Re: [quicwg/base-drafts] Required state for retaining unacked RETIRE_CONNECTION_ID frames is unbound (#3509)

Kazuho Oku <> Fri, 27 March 2020 02:36 UTC

Return-Path: <>
Received: from localhost (localhost []) by (Postfix) with ESMTP id 49CC63A08C9 for <>; Thu, 26 Mar 2020 19:36:01 -0700 (PDT)
X-Virus-Scanned: amavisd-new at
X-Spam-Flag: NO
X-Spam-Score: -3.098
X-Spam-Status: No, score=-3.098 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIMWL_WL_HIGH=-0.001, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, DKIM_VALID_EF=-0.1, HTML_MESSAGE=0.001, MAILING_LIST_MULTI=-1, RCVD_IN_DNSWL_BLOCKED=0.001, SPF_HELO_NONE=0.001, SPF_PASS=-0.001, URIBL_BLOCKED=0.001] autolearn=ham autolearn_force=no
Authentication-Results: (amavisd-new); dkim=pass (1024-bit key)
Received: from ([]) by localhost ( []) (amavisd-new, port 10024) with ESMTP id MrcituPZYUyh for <>; Thu, 26 Mar 2020 19:35:59 -0700 (PDT)
Received: from ( []) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by (Postfix) with ESMTPS id 94AFD3A08C1 for <>; Thu, 26 Mar 2020 19:35:59 -0700 (PDT)
Received: from ( []) by (Postfix) with ESMTP id 69E2E1C040A for <>; Thu, 26 Mar 2020 19:35:58 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed;; s=pf2014; t=1585276558; bh=0HSKfWC4ycNO4kWeEUlVCeo9i4FzHNwGlUzi3hGe/H8=; h=Date:From:Reply-To:To:Cc:In-Reply-To:References:Subject:List-ID: List-Archive:List-Post:List-Unsubscribe:From; b=eaHt0YknGbBb49YhomeEBwAGx/GBFst/JsK66JzXJbgSLZoD4oqmzdK7uKaMkQw3Y DqdWMzo0bFjz9oX5/KzFLQ0LFvo3b/Tvi4adffNqNeANjRaM64kxZM85JRzC0hne1N cqvs59svdOVq4O/xMfPEwGOdAmvLlBa6dovm06qU=
Date: Thu, 26 Mar 2020 19:35:58 -0700
From: Kazuho Oku <>
Reply-To: quicwg/base-drafts <>
To: quicwg/base-drafts <>
Cc: Subscribed <>
Message-ID: <quicwg/base-drafts/issues/3509/>
In-Reply-To: <quicwg/base-drafts/issues/>
References: <quicwg/base-drafts/issues/>
Subject: Re: [quicwg/base-drafts] Required state for retaining unacked RETIRE_CONNECTION_ID frames is unbound (#3509)
Mime-Version: 1.0
Content-Type: multipart/alternative; boundary="--==_mimepart_5e7d668e5ac14_2ec53fa2deccd95c1368f"; charset="UTF-8"
Content-Transfer-Encoding: 7bit
Precedence: list
X-GitHub-Sender: kazuho
X-GitHub-Recipient: quic-issues
X-GitHub-Reason: subscribed
X-Auto-Response-Suppress: All
Archived-At: <>
X-Mailman-Version: 2.1.29
List-Id: Notification list for GitHub issues related to the QUIC WG <>
List-Unsubscribe: <>, <>
List-Archive: <>
List-Post: <>
List-Help: <>
List-Subscribe: <>, <>
X-List-Received-Date: Fri, 27 Mar 2020 02:36:02 -0000

> If the peer doesn't acknowledge your retirement, then you shouldn't be issuing more CIDs, because you might exceed their limit.

I think the framing is incorrect. The peer can enforce an endpoint to retire a CID (e.g., by changing RPT or discarding a path). As @martinduke points out, an endpoint is encouraged issue a new CID at this point, rather than withholding than until the ACK of RCID is received.

Note also that RCID carries the CID that *the endpoint sends* carries the CID that the peer has issued, while NCID that the endpoint carries the CID that that endpoint issues. Therefore, I am not sure what the rationale for withholding the emission of NCID until receiving an ACK for RCID would be.

> In my opinion, running out of CIDs is not a big crisis. You can always create a new connection with 0-RTT with a relatively small overhead. We should be biased towards simplicity rather than never running out of CIDs.

I agree. An endpoint is not expected to consume CID at high rate. We already discourage such behavior; see

That's why I'm fine with the leaky behavior that @martinthomson has suggested.

> I'm really coming around to your cumulative retire idea, but will have another go at this tomorrow.

As previously stated, I am hesitant to have a new machinery that is unique to this purpose. There is a risk of creating something buggy, both in terms of specification and implementation. My preference goes to accepting the risk (and recommending mitigations, such as the leaky behavior), or reusing a design pattern that we already have (i.e. introduce MAX_CONNECTION_IDS frame).

You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub: