1
0
forked from wrenn/wrenn

fix: prevent sandbox halt after resume by fixing HTTP/2 HOL blocking and adding timeouts

Disable HTTP/2 on both host agent server and CP→agent transport — multiplexing
caused head-of-line blocking when a slow sandbox RPC stalled the shared connection.
Add ResponseHeaderTimeout to envd HTTP clients. Merge SetDefaults into Resume's
PostInit call to eliminate an extra round-trip that could hang on a stale connection.
This commit is contained in:
2026-05-02 13:48:51 +06:00
parent 7ef9a64613
commit bb582deefa
5 changed files with 30 additions and 16 deletions

View File

@ -154,6 +154,11 @@ func main() {
Addr: listenAddr,
ReadHeaderTimeout: 10 * time.Second,
IdleTimeout: 620 * time.Second, // > typical LB upstream timeout (600s)
// Disable HTTP/2: empty non-nil map prevents Go from registering
// the h2 ALPN token. Connect RPC works over HTTP/1.1; HTTP/2
// multiplexing causes HOL blocking when a slow sandbox RPC stalls
// the shared connection.
TLSNextProto: make(map[string]func(*http.Server, *tls.Conn, http.Handler)),
}
// mTLS is mandatory — refuse to start without a valid certificate.