder, is that intentional?")
}
var shortDisabled bool
for _, e := range long.Env() {
if e == "GO_TEST_SHORT=0" {
shortDisabled = true
}
}
if !shortDisabled {
t.Error("the linux-amd64-longtest builder doesn't set GO_TEST_SHORT=0, is that intentional?")
}
}
func TestHostConfigIsVM(t *testing.T) {
testCases := []struct {
desc string
config *HostConfig
want bool
}{
{
desc: "non-ec2-vm",
config: &HostConfig{
VMImage: "image-x",
ContainerImage: "",
isEC2: false,
},
want: true,
},
{
desc: "non-ec2-container",
config: &HostConfig{
VMImage: "",
ContainerImage: "container-image-x",
isEC2: false,
},
want: false,
},
{
desc: "ec2-container",
config: &HostConfig{
VMImage: "image-x",
ContainerImage: "container-image-x",
isEC2: true,
},
want: false,
},
{
desc: "ec2-vm",
config: &HostConfig{
VMImage: "image-x",
ContainerImage: "",
isEC2: true,
},
want: true,
},
}
for _, tc := range testCases {
t.Run(fmt.Sprintf(tc.desc), func(t *testing.T) {
if got := tc.config.IsVM(); got != tc.want {
t.Errorf("HostConfig.IsVM() = %t; want %t", got, tc.want)
}
})
}
}
func TestModulesEnv(t *testing.T) {
testCases := []struct {
desc string
buildConfig *BuildConfig
repo string
want []string
}{
{
desc: "ec2-builder-repo-non-go",
buildConfig: &BuildConfig{
testHostConf: &HostConfig{
IsReverse: false,
isEC2: true,
},
},
repo: "bar",
want: []string{"GOPROXY=https://proxy.golang.org"},
},
{
desc: "reverse-builder-repo-non-go",
buildConfig: &BuildConfig{
testHostConf: &HostConfig{
IsReverse: true,
isEC2: false,
},
},
repo: "bar",
want: []string{"GOPROXY=https://proxy.golang.org"},
},
{
desc: "reverse-builder-repo-go",
buildConfig: &BuildConfig{
testHostConf: &HostConfig{
IsReverse: true,
isEC2: false,
},
},
repo: "go",
want: []string{"GOPROXY=off"},
},
{
desc: "builder-repo-go",
buildConfig: &BuildConfig{
testHostConf: &HostConfig{
IsReverse: false,
isEC2: false,
},
},
repo: "go",
want: []string{"GOPROXY=off"},
},
{
desc: "builder-repo-go-outbound-network-allowed",
buildConfig: &BuildConfig{
Name: "test-longtest",
testHostConf: &HostConfig{
IsReverse: false,
isEC2: false,
},
},
repo: "go",
want: nil,
},
{
desc: "builder-repo-special-case",
buildConfig: &BuildConfig{
testHostConf: &HostConfig{
IsReverse: false,
isEC2: false,
},
},
repo: "build",
want: []string{"GO111MODULE=on"},
},
{
desc: "reverse-builder-repo-special-case",
buildConfig: &BuildConfig{
testHostConf: &HostConfig{
IsReverse: true,
isEC2: false,
},
},
repo: "build",
want: []string{"GOPROXY=https://proxy.golang.org", "GO111MODULE=on"},
},
{
desc: "builder-repo-non-special-case",
buildConfig: &BuildConfig{
testHostConf: &HostConfig{
IsReverse: false,
isEC2: false,
},
},
repo: "bar",
want: nil,
},
}
for _, tc := range testCases {
t.Run(tc.desc, func(t *testing.T) {
got := tc.buildConfig.ModulesEnv(tc.repo)
if diff := cmp.Diff(tc.want, got); diff != "" {
t.Errorf("BuildConfig.ModulesEnv(%q) mismatch (-want, +got)\n%s", tc.repo, diff)
}
})
}
}