src.nth.io/

summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorLuke Hoersten <[email protected]>2026-06-19 19:44:08 -0500
committerLuke Hoersten <[email protected]>2026-06-19 19:44:08 -0500
commite88aa51627a727d1883d1d803c9674d92dac643c (patch)
treec947f74fcf06818f22d0927c7f1c732ec3b9f566
parent6df19c53a37eadee6c132fc6c1a7a2f9bd53e505 (diff)
scrypted: noDelay:true on keep-alive sockets — kills 200ms Nagle deadlock
After enabling keep-alive in bccf0fa, fw_recv p95 spiked from ~25ms to ~230ms on a significant fraction of frames and net_up p95 went from <15ms to 200+ms simultaneously. Worst frames hit 467ms wall. Root cause: Node http.Agent's outgoing sockets default to noDelay false (Nagle ON). When a ~128KB JPEG body doesn't end on an exact MTU boundary, the sender holds the final partial packet for up to 200ms waiting for either more data or a peer ACK. Firmware-side TCP_NODELAY only controls what the firmware *sends* (its ACKs and response packets); it has no effect on what arrives at the firmware from a Nagling sender. Setting noDelay: true on the http.Agent makes new sockets in the pool opt out of Nagle on creation. Plus a belt-and-suspenders setNoDelay on the request's socket event in case the Agent-level option isn't honored on this Node version. The 230ms signature is the classic Nagle + delayed-ACK deadlock: - Sender: "I have a partial packet; I'll wait for more data or an ACK before sending." - Receiver: "I'll delay this ACK up to 200ms in case I can piggyback it on a response." - Result: 200ms stall on every send that doesn't fill an MTU.
-rw-r--r--scrypted/scrypted-viewport.ts18
1 files changed, 17 insertions, 1 deletions
diff --git a/scrypted/scrypted-viewport.ts b/scrypted/scrypted-viewport.ts
index df6c041..344d58c 100644
--- a/scrypted/scrypted-viewport.ts
+++ b/scrypted/scrypted-viewport.ts
@@ -6,7 +6,7 @@
// short git hash of the commit that added this constant — if the
// hash in the log doesn't match the HEAD this file came from, the
// Scrypted Script editor is still on stale code.
-const SCRIPT_VERSION = "1063a4e";
+const SCRIPT_VERSION = "pending";
//
// Architecture
// ------------
@@ -338,6 +338,16 @@ class ScryptedViewportProvider extends ScryptedDeviceBase
maxSockets: 2,
maxFreeSockets: 2,
timeout: 30_000,
+ // CRITICAL: without this, the keep-alive socket inherits
+ // Nagle ON by default. When a 128KB JPEG body doesn't
+ // end on an MTU boundary, the kernel sits on the final
+ // partial packet for up to 200ms (delayed-ACK window)
+ // waiting for either more data or an ACK. Firmware-side
+ // TCP_NODELAY only affects firmware's sends; the sender
+ // also has to opt out. Without noDelay we measured
+ // fw_recv p95 spiking from ~25ms to ~230ms — the exact
+ // 200ms Nagle+delayed-ACK deadlock signature.
+ noDelay: true,
});
this.agents.set(host, a);
}
@@ -387,6 +397,12 @@ class ScryptedViewportProvider extends ScryptedDeviceBase
}
req.on("timeout", () => req.destroy(new Error("request timeout")));
req.on("error", reject);
+ // Belt-and-suspenders: in case the Agent's noDelay option
+ // isn't honored on every Node version, force it on the
+ // socket as soon as it's allocated. Without this the
+ // socket inherits Nagle ON and a JPEG body ending mid-MTU
+ // stalls for the 200ms delayed-ACK window.
+ req.on("socket", (s: any) => { try { s.setNoDelay(true); } catch {} });
if (opts.body != null) req.write(opts.body);
req.end();
});