zap-proto/zip
1
0
Fork 0
mirror of https://github.com/zap-proto/zip.git synced 2026-10-19 15:38:57 +00:00
Code Issues Packages Projects Releases Wiki Activity

A plugin's dying reply is a race, and not the one under test

Browse Source 
The crash trigger asserted 200, which says the panicking plugin flushed its
answer before the process went. Whether it wins that race is a scheduling
accident; what the test is for is the sentence after it — the host survives and
the plugin comes back.

Co-authored-by: Hanzo Dev <[email protected]>
This commit is contained in:
zeekay
2026-09-17 01:05:04 -07:00
co-authored by Hanzo Dev
parent e24128f57b
commit a69381ca22
1 changed files with 7 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
load_test.go
+7 -2
View File
@@ -481,8 +481,13 @@ func TestPlugins_SurvivesPanic(t *testing.T) {
}
// Panic it on a goroutine — unrecoverable, so the process really dies.
if status, _ := call(t, app, "GET", "/v1/demo/crash", ""); status != 200 {
t.Fatalf("crash trigger: status %d", status)
//
// Whether it flushes its answer before it dies is a RACE, and not the one
// under test: 200 means the reply won, 502 means the socket went first. The
// claim is what follows — the host survives and the plugin comes back — so
// asserting either side of that race is asserting a scheduling accident.
if status, _ := call(t, app, "GET", "/v1/demo/crash", ""); status != 200 && status != 502 {
t.Fatalf("crash trigger: status %d, want 200 or a dead socket", status)
}
// THE HOST MUST SURVIVE. If a plugin panic could kill the host, none of