selfhosted
/
coder
mirror of https://github.com/coder/coder.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

fix: stop running tests that exec sh scripts in parallel (#11834) 

Ok, so my last attempt at a fix here failed

https://github.com/coder/coder/actions/runs/7666229961/job/20893608286

I have a new theory: it's not the `terraform` binary that's busy, it's actually `fake_cancel.sh` and it gets marked busy when we `exec` it from the script we write.

Use of `exec` also replaces the executing code in place, rather than starting a new process/shell, so that's why the error we get says `terraform` is busy.
This commit is contained in:
Spike Curtis 2024-01-26 19:22:35 +04:00 committed by GitHub
parent 0ba035a16d
commit d6baa3cab0
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 10 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

25
provisioner/terraform/provision_test.go
View File

@ -14,7 +14,6 @@ import (
"runtime"
"sort"
"strings"
"syscall"
"testing"
"time"
@ -126,8 +125,10 @@ func sendApply(sess proto.DRPCProvisioner_SessionClient, transition proto.Worksp
}}})
}
// below we exec fake_cancel.sh, which causes the kernel to execute it, and if more than
// one process tries to do this simultaneously, it can cause "text file busy"
// nolint: paralleltest
func TestProvision_Cancel(t *testing.T) {
t.Parallel()
if runtime.GOOS == "windows" {
t.Skip("This test uses interrupts and is not supported on Windows")
}
@ -158,24 +159,16 @@ func TestProvision_Cancel(t *testing.T) {
}
for _, tt := range tests {
tt := tt
// below we exec fake_cancel.sh, which causes the kernel to execute it, and if more than
// one process tries to do this, it can cause "text file busy"
// nolint: paralleltest
t.Run(tt.name, func(t *testing.T) {
t.Parallel()
dir := t.TempDir()
binPath := filepath.Join(dir, "terraform")
// Example: exec /path/to/terrafork_fake_cancel.sh 1.2.1 apply "$@"
content := fmt.Sprintf("#!/bin/sh\nexec %q %s %s \"$@\"\n", fakeBin, terraform.TerraformVersion.String(), tt.mode)
// golang's standard OS library can sometimes leave the file descriptor open even after
// "Closing" the file (which can then lead to a "text file busy" error, so we bypass this
// and use syscall directly).
fd, err := syscall.Open(binPath, syscall.O_WRONLY|syscall.O_CREAT, 0o755)
require.NoError(t, err)
n, err := syscall.Write(fd, []byte(content))
require.NoError(t, err)
require.Equal(t, len(content), n)
err = syscall.Close(fd)
err := os.WriteFile(binPath, []byte(content), 0o755) //#nosec
require.NoError(t, err)
t.Logf("wrote fake terraform script to %s", binPath)
@ -228,8 +221,10 @@ func TestProvision_Cancel(t *testing.T) {
}
}
// below we exec fake_cancel_hang.sh, which causes the kernel to execute it, and if more than
// one process tries to do this, it can cause "text file busy"
// nolint: paralleltest
func TestProvision_CancelTimeout(t *testing.T) {
t.Parallel()
if runtime.GOOS == "windows" {
t.Skip("This test uses interrupts and is not supported on Windows")
}