Return-Path: X-Original-To: netmod@ietfa.amsl.com Delivered-To: netmod@ietfa.amsl.com Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 1B6B9C1DFD4A; Mon, 13 May 2024 11:56:52 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -2.668 X-Spam-Level: X-Spam-Status: No, score=-2.668 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIMWL_WL_HIGH=-0.582, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, DKIM_VALID_EF=-0.1, HTML_MESSAGE=0.001, RCVD_IN_MSPIKE_H2=-0.001, RCVD_IN_ZEN_BLOCKED_OPENDNS=0.001, SPF_NONE=0.001, T_KAM_HTML_FONT_INVALID=0.01, URIBL_DBL_BLOCKED_OPENDNS=0.001, URIBL_ZEN_BLOCKED_OPENDNS=0.001] autolearn=ham autolearn_force=no Authentication-Results: ietfa.amsl.com (amavisd-new); dkim=pass (2048-bit key) header.d=nokia.com Received: from mail.ietf.org ([50.223.129.194]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id U6N318RhvgWm; Mon, 13 May 2024 11:56:47 -0700 (PDT) Received: from NAM10-BN7-obe.outbound.protection.outlook.com (mail-bn7nam10on2054.outbound.protection.outlook.com [40.107.92.54]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id 29DC6C180B5D; Mon, 13 May 2024 11:56:46 -0700 (PDT) ARC-Seal: i=1; a=rsa-sha256; s=arcselector9901; d=microsoft.com; cv=none; b=MpkD2FyZXifK/vre9IqsD41WB5OrDCaJMjqGAbLIqSIBCInjfKK9VtNVUaGaXLFVqQeKExxYVDj2RbH6oFNFIkawli4eOet5AaPprsrUlpTFmAqAXC91GF7s5r5ts4YraidkmVlXq+exKxXt0j01pCknh86rGIsgahEhHJFVyDQ8Qp4foL1QwFteJkgp6xFgW1/YXp+8pl6rsX/+qnPqNBeZyqHdystYzIApsI19U/3HU68aqSEqn3bgllTcjyYyF6aMF/8QPRuLEzMg2O17KCvps1ZRqFj+7LHApssDVW9MbMtPInsaFjqL9Q68Yxq5tlYvTyALsUtyyk+JcDo6jA== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=microsoft.com; s=arcselector9901; h=From:Date:Subject:Message-ID:Content-Type:MIME-Version:X-MS-Exchange-AntiSpam-MessageData-ChunkCount:X-MS-Exchange-AntiSpam-MessageData-0:X-MS-Exchange-AntiSpam-MessageData-1; bh=L8r+rbYyR7F4sDh95dUFei4C3fr8hAj7bellgLqztD8=; b=NN5Rws3aZY1ysptG8G4Xk34yIa5IKr97W6xX2LNLOQKZ0uUqa7lAheygxv4t0SvOhWi9cSGw/GmaM2MitR4Jw2JXR4gz1rZ3xm/bSkYBTORdZJLmtW0nLbrLMf3QfZvnMTTusn6YuSsjWwyzEVvsrreT1IHI9c8TsA4y0SbzNKLVofzbsCWZ8rmVoeoh4Bt5C0aKlkqvtJ1pydUG9HBicl2/gg6iOtsSWKWdlllfDJIpZSr0n3t3mrJK6c0scWDG+ZSW7W7bYhrARUuXTps/q5HzVz6UOZVTpkKefDAEj3rjp1aWSNrwEoWyuEfHnmKl8lbM7L21VOiQ6Pu4CJ4YVQ== ARC-Authentication-Results: i=1; mx.microsoft.com 1; spf=pass smtp.mailfrom=nokia.com; dmarc=pass action=none header.from=nokia.com; dkim=pass header.d=nokia.com; arc=none DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=nokia.com; s=selector2; h=From:Date:Subject:Message-ID:Content-Type:MIME-Version:X-MS-Exchange-SenderADCheck; bh=L8r+rbYyR7F4sDh95dUFei4C3fr8hAj7bellgLqztD8=; b=Q5zsOuRo1oLI5JLfJhldUdXgsNPIcJYVP9c8pKXA1VIfhsJqH/pvq510JYR5P9Y6DaPMqY0wuvv6yLXDuUgnZUAtnXQSbLv1X4JrD6acaYs9O6YRboMqYCn8zD6co7vz8ydjHFtoKyKPgkpi+gwBEUvLSF1XYVvDPlgmrBaDzBoUHbxtZ85Yodtkb2lCyxVlxPi/mTw2E+6L6+xUcyPsXIai4lzxtJzSKVFysNyWRZ/1MSmxJ2xranohJIJs9Qw4zt7TeZtqLB8WDE8O76FFuy4basRdQVpDsLwjhNeT8NObZuxLuW1Fsulkbo/NWdh5wirAuBB6s58WnuPlBus0YQ== Received: from SN6PR08MB4847.namprd08.prod.outlook.com (2603:10b6:805:6b::12) by SA1PR08MB8385.namprd08.prod.outlook.com (2603:10b6:806:339::9) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384) id 15.20.7544.55; Mon, 13 May 2024 18:56:41 +0000 Received: from SN6PR08MB4847.namprd08.prod.outlook.com ([fe80::9959:84e9:c6c3:7da]) by SN6PR08MB4847.namprd08.prod.outlook.com ([fe80::9959:84e9:c6c3:7da%7]) with mapi id 15.20.7544.052; Mon, 13 May 2024 18:56:41 +0000 From: "Jason Sterne (Nokia)" To: "maqiufang (A)" , "Rob Wilton (rwilton)" , "kent+ietf@watsen.net" , NETMOD Group , "draft-ietf-netmod-system-config@ietf.org" Thread-Topic: [netmod] Re: WGLC on system-config-05 Thread-Index: AQHaohhU7eR1shoyaEmnxWUhiibkwLGQpSXAgARcjQCAAIDxoA== Date: Mon, 13 May 2024 18:56:40 +0000 Message-ID: References: <18094edcb1b144e381e653820c2155af@huawei.com> In-Reply-To: <18094edcb1b144e381e653820c2155af@huawei.com> Accept-Language: en-US Content-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: authentication-results: dkim=none (message not signed) header.d=none;dmarc=none action=none header.from=nokia.com; x-ms-publictraffictype: Email x-ms-traffictypediagnostic: SN6PR08MB4847:EE_|SA1PR08MB8385:EE_ x-ms-office365-filtering-correlation-id: dbf61388-5389-42be-2c16-08dc737e6cc7 x-ms-exchange-senderadcheck: 1 x-ms-exchange-antispam-relay: 0 x-microsoft-antispam: BCL:0;ARA:13230031|376005|366007|1800799015|38070700009; x-microsoft-antispam-message-info: =?us-ascii?Q?/ryL9IRHpIVM1O+3hHN2rGKpkex/6fKwfODJLptdl6v0wobI/7C5OrrDRNDN?= =?us-ascii?Q?Y/idDmm78yWGs8tzkTSx/NtZ2XfcrylbSSxEG19ERQ7beDhnpkegJ7hHkM1C?= =?us-ascii?Q?5l7ZSJhTvORfKgmyrlcV6e1Zc+1ljkUGk45+5NOWg8ANSePWjmYCX89fyllS?= =?us-ascii?Q?KcKqVI6yZjn3jLd1Upv7jvwVf8hfsa06l0/xm/GsRybZt3zgafiKJGxn8JvU?= =?us-ascii?Q?31/ccgplquHh98gCCRE97qLJM9Z+pSMkXax1th7gczfjrJzDBIywWGVO1fmE?= =?us-ascii?Q?Wk5XVwxRaDJaqAUIDE8AiBMdzvIIlF5WPDbY3ARjBwYmm2QosCqathzksGR0?= =?us-ascii?Q?7fXlfh1J1pFhnYTeuzecwxKAUnvZQhTGwfGSM/KhPEm5FabjhjZXkcugysu+?= =?us-ascii?Q?TW8eB4LX/tR7KKPFve741d9uz1pvkLpkDOa1r1MsF5HO5YsuPIwSt51LTZCb?= =?us-ascii?Q?U6PY+6LZ9psKnHyEgJnq4w96gzk68PUhOBCkiIPwc1pBFx7SW5jblBq02JIU?= =?us-ascii?Q?ezA1wxsGt3GGDqZMuyu4u+f/Fz/RC2FfnPTXyXmfXNf2RLrwEIV1g3NNpD7v?= =?us-ascii?Q?CT1BCO7cMzHKoa6n73H2d9hUZSizxtBSf4gTMyAd9QIhIH+tGCRr7ZSk3ohM?= =?us-ascii?Q?KgiDzP/ExfjkykbLEE/AGHawZejklMA4Yu1G1NznHN27Ql/yYIzzDPb+fyLG?= =?us-ascii?Q?n+JD9G3Ue0Jq+uX5SjSGrEhJZRJYTFtoEX9YR/fQjAkQwfb/LhrLjeYDDzVX?= =?us-ascii?Q?t3oyWv8EFE+07Q/9yQSO4Du41ZYX6cev7lKD4+XmlgsyuDdINGBQoc/Foqqo?= =?us-ascii?Q?NAinuoDgBdUQmO/Jvzg3hPmOXFXIbPGRBeJ0ziOjSaFah3lWcQUScbvh7KwT?= =?us-ascii?Q?FOlpimUwuajrtD/ceOcP+dNMUXs7Sb673MHa6bDBBOyJcN8StZwozI3GLNre?= =?us-ascii?Q?McHsZ9KLynhc1ZfAo88oOBb65r9s6lbRpY/0l8/RvOej6K9REQEEADmIHvUD?= =?us-ascii?Q?i9dP99iMf9Td7+lnL06MXK/MNqcF83IfY8LzZHpasUwbwOj0N3she2dmq/Bl?= =?us-ascii?Q?rpK3bdHzdUkHFZTp1uiHseJvv3KRCi0lrKhTt7yysyp1gIu0vCVR+7cjPbtV?= =?us-ascii?Q?PQ2UzrNWFCLGOyfBQZf9Rg7ZJrzMBzMapGWnf9y6XeJqCQ1VDFvrQa0eX3ha?= =?us-ascii?Q?D0Ia09g+22pJNpRST1Znq5IG3dMz5v1YXi1KBTof+bs4iYdPjej2gT6+ldO4?= =?us-ascii?Q?G57TrG6KvvxfxjafBcu/h4tOAGsh0xNp7v5f4NaqB0SsVbv4MGIwp/Pi//X9?= =?us-ascii?Q?11Q=3D?= x-forefront-antispam-report: CIP:255.255.255.255;CTRY:;LANG:en;SCL:1;SRV:;IPV:NLI;SFV:NSPM;H:SN6PR08MB4847.namprd08.prod.outlook.com;PTR:;CAT:NONE;SFS:(13230031)(376005)(366007)(1800799015)(38070700009);DIR:OUT;SFP:1101; x-ms-exchange-antispam-messagedata-chunkcount: 1 x-ms-exchange-antispam-messagedata-0: =?us-ascii?Q?+veqpxaU6q+e3n44fHH1lah37cNUq1mfmcaUcrAcUxUydt6F3YWfi9W1mxks?= =?us-ascii?Q?icHD7Wd/PQz+iH9Au1Yh5pj9/mH4KPFDA57ohx7RDVnxidOJ1Nr531vvfXg7?= =?us-ascii?Q?LPVSD2twtWaRHEzrEsTXRRhuk8yCH7PxM7Jy9UQgoMpj7DG4lsPf6KvTbjzQ?= =?us-ascii?Q?UgH1bWruarMNkIF+xe+ZlxPhIwsupJUEP7XD9XtwZ7rDgCFBYZlgJl7ShtqF?= =?us-ascii?Q?w88rx5tv8GBwhjffYhO1pH8cu4Clj46nQn0CbVT3l7HUEuGbipsY0vAwl9IB?= =?us-ascii?Q?MUPB36je3CJzk7xjl3QHYjJ1Wxs5jHYSwDFbFFbxN4oWoA1eBXkyDxR7qUzv?= =?us-ascii?Q?6fZHeJfjzwCfs8qjD351E6aQtV5ReDSw6fMGrXmGehuIXZ7zdpyowJ48vlDC?= =?us-ascii?Q?xOgR75LAsjzu4ttBwJtvWN57s7DvGspsgFmDXkGlu+fOrJQ9GL17PF/gJJv/?= =?us-ascii?Q?flBBeN9K6C53/h2u9cd+F7Dq63Ea7p802YZ3L7GZVuVyXMJLfa3xoFbZT7bm?= =?us-ascii?Q?an5gqUBUeqoLJE64QSUHBv+LA8yyFyUe56HxYgwaKfOH2I6wOrV7u1M1f7jq?= =?us-ascii?Q?Zm99/WtEHgr+LWoA7dLG8f1abcJNF5PDRw/Ip/RuxdqXLY7Q/BB9B32WrnsQ?= =?us-ascii?Q?PEu4xJ/t8KxzSoD4LqhDy+eQRWPXNbc3NTqLCM/0GeS66e8gqfPjx8dicjyM?= =?us-ascii?Q?a1bmpKWYd1hsIA2zTYV953bju0xMGydw1r39UXOsfBsodDpGEKjk1rMcJdvV?= =?us-ascii?Q?RwZk0ZHD9xAJjr4BWe6uDyEhRdJiRbIwnjBw5WtEUOSci7XTXgoDAvcKBAZI?= =?us-ascii?Q?39D5SM/gPeJbmCybnbkCD4JfrizuaPtE5kBrY7LXVje1OxLj0w8NNhiGnPpr?= =?us-ascii?Q?mlii3RUCnkK3WU1ug8XVq9LXmTb7tj9uyHfkeG/bsYVpOOsquXZSrF2pjiF4?= =?us-ascii?Q?06/6oeRoZ5ZJgxHoEOSH7ewMWlsxqsTJQaB7bdl2D285aAcWDJFuFLiGOiTG?= =?us-ascii?Q?2llga6TYCcxGqHa3Rz1OidGCm+zIJ1mYZ4BNDY8DtzRCnvkhuu7sg9pGZxZ6?= =?us-ascii?Q?tcSxU/nfzhJlXc9B/zcQM+hOQ9L7YRlMXnRvHpRoLi2nsPn/PlVDkVnLtO7d?= =?us-ascii?Q?/wnwq8079i6dw1VDs9mvHYKhV/XW8Pq76gYgqfR2js8Oig1SSezi08QqdRo2?= =?us-ascii?Q?HqB4FG1MdbDiBp6spaMrH9MOmrQBV8UevakVLUJ0X3eufzQl9hA3UtXdeMKs?= =?us-ascii?Q?G+Dte+2HxPeutGDH37RZ1x0mzJQkOzn/WiKwxfP/4TQDPRbh4+OGG/8TyoAq?= =?us-ascii?Q?TSwpYYTKM+gkZ9WNhoML5UdUWw6GDxHeH8BdA+lW9mV0AlRGUVyny95UWXDR?= =?us-ascii?Q?AEIhoWR2H2YAlYnLyWNUWRBhdaYqg6NMWmOYGiIUL3/P33Ywfwgs9aVjQ8MB?= =?us-ascii?Q?CLpGH71o7Ste0OLqz3OOgHc8oPvRR4M2MdblmlMCH01YaebbolXvouKbbd64?= =?us-ascii?Q?Pxp1/tEp6Ja+VcZlksYgUFCDE1zr6qQb3FGiMY/RygNLL0jx70YIAoxVFWSK?= =?us-ascii?Q?4b0R5ynzEkIUqG8vMzr/eiKtXImQ18cbNg1mCRDH?= Content-Type: multipart/alternative; boundary="_000_SN6PR08MB4847F2005D9219409B6CDCC29BE22SN6PR08MB4847namp_" MIME-Version: 1.0 X-OriginatorOrg: nokia.com X-MS-Exchange-CrossTenant-AuthAs: Internal X-MS-Exchange-CrossTenant-AuthSource: SN6PR08MB4847.namprd08.prod.outlook.com X-MS-Exchange-CrossTenant-Network-Message-Id: dbf61388-5389-42be-2c16-08dc737e6cc7 X-MS-Exchange-CrossTenant-originalarrivaltime: 13 May 2024 18:56:40.7737 (UTC) X-MS-Exchange-CrossTenant-fromentityheader: Hosted X-MS-Exchange-CrossTenant-id: 5d471751-9675-428d-917b-70f44f9630b0 X-MS-Exchange-CrossTenant-mailboxtype: HOSTED X-MS-Exchange-CrossTenant-userprincipalname: 30rY0qhRoInx3el1HKVBPSIdYjf1l3JC0shpv3ZyZBZdg/KifnqIn3pStauEntoakI60GLNVAVl5NVXtVWWd0w== X-MS-Exchange-Transport-CrossTenantHeadersStamped: SA1PR08MB8385 Message-ID-Hash: QQ5MLREEDYDVG2GSXJ24CU4OFCBUE44I X-Message-ID-Hash: QQ5MLREEDYDVG2GSXJ24CU4OFCBUE44I X-MailFrom: jason.sterne@nokia.com X-Mailman-Rule-Misses: dmarc-mitigation; no-senders; approved; emergency; loop; banned-address; member-moderation; header-match-netmod.ietf.org-0; nonmember-moderation; administrivia; implicit-dest; max-recipients; max-size; news-moderation; no-subject; digests; suspicious-header X-Mailman-Version: 3.3.9rc4 Precedence: list Subject: =?utf-8?q?=5Bnetmod=5D_Re=3A_WGLC_on_system-config-05?= List-Id: NETMOD WG list Archived-At: List-Archive: List-Help: List-Owner: List-Post: List-Subscribe: List-Unsubscribe: --_000_SN6PR08MB4847F2005D9219409B6CDCC29BE22SN6PR08MB4847namp_ Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Thx Qiufang. Please see inline. Jason From: maqiufang (A) Sent: Monday, May 13, 2024 3:47 AM To: Jason Sterne (Nokia) ; Rob Wilton (rwilton) ; kent+ietf@watsen.net; NETMOD Group ; dra= ft-ietf-netmod-system-config@ietf.org Subject: RE: [netmod] Re: WGLC on system-config-05 Hi, Jason and Rob, Thanks you both for your valuable comments, all good points, much appreciat= ed. Please also find my reply below inline... -----Original Message----- From: Jason Sterne (Nokia) [mailto:jason.sterne@nokia.com] Sent: Saturday, May 11, 2024 12:29 AM To: Rob Wilton (rwilton) >; Kent Watsen >; netmod@ietf.org; draft-iet= f-netmod-system-config@ietf.org Subject: RE: [netmod] Re: WGLC on system-config-05 Please see inline for comments on the "moderate level comments". I'll try = to reply later with more feedback on further items below. Jason From: Rob Wilton (rwilton) > Sent: Thursday, May 9, 2024 9:55 AM To: Kent Watsen >; netmod= @ietf.org; draft-ietf-netmod-system-config@ietf.org= Subject: [netmod] Re: WGLC on system-config-05 CAUTION: This is an external email. Please be very careful when clicking li= nks or opening attachments. See the URL nok.it/ext for additional informati= on. [Resending due to mailer issues.] [Qiufang] Thanks, I see it's been accurately archived now. Hi authors, chairs, WG, I'm generally supportive of this work, but I think that there are still som= e potential corner cases that are not covered, or it isn't entirely obvious= how they are handled. Comments below. Moderate level comments: (1) p 7, sec 2.3. Inactive-Until-Referenced There are some system configuration predefined (e.g., application ids, anti-x signatures, trust anchor certs, etc.) as a convenience for the clients, which must be referenced to be active. The clients can also define their own configurations for their unique requirements. Inactive-until-referenced system configuration is generated in immediately when the device is powered on, but it is not active until being referenced. I'm not sure whether Inactive-Until-Referenced actually needs to be defined= , or to put it another way, I'm not sure whether this type of configuration= is special to system datastores at all. If a configuration (either explic= itly in or implicitly from ) defines a QoS policy that is= not referenced from anywhere, (e.g., not applied to any interfaces) then I= think that it up to the server to decide whether that unreferenced QoS pol= icy is reported in operational or not, depending on server implementation. [>>JTS:] I agree. I think section 2 may be mixing up two concepts: 1. Data dynamically being populated/removed from the system datastore, a= nd 2. For data that is in the system datastore, whether it is "active" and = present in the operational datastore [Qiufang] Yes, it's actually from 2 dimensions to differentiate different k= inds of system config: 1. Time of being generated; 2. Time of being applied= . For #2 I don't think there should be anything special. We could say somethi= ng like: As with the running datastore, data present in the system datastor= e may or may not be present in the operational datastore depending on wheth= er it is considered as active configuration or not by the server. [Qiufang] I am personally okay to remove the third kind of definition to ke= ep the definition dimension consistent, and maybe for system configuration = being generated at both different times (immediately vs. conditional), they= may be either applied by the server immediately or only after being refere= nced. I also think it's worth adding some text as Jason suggested, the clie= nt would benefit from the knowledge that there might be some system configu= ration that is defined there but not actually in use. [>>JTS:] I think it is more than just removing the 3rd type defined in sect= ion 2. We probably need to rework to just have two types (for the one dimen= sion): 1. Immediately-present 2. Conditionally-present They would both related purely to presence of config in (and not s= ay anything about whether they are applied or active). For #1 there is a bit of repeat with section 4.2. We should probably just t= alk about system config coming and going (and changing) in one place. [Qiufang] I think you are referring to the QoS examples in the first paragr= aph of sec.4.2, right? We will move this to sec.2 in the new revision. [>>JTS:] I wasn't just referring to the QoS example. It was the entire conc= ept that the contents of the datastore can change. But maybe it is= OK if that concept is mentioned in the 2 places. Perhaps we should just re= fer back to section 2 from 4.2 when we mention things dynamically showing u= p in . I would not necessarily move your QoS example out of 4.2 - = it seems useful there (as well as useful in section 2). We'd have to update other parts of the doc (e.g. 5.1) where these 3 types o= f data are mentioned. [Qiufang]Yes, I agree. (2) p 9, sec 5.1. Conceptual Model of Datastores When the device is powered on, immediately-active system configuration is generated in and active immediately, but inactive-until-referenced system configuration only becomes active if referenced by client-defined configuration. However, conditionally- active system configuration will only be created and active when specific conditions on system resources are met. I think that it should be "merged with system" not "merged into system" sin= ce the running configuration never ends up in the system datastore. [>>JTS:] Yes. Jan mentioned the same. [Qiufang] This will be fixed in the new revision, thanks! (3) p 9, sec 5.1. Conceptual Model of Datastores additional nodes to a list entry or new list/leaf- list entries appearing in extends the list entry or the whole list/leaf-list defined in if the server allows the list/leaf-list to be updated. How is this achieved? This appears to suggest that there are two different= merging behaviours (one choice is to be additive, the other is to replace)= , and it seems to be down to the server to choose what to do on a case-by-c= ase basis. I think that it would be cleaner to define a single merge behav= iour if that is feasible (even if it is slightly less flexible). Also, pot= entially it is appropriate for the merge behaviour to be different for list= vs leaf-list (e.g., always merge list entries, but do a simple replace on = leaf-lists). [>>JTS:] I think part of the complication is that we want to allow the poss= ibility that there is a list that contains entries in , and it is a= completely non-modifiable list (no new entries allowed). I think maybe tha= t's what the "if the server allows" part is about. I agree though that it should be a merge for lists (and the server can just= error if that merge makes the list invalid, i.e. no additional entries all= owed on top of what's in system). For leaf-lists that a tough one (merge vs replace). I'm not sure what to do= there (and can imagine use cases for both merge and replace). [Qiufang] I think for both leaf-list and list cases, additive should be the= right answer when merging(note that client has no way to remove system con= fig, its lifecycle is beyond client control). But then ordering is another = question if it is "ordered-by user". I am not sure this draft is right place to define how the merge should happ= en, merge operation is there as early as 6241. Is there any difference when= it comes to being merged with , no? I expect the merge b= ehavior in this document be consistent with what's defined elsewhere. Just noticed that Kent has already raised a netconf-next issue for this: ht= tps://github.com/netconf-wg/netconf-next/issues/19. Maybe the right thing to do is to remove any text related to the merge beha= vior which also related to this sentence? [>>JTS:] You raise a good point about ordered-by user. That's going to make= the merge problematic. I don't think currently existing definitions really= address it for this draft. I'm doubtful we should actually define how merg= e of ordered-by-user lists should work for system->running. We may need to = leave that undefined (system specific) or disallow it (error, or make it a = replace). I'd lean towards leaving it undefined. (4) p 9, sec 5.1. Conceptual Model of Datastores If a server implements , MUST be merged into . This sentence is just repetition and can be deleted. The text above is sti= ll normative without the RFC 2119 MUST. [Qiufang] Yes! It is always the case that is merged into . Even though the server doesn't implement an explicit , there c= ould also be a conceptual one. Will remove it in the new revision. (5) p 13, sec 5.4. Modifying (Overriding) System Configuration For instance, descendant nodes in a system-defined list entry may be modifiable or not, even if some system configuration has been copied into earlier. If a system node is non-modifiable, then writing a different value for that node MUST return an error. The immutability of system configuration is defined in [I-D.ma-netmod-immutable-flag]. I think that some care is needed here. E.g., if the modification was being= done to , then it isn't writing a different value to that would return an error, but instead the or operat= ion that would fail. [>>JTS:] Agree. Maybe we can just add this? If a system node is non-modifiable, then writing a different value for that= node MUST return an error during a validate or commit operation. [Qiufang]"A validate operation" is easy to be confused with the = RPC operation, I think you're referring to the server's validation process,= not just RPC operation, right? Or we can just state writing a different value into MUST return a= n error. I am okay with either. [>>JTS:] I was talking about the operation. We should probably c= larify this in the text and not just keep the current sentence. (6) p 13, sec 5.4. Modifying (Overriding) System Configuration A server may also allow a client to add data nodes to a list entry in by writing those additional nodes in . Those additional data nodes may not exist in (i.e., an *addition* rather than an override). Earlier, the text in 5.1 seems to suggest that a list-entry could be overwr= itten. Is the intention that this is always a merge? I.e., it is possible= to override entries, but there is no way that running can remove a list en= try that is defined in . [Qiufang] I think it is an overriding case (e.g., clients overwrites a list= entry with the same key value), instead of different list/leaf-list entrie= s being merged together. [>>JTS:] I don't think there is a way to remove an entry in . You = can configure the *same* entry (key) in , and then potentially mod= ify child nodes if the entry is modifiable. [Qiufang] Agree. This section, 5.4., seems somewhat of a repeat of what is specified in sect= ion 5.1, and arguably it would be nice if this text could be co-located and= only specified once (for brevity and to avoid ambiguity). I'm wondering i= f the merge behaviour generally needs to be specified more explicitly. [Qiufang] will remove the duplicate text in the new revision. See comments = above, I am unsure to what extent the merge behavior should be specified in= this draft. Minor level comments: (7) p 0, sec This document defines how a management client and server handle YANG- modeled configuration data that is defined by the server itself. The system-defined configuration can be referenced (e.g. leafref) by configuration explicitly created by a client. Perhaps 'instantiated' by the server itself rather than 'defined' by the se= rver. [Qiufang] Okay, will fix this in the new revision, thanks. (8) p 0, sec The Network Management Datastore Architecture (NMDA) defined in RFC 8342 is updated with a read-only conventional configuration datastore called "system" to hold system-defined configuration. Perhaps 'expose system-defined configuration' to clients rather than 'hold' [Qiufang] Okay, will fix this in the new revision, thanks. (9) p 0, sec As an alternative to clients explicitly copying referenced system- defined configuration into the target configuration datastore (e.g., ) so that the datastore is valid, a "resolve-system" parameter is defined to allow the server acting as a "system client" to copy referenced system nodes automatically. This solution enables clients manipulating the target configuration datastore (e.g., ) to reference nodes defined in , override system- provided values, and configure descendant nodes of system-defined configuration. I think that this paragraph is too detailed to be in the abstract and shoul= d be removed from the abstract. [Qiufang] Will remove this paragraph in the new revision, thanks. (10) p 4, sec 1.1. Terminology The following terms are defined in this document: System configuration: Configuration that is provided by the system itself. System configuration is present in the system configuration datastore (regardless of whether it is applied or referenced) and appears in unless explicitly overridden. System configuration that is considered active appears in with origin=3D"system". It is a different and separate concept from factory default configuration defined in RFC 8808 (which represents a preset initial configuration that is used to initialize the configuration of a server). RFC 8808 should turn into a proper reference, it looks like it is just text= here. [Qiufang] Agree, will fix this. (11) p 5, sec 1.4. Updates to RFC 6241 and RFC 8526 This document defines a NETCONF protocol capability to indicate support for this parameter. NETCONF server that supports "resolve- system" parameter MUST advertise the following capability identifier: Are we ambiguous as to whether this must be supported, or is optional to im= plement? Ah, I see that this is specified later in the document (which is = arguably the right place). Is the capability really an update to RFC 6241 = and 8526? I wonder whether this last paragraph (i.e., the capability defin= ition) would be better under section 5.3. [Qiufang] You're right, I'll move the capability definition to sec.5.3, but= note that this is only specific to NETCONF protocol and sec.5.3 generally = is intends to be protocol-independent. (12) p 5, sec 1.5. Updates to RFC 8040 This document extends Sections 4.8 and 9.1.1 of [RFC8040] to add a new query parameter "resolve-system" and corresponding query parameter capability URI. Again, I think that possibly sections 1.5.1 and 1.5.2 would be better outsi= de of the introduction, perhaps as subsections of 5.3. Then section 1.5, c= ould then forward reference to those sections. [Qiufang]Sure, and this is the part that is only specific to RESTCONF proto= col. (13) p 6, sec 2. Kinds of System Configuration Active system configuration refers to system configuration that is currently in use. As per definition of the operational state datastore in [RFC8342], if system configuration is inactive, it does not appear in . However, system configuration is present in once it is generated, regardless of whether it is active or not. I'm not sure that calling this "active configuration" is a great choice, be= cause it seems to be a slightly different concept to inactive configuration= defined in RFC 8342. Specifically, I thought that the inactive configurat= ion in RFC 8342 controlled whether or not it would appear in , bu= t in this case, presumably it always turns up in if it is in and instead doesn't appear in ? [Qiufang] You are right, and maybe it also includes the naming of immediate= ly-active vs. conditionally-active? I agree it would be good to make a dist= inction here, maybe use "applied" instead of "active"? (15) p 7, sec 3. The System Configuration Datastore () * Management operations: The content of the datastore is set by the server in an implementation dependent manner. The content can not be changed by management operations via protocols such as NETCONF, RESTCONF, but may change itself by license change, device upgrade and/or system-controlled resources change. The datastore can be read using the standard network management protocols such as NETCONF and RESCTCONF. Rather than saying that the contents can change itself, I think that it wou= ld be better to say that the server may change the contents under various c= onditions, such as ... [Qiufang]Yes, it should be fixed. Thanks for pointing this out. (16) p 7, sec 3. The System Configuration Datastore () * Origin: This document does not define any new origin identity when it interacts with and flows into . The "system" origin Metadata Annotation [RFC7952] is used to indicate the origin of a data item is system, which is achieved by updating the definition of "intended" origin metadata annotation in [RFC8342]. If a different value is configured in that overrides a value in <= system> then it is clear that the origin should be . Do we speci= fy what the origin should be if the same value exists in both and= (which could be a very common occurrence if resolve-system is use= d)? [Qiufang] Yes, it is still "intended" for system config copied into , being specified in sec.5.1. Do you expect it to be specified here? (17) p 8, sec 3. The System Configuration Datastore () The system datastore is defined as a conventional configuration datastore and shares a common datastore schema with other conventional datastores. This paragraph should probably move up to "YANG modules". [Qiufang] Sure, will move to sec.6.1. (18) p 8, sec 4.2. May Change via Software Upgrades or Resource Changes * Servers rejects the operation to change system configuration (e.g., device upgrade fails) and needs the client to correct the configuration in as a prerequisite to ensure validity Should we add a recommendation for servers to document how they handle thes= e issues? [Qiufang] Okay, so how about adding the following text: Servers are recommended to include some hints in error responses to help cl= ients understand how should be updated. (19) p 10, sec 5.1. Conceptual Model of Datastores ct =3D config true; cf =3D config false rw =3D read-write; ro =3D read-only boxes denote named datastores In this diagram, (1) please move the system box 1 line to the left to keep = is more cleanly separate from the arrow into running. [Qiufang] That is easy to fix. (2) I think that we should discuss whether the running and system arrows sh= ould merge at a common point rather than running flowing into the side. [Qiufang] By "running flowing into the side", are you referring to running = directly flowing into intended? In the previous version (https://datatracker.ietf.org/doc/html/draft-ietf-n= etmod-system-config-04#section-5.1), we do have both running and system (eq= ually) flowing into intended. But I feel that way we cannot emphasize runni= ng takes precedence over system, and now it is updated like, to have system= as the "overlay", and running is merged into system to create intended. Bu= t I am unsure if that makes sense, e.g., to have one flow arrow points to a= nother one. Thoughts? (20) p 10, sec 5.1. Conceptual Model of Datastores ct =3D config true; cf =3D config false rw =3D read-write; ro =3D read-only boxes denote named datastores I know that it isn't directly related to this work, but I wonder whether th= e "default configuration" arrow is really in the right place, and whether t= hat shouldn't also be feeding this arrow into , since validation = would surely take default values into account. But this is perhaps a quest= ion for another day ... [Qiufang] I agree that validation should take default configuration into ac= count. Should we allow the default configuration to be present in = ? Jan raised a comment which is about the interplay between system config and= default configuration, which might not tightly relate to your comment, I a= m not sure that discussion might end up with, but maybe it is worth documen= ting some outcome and also pointing this out? (21) p 11, sec 5.1. Conceptual Model of Datastores Any deletable system-provided configuration that is populated as part of by the system at boot up, without being part of the contents of a datastore, must be defined in [RFC8808], which is used to initialize when the device is first-time powered on or reset to its factory default condition. I agree with the sentiment of what is written here, but I'm not sure that i= t is wise to restate it in this document, or whether it would be better to = delete this paragraph and just reference RFC 8808. E.g., maybe something like .. [RFC8808] defines a mechanism for populating at= system boot up with regular configuration data nodes, that hence can be de= leted. [Qiufang] I think there might be some slight difference, the intent here is= not to highlight contents in is deletable, but deletable= system config must be defined in and outside the scope o= f this draft. Would it be okay for you if we remove the part starting from = "which is used to initialize ..."? (22) p 12, sec 5.3. Servers Auto-configuring Referenced System Configurati= on ("resolve-system" parameter) The "resolve-system" parameter is optional and has no value. If it is present, and the server supports this capability, the server MUST copy referenced system nodes into the target datastore (e.g., ) without the client doing the copy/paste explicitly, to resolve any references not resolved by the client. The server acting as a "system client" like any other remote clients copies the referenced system-defined nodes when triggered by the "resolve- system" parameter. Legacy clients interacting with servers that support this parameter don't see any changes in / and behaviors. How does resolve-system interplay with the candidate configuration datastor= e? E.g., should it also be listed in the examples of datastores. What abo= ut the or operations? Is there any impact of private-c= andidate datastores, and if so, where should that be documented? [Qiufang] I assume the resolve-system parameter can be used for any r-w con= fig datastore, including private-candidate (will list them all in the examp= les). Given the auto-copy happens during validation time, for candidate/pri= vate-candidate, the does not necessarily cause the server to = perform the auto-copy, e.g., if the test-option value in is "= set". And in another mail response to the comments from Jan, some question = in mind is whether we should also augment the and RPCs= to support this parameter. Regarding the impact of introducing this parameter to private-candidate dat= astores, generally, once the config is copied into private-candidate, it's = no different from the configuration explicitly configured by the client, an= d thus IMO needs no specific handling. For example: 1. If contains an interface named "loopback" with "enabled" val= ue "true" which is not in at t0 time; 2. Client1 issues an to create priv-candidate #1 and leafr= efs the loopback interface without specifying the "resolve-system" at t1 ti= me; 3. Client2 issues an to set the loopback interface "enable= " value to "false" towards at t2; 4. Client1 issues the (assume "resolve-system" is carried) whic= h will automatically issue the as per priv-candidate draft, and th= us what is set in (step3) will be synced into priv-candidate #1, = no extra copy for the server needs to be done; and if we don't have step 3,= the server just copies config of loopback interface from ; Make sense? (23) p 12, sec 5.3. Servers Auto-configuring Referenced System Configurati= on ("resolve-system" parameter) The server's copy referenced nodes from to the target datastore MUST be enforced at the end of the / or operations during the validation processing, regardless of which target datastore it is. This probably means that it isn't a separate "system client" because I woul= d expect that to turn in the commit history as a separate commit, but inste= ad, the update to running via resolve-system is exactly the same as if the = client had made the modification directly as part of an edit-data (or simil= ar) operation. [Qiufang] The idea is that the server writing some configuration into when triggered by the "resolve-system" behaves just like a client writ= ing into it, because the server won't make a distinction in then,= origin value for both type is reported as "intended". But you interpretati= on makes sense to me, I guess the "system client" expression should be remo= ved in the next revision to avoid any potential confusion. (24) p 13, sec 5.3. Servers Auto-configuring Referenced System Configurati= on ("resolve-system" parameter) If the "resolve-system" parameter is not given by the client, the server should not modify in any way otherwise not specified by the client. Not using capitalized "SHOULD NOT" in the previous sentence is intentional. The intention is to bring awareness to the general need to not surprise clients with unexpected changes. It is desirable for clients to always opt into using mechanisms having server-side changes. This document enables a client to opt into this behavior using the "resolve-system" parameter. An example of this type of opt-in behavior can also be found in RFC 7317, which enables a client to opt into its behavior using a "$0$" prefix (see ianach:crypt-hash type defined in [RFC7317]). Arguably, I don't think that above paragraph is needed at all and can just = be removed. Otherwise, you could argue that it perhaps conflicts with the = text in 4.2? [Qiufang]As the co-author, I want to keep this paragraph as it reflects wha= t the WG has been expecting for : to have control over it. The ini= tial proposal for this draft is to allow the server to populate system conf= ig into when the device is powered on, but it has evolved into wh= at it is now, I think this general principle of client-control is important= . Or would it work for you if we only keep the first sentence, i.e., "If th= e "resolve-system" parameter is not given by the client, the server should = not modify in any way otherwise not specified by the client. "? But you are right that this seems to contradict the text in 4.2. For sec.4.= 2, the case is (or should be, will make it clear in sec.4.2) locked down to= software upgrades. And above text is applied to the scenarios other than s= oftware upgrades. (25) p 13, sec 5.3. Servers Auto-configuring Referenced System Configurati= on ("resolve-system" parameter) Implementation specifics are beyond the scope of this document, however, due to the extra complexity brought by the "resolve-system" parameter, clients should be aware that it would cost a reasonable amount of time for the server to resolve reference, retrieve and copy the referenced system configuration from , which could take multiple rounds since some errors may depend on the resolution of previous ones. Suggest changing "it would cost" to "it may take". But I'm also not really= sure that this paragraph should be in the document (e.g., what is a reason= able amount of time? Is it 1 second, or a minute, or a few minutes). [Qiufang] This is about some implementation consideration for "resolve-syst= em" parameter, the authors added this due to one comment received from the = WG that this parameter might be expensive to implement properly because of= the reason illustrated above. So do you think the extra complexity introdu= ced by this parameter, and the delayed response time of operation deserved = to be mentioned here? Or any suggestion to improve the text? I am really op= en to this. (26) p 24, sec 6.2. Example Usage The local port and remote port are used when the BGP peer connection is established. Since both are not supplied explicitly in and , the default value for "bgp/peer/remote-port" is used, and there is no default statement for "bgp/peer/local-port", the system will select a value for it. So the contents of are shown as follows: There is some level of interplay here between YANG default values and the i= nformation present in system. E.g., depending on how the YANG data model i= s written (i.e., sometimes complex default values are specifying in descrip= tion statements rather than as formal YANG defaults), then the choice as to= whether to report a value in system vs a default value in the configuratio= n may be a bit ambiguous. [Qiufang] Agree, note that the "default" origin value defined in RFC8342 is= specified as follows: default: represents configuration using a default value specified in the data model, using either values in the "default" statement or any values described in the "description" statement. The default origin is only used when the configuration has not been provided by any other source. Currently doesn't contain any default configuration, and mainly to= seek consistence with the origin definition in 8342, i.e., only the config= uration with origin value "system" is treated as system config and thus def= ined in . Is the consistence necessary and worth maintaining? E.g., this example of using the local-port as an example of system configur= ation potentially feels like the weakest of the alternative justifications = that have been provided. [Qiufang] Are you suggesting to document this explicitly? I am wondering wh= ether we need a brand-new subsection to talk about the interplay between de= fault vs. system, once agreement is reached, maybe this is worth documentin= g. (27) p 29, sec 7.3. YANG Module description "When present, the server is allowed to automatically configure referenced system configuration into the target configuration datastore."; Should this be "is allowed to automatically configure", or should it be "mu= st automatically configure"? [Qiufang] Yes, you are right, the description needs to be updated to be con= sistent with the related description in sec.5.3. Thanks for pointing it out= . I will also check other description statements in the yang module. (28) p 31, sec 9.2. Regarding the "ietf-netconf-resolve-system" YANG Modul= e The security considerations for the base NETCONF protocol operations (see Section 9 of [RFC6241] apply to the new extended RPC operations defined in this document. Possibly, this section should say a bit more about the security impacts of = supporting the resolve-system option, i.e., that there aren't any beyond th= e potential performance impacts of implementing resolve-system, which may m= ean that employing some form of rate limiting of requests specifying this o= ption might be a good idea to avoid DoS attacks. [Qiufang]Having to say I don't have much security knowledge, but as far as = I know, some systems uses rate limiting already, but given the extra perfor= mance impacts introduced by resolve-system parameter, we might ask implemen= tations to adjust the limited rate threshold for better protection. Would t= he following tweaking work for you: There is not any beyond the potential performance impacts of implementing t= he "resolve-system" parameter, which may mean employing some form of rate l= imiting or adapting the rate threshold might be a good idea to avoid DoS at= tacks. (29) p 35, sec Appendix A. Key Use Cases A.1. Device Powers On Please provide a short prose description of what the example illustrates. [Qiufang]Yes, I agree this needs to be provided. (30) p 35, sec Appendix A. Key Use Cases : Please expand these, e.g. The datastore contains: [Qiufang] Will expand these in the new revision, thanks. (31) p 35, sec Appendix A. Key Use Cases : Please expand these, e.g. The datastore contains: [Qiufang] Will expand these in the new revision, thanks. (32) p 35, sec Appendix A. Key Use Cases : Please expand these, e.g. After merging, the datastore contains= : [Qiufang] Will expand these in the new revision, thanks. (33) p 35, sec Appendix A. Key Use Cases : Please expand these, e.g. Once the configuration is applied, the datastore contains: [Qiufang] Will expand these in the new revision, thanks. (34) p 36, sec Appendix A. Key Use Cases : Please expand these simiarly to above for the other examples, A.2 and A.3. [Qiufang] Sure, will fix them all. (35) p 36, sec Appendix A. Key Use Cases As per a previous comment, I wonder whether the origin of the 'interfaces' = container itself should be 'intended' or 'system' (given than loopback alwa= ys exists and hence it can never be removed). [Qiufang] Yes, config in always exists and can never be removed. I failed to mention in response to your above comments that this is also th= e very first issue that we discussed at the interim we had in January (what= 's the origin value for config copied from into ), and th= e consensus of the participants was that system configuration copied from <= system> into should have origin value "intended" when flowing int= o . So configuration once written into , it reflects the intent of ope= rators and thus always takes precedence over what's in , this is in= dependent of whether it's an overridden or copy of system config. Does this= make sense? (36) p 37, sec Appendix A. Key Use Cases lo0 127.0.0.1 ::1 A.3. Operator Installs Card into a Chassis Please provide a short prose description of what the example illustrates. [Qiufang]Sure, will do. Nit level comments: (37) p 5, sec 1.3. Updates to RFC 8342 Configuration in is merged into to create the contents of after the configuration transformations to (e.g., template expansion, removal of inactive configuration defined in [RFC8342]) have been performed. This document updates the definition of "intended" origin metadata annotation identity to allow a subset of configuration provided by to use "system" as origin value as it flows into . Applied system configuration appears in with origin value being reported as "system" (Section 5.1). I think that " is merged into " is confusing. I would say= that is merged with the contents of and how that merge = is performed must be specified. [Qiufang] Will update it to " is merged with the contents of ", and let's discuss the merge behavior in your comments #3? (38) p 8, sec 3. The System Configuration Datastore () * Defining YANG module: "ietf-system-datastore". The datastore's content is defined by the server and read-only to clients. Upon the content is created or changed, it will be merged into . Unlike [RFC8808], it MAY change dynamically, e.g., depending on factors like license change, device upgrade or system-controlled resources change (e.g., HW available). The system configuration datastore doesn't persist across reboots; RPC operation defined in [RFC8808] can reset it to its factory default configuration without including configuration generated due to the system update or client-enabled functionality. Upon the content =3D> When the content. Some of the content here seems to = repeat the text in "Management operations", I think the examples would be b= etter in only a single place. [Qiufang]You're right, will remove the example description. Thanks. (39) p 8, sec 4.2. May Change via Software Upgrades or Resource Changes If system configuration changes (e.g., due to device upgrade), MAY become invalid. The server behaviors of migrating updated system data into is beyond the scope of this document. That said, the following gives a list of examples of server implementations that might be possible: Suggest rewording to: "That said, here are some examples of how a server mi= ght handle this scenario:" [Qiufang]Sure, will update this in the next version. (40) p 9, sec 4.3. No Impact to This work intends to have no impact to . System configuration appears in with origin value being reported as "system" if not configured or overridden explicitly in . This document enables a subset of those system generated nodes to be defined like configuration, i.e., made visible to clients in order for being referenced or configurable prior to present in . "Config false" nodes are out of scope, hence existing "config false" nodes are not impacted by this work. As per above, does "Overridden explicitly" mean "has a different value" in = running? [Qiufang]Yes, having a different value can be interpreted as "overridden",= and there are other cases, e.g., clients expanding a list entry to add new= descendant nodes which are not in is overriding the system list e= ntry. maybe also refer to sec.5.4. Generally, configuration in with its = origin being reported as unless the system configuration is copied= or overridden in . Is the existing text clear enough? Or please f= eel free to propose text. Note sec.5.1 also specifies that "Configuration copied from into <= running> has its origin value reported as "intended" when it flows into ." Maybe things related to the origin should be specified in one = place, e.g., adding a new subsection to talk about origin value? (41) p 12, sec 5.3. Servers Auto-configuring Referenced System Configurati= on ("resolve-system" parameter) Note that even an auto-configured node is allowed to be deleted from the target datastore by the client, the system may automatically configure the deleted node again to make configuration valid, when a "resolve-system" parameter is carried. It is also possible that the operation request (e.g., ) may not succeed due to incomplete referential integrity. Perhaps "recreate the deleted node" rather than "configure the deleted node= ". [Qiufang] Yes, your proposal is better. (42) p 12, sec 5.3. Servers Auto-configuring Referenced System Configurati= on ("resolve-system" parameter) Support for the "resolve-system" parameter is OPTIONAL. Servers not supporting NMDA [RFC8342] MAY also implement this parameter without implementing the system configuration datastore, which would only eliminate the ability to expose the system configuration via protocol operations. If a server implements , referenced system configuration is copied from into the target datastore (e.g., ) when the "resolve-system" parameter is used; otherwise it is an implementation decision where to copy referenced system configuration into the target datastore (e.g., ). Perhaps 'examine' rather than 'expose'. [Qiufang] I am not sure if word "examine" is proper here. By using "expose"= , it means that to enable the client to read the system config via standard= network management protocols (e.g., netconf and restconf). Could you please explain the intent to use "examine" here? Maybe "retrieve"= ? (43) p 21, sec 5.5.3. Modifying a System-instantiated Leaf's Value lo0 65536 127.0.0.1 ::1 A client modifies the value of MTU to 65535 and adds the following configuration into : I initially hadn't spotted the subtle change, perhaps use an MTU value that= is more obviosuly different from the value in system. E.g., perhaps 9216. [Qiufang] Sure, will fix this in the new revision! Thanks a lot taking the time for reviewing this document! All excellent com= ments! Regards, Rob Best Regards, Qiufang --_000_SN6PR08MB4847F2005D9219409B6CDCC29BE22SN6PR08MB4847namp_ Content-Type: text/html; charset="us-ascii" Content-Transfer-Encoding: quoted-printable

Thx Qiufa= ng. Please see inline.

Jason

&nbs= p;

From: maqiufang (A) <maqiufang1@huawei.com>
Sent: Monday, May 13, 2024 3:47 AM
To: Jason Sterne (Nokia) <jason.sterne@nokia.com>; Rob Wilton = (rwilton) <rwilton@cisco.com>; kent+ietf@watsen.net; NETMOD Group <= ;netmod@ietf.org>; draft-ietf-netmod-system-config@ietf.org
Subject: RE: [netmod] Re: WGLC on system-config-05=

 

Hi, = Jason and Rob,

 

Than= ks you both for your valuable comments, all good points, much appreciated. = Please also find my reply below inline...

 

-----Original Message----- From: Jason Sterne (Nokia) [mailt= o:jason.sterne@nokia.com]
Sent: Saturday, May 11, 2024 12:29 AM
To: Rob Wilton (rwilton) <rwilton=3D40cisco.com@dmarc.ietf.org>; Kent Watsen <kent+ietf@watsen.net>; netmod@ietf.org; draft-ietf-netmod-system-config@ietf.org
Subject: RE: [netmod] Re: WGLC on system-config-05

 

Please see inline for commen= ts on the "moderate level comments".  I'll try to reply late= r with more feedback on further items below.

Jason

 

From: Rob Wilton (rwilton) &= lt;rwilton=3D40cisco.com@dmarc.ietf.org<= /span>>

Sent: Thursday, May 9, 2024 = 9:55 AM

To: Kent Watsen <kent+ietf@watsen.net>; netmod@ietf.org; draft-ietf-netmod-system-config@ietf= .org

Subject: [netmod] Re: WGLC o= n system-config-05

 

CAUTION: This is an external= email. Please be very careful when clicking links or opening attachments. = See the URL nok.it/ext for additional information.

&= nbsp;

[Resending due to mailer iss= ues.]

[Qiu= fang] Thanks, I see it's been accurately archived now.

 

Hi authors, chairs, WG,=

 

I'm generally supportive of = this work, but I think that there are still some potential corner cases tha= t are not covered, or it isn't entirely obvious how they are handled.<= /o:p>

 

Comments below.

 

Moderate level comments:

 

(1) p 7, sec 2.3.  Inac= tive-Until-Referenced

 

   There are some = system configuration predefined (e.g., application

 

   ids, anti-x sig= natures, trust anchor certs, etc.) as a convenience

 

   for the clients= , which must be referenced to be active.  The clients

 

   can also define= their own configurations for their unique

 

   requirements.&n= bsp; Inactive-until-referenced system configuration is

 

   generated in &l= t;system> immediately when the device is powered on, but

 

   it is not activ= e until being referenced.

 

I'm not sure whether Inactiv= e-Until-Referenced actually needs to be defined, or to put it another way, = I'm not sure whether this type of configuration is special to system datast= ores at all.  If a configuration (either explicitly in <running> or implicitly from <system>) defines a= QoS policy that is not referenced from anywhere, (e.g., not applied to any= interfaces) then I think that it up to the server to decide whether that u= nreferenced QoS policy is reported in operational or not, depending on server implementation.

 

[>>JTS:] I agree. I th= ink section 2 may be mixing up two concepts:

 

  1.  Data dynamic= ally being populated/removed from the system datastore, and

  2.  For data tha= t is in the system datastore, whether it is "active" and present = in the operational datastore

[Qiu= fang] Yes, it’s actually from 2 dimensions to differentiate different= kinds of system config: 1. Time of being generated; 2. Time of being appli= ed.

 

For #2 I don't think there s= hould be anything special. We could say something like: As with the running= datastore, data present in the system datastore may or may not be present = in the operational datastore depending on whether it is considered as active configuration or not by the server.<= o:p>

[Qiu= fang] I am personally okay to remove the third kind of definition to keep t= he definition dimension consistent, and maybe for system configuration bein= g generated at both different times (immediately vs. conditional), they may be either applied by the server immediately or = only after being referenced. I also think it’s worth adding some text= as Jason suggested, the client would benefit from the knowledge that there= might be some system configuration that is defined there but not actually in use.

[>>JTS:] I think= it is more than just removing the 3rd type defined in section 2= . We probably need to rework to just have two types (for the one dimension)= :

  1. Immediately-present
  2. Conditionally-present

They would both relate= d purely to presence of config in <system> (and not say anything abou= t whether they are applied or active).

&= nbsp;

For #1 there is a bit of rep= eat with section 4.2. We should probably just talk about system config comi= ng and going (and changing) in one place.

[Qiu= fang] I think you are referring to the QoS examples in the first paragraph = of sec.4.2, right? We will move this to sec.2 in the new revision.

[>>JTS:] I wasn&= #8217;t just referring to the QoS example. It was the entire concept that t= he contents of the <system> datastore can change. But maybe it is OK = if that concept is mentioned in the 2 places. Perhaps we should just refer back to section 2 from 4.2 when we mention things dyn= amically showing up in <system>.  I would not necessarily move y= our QoS example out of 4.2 – it seems useful there (as well as useful= in section 2).

 

We'd have to update other pa= rts of the doc (e.g. 5.1) where these 3 types of data are mentioned.

[Qiu= fang]Yes, I agree.

 

(2) p 9, sec 5.1.  Conc= eptual Model of Datastores

 

   When the device= is powered on, immediately-active system

 

   configuration i= s generated in <system> and active immediately, but=

 

   inactive-until-= referenced system configuration only becomes active if

 

   referenced by c= lient-defined configuration.  However, conditionally-

 

   active system c= onfiguration will only be created and active when

 

   specific condit= ions on system resources are met.

 

I think that it should be &q= uot;merged with system" not "merged into system" since the r= unning configuration never ends up in the system datastore.

 

[>>JTS:] Yes. Jan ment= ioned the same.

[Qiu= fang] This will be fixed in the new revision, thanks!

 

(3) p 9, sec 5.1.  Conc= eptual Model of Datastores

 

    &nbs= p;             = additional nodes to a list entry or new list/leaf-

 

   list entries ap= pearing in <running> extends the list entry or the<= /p>

 

   whole list/leaf= -list defined in <system> if the server allows the<= /p>

 

   list/leaf-list = to be updated.

 

How is this achieved?  = This appears to suggest that there are two different merging behaviours (on= e choice is to be additive, the other is to replace), and it seems to be do= wn to the server to choose what to do on a case-by-case basis.  I think that it would be cleaner to define a s= ingle merge behaviour if that is feasible (even if it is slightly less flex= ible).  Also, potentially it is appropriate for the merge behaviour to= be different for list vs leaf-list (e.g., always merge list entries, but do a simple replace on leaf-lists).

 

[>>JTS:] I think part = of the complication is that we want to allow the possibility that there is = a list that contains entries in <system>, and it is a completely non-= modifiable list (no new entries allowed). I think maybe that's what the "if the server allows" part is about.=

 

I agree though that it shoul= d be a merge for lists (and the server can just error if that merge makes t= he list invalid, i.e. no additional entries allowed on top of what's in sys= tem).

 

For leaf-lists that a tough = one (merge vs replace). I'm not sure what to do there (and can imagine use = cases for both merge and replace).

[Qiu= fang] I think for both leaf-list and list cases, additive should be the rig= ht answer when merging(note that client has no way to remove system config,= its lifecycle is beyond client control). But then ordering is another question if it is “ordered-by user̶= 1;.

I am= not sure this draft is right place to define how the merge should happen, = merge operation is there as early as 6241. Is there any difference when it = comes to <running> being merged with <system>, no? I expect the merge behavior in this document be consistent with what&#= 8217;s defined elsewhere.

Just= noticed that Kent has already raised a netconf-next issue for this:= https://github.com/netcon= f-wg/netconf-next/issues/19.

Mayb= e the right thing to do is to remove any text related to the merge behavior= which also related to this sentence?

[>>JTS:] You rai= se a good point about ordered-by user. That’s going to make the merge= problematic. I don’t think currently existing definitions really add= ress it for this draft. I’m doubtful we should actually define how merge of ordered-by-user lists should work for system->runni= ng. We may need to leave that undefined (system specific) or disallow it (e= rror, or make it a replace). I’d lean towards leaving it undefined.

 

(4) p 9, sec 5.1.  Conc= eptual Model of Datastores

    &nbs= p;            &= nbsp;           &nbs= p;        If a server implements

 

   <intended>= ;, <system> MUST be merged into <intended>.

 

This sentence is just repeti= tion and can be deleted.  The text above is still normative without th= e RFC 2119 MUST.

[Qiu= fang] Yes! It is always the case that <system> is merged into <int= ended>. Even though the server doesn’t implement an explicit <i= ntended>, there could also be a conceptual one. Will remove it in the new revision.

 

(5) p 13, sec 5.4.  Mod= ifying (Overriding) System Configuration

 

   For instance, d= escendant nodes in a system-defined list entry may be

 

   modifiable or n= ot, even if some system configuration has been copied

 

   into <runnin= g> earlier.  If a system node is non-modifiable, then

 

   writing a diffe= rent value for that node MUST return an error.  The<= /p>

 

   immutability of= system configuration is defined in

 

   [I-D.ma-netmod-= immutable-flag].

 

I think that some care is ne= eded here.  E.g., if the modification was being done to <candidate&= gt;, then it isn't writing a different value to <candidate> that woul= d return an error, but instead the <validate> or <commit> operation that would fail.

 

[>>JTS:] Agree. Maybe = we can just add this?

 

If a system node is non-modi= fiable, then writing a different value for that node MUST return an error d= uring a validate or commit operation.

[Qiu= fang]”A validate operation” is easy to be confused with the <= ;validate> RPC operation, I think you’re referring to the server&#= 8217;s validation process, not just <validate> RPC operation, right?<= o:p>

Or w= e can just state writing a different value into <running> MUST return= an error. I am okay with either.

[>>JTS:] I was t= alking about the <validate> operation. We should probably clarify thi= s in the text and not just keep the current sentence.

 

(6) p 13, sec 5.4.  Mod= ifying (Overriding) System Configuration

 

   A server may al= so allow a client to add data nodes to a list entry in

 

   <system> = by writing those additional nodes in <running>.  Those

 

   additional data= nodes may not exist in <system> (i.e., an *addition*

 

   rather than an = override).

 

Earlier, the text in 5.1 see= ms to suggest that a list-entry could be overwritten.  Is the intentio= n that this is always a merge?  I.e., it is possible to override entri= es, but there is no way that running can remove a list entry that is defined in <system>.

[Qiu= fang] I think it is an overriding case (e.g., clients overwrites a list ent= ry with the same key value), instead of different list/leaf-list entries be= ing merged together.=

[>>JTS:] I don't think= there is a way to remove an entry in <system>.  You can configu= re the *same* entry (key) in <running>, and then potentially modify c= hild nodes if the entry is modifiable.

[Qiu= fang] Agree.

 

This section, 5.4., seems so= mewhat of a repeat of what is specified in section 5.1, and arguably it wou= ld be nice if this text could be co-located and only specified once (for br= evity and to avoid ambiguity).  I'm wondering if the merge behaviour generally needs to be specified more expl= icitly.

[Qiu= fang] will remove the duplicate text in the new revision. See comments abov= e, I am unsure to what extent the merge behavior should be specified in thi= s draft.

 

Minor level comments:

 

(7) p 0, sec

 

   This document d= efines how a management client and server handle YANG-

 

   modeled configu= ration data that is defined by the server itself.  The

 

   system-defined = configuration can be referenced (e.g. leafref) by

 

   configuration e= xplicitly created by a client.

 

Perhaps 'instantiated' by th= e server itself rather than 'defined' by the server.

[Qiu= fang] Okay, will fix this in the new revision, thanks.

 

(8) p 0, sec

 

   The Network Man= agement Datastore Architecture (NMDA) defined in RFC

 

   8342 is updated= with a read-only conventional configuration datastore

 

   called "sy= stem" to hold system-defined configuration.

 

Perhaps 'expose system-defin= ed configuration' to clients rather than 'hold'

[Qiu= fang] Okay, will fix this in the new revision, thanks.

 

(9) p 0, sec

 

   As an alternative to clients explicitly copy= ing referenced system-

 

   defined configuration into the target config= uration datastore (e.g.,

 

   <running>) so that the datastore is va= lid, a "resolve-system"

 

   parameter is defined to allow the server act= ing as a "system client"

 

   to copy referenced system nodes automaticall= y.  This solution enables

 

   clients manipulating the target configuratio= n datastore (e.g.,

 

   <running>) to reference nodes defined = in <system>, override system-<= /span>

 

   provided values, and configure descendant no= des of system-defined

 

   configuration.

 

I think that this paragraph is too detailed to be in the = abstract and should be removed from the abstract.

[Qiufang] Will remove this paragraph in the new revisio= n, thanks.

 

(10) p 4, sec 1.1.  Terminology

 

   The following terms are defined in this docu= ment:

 

   System configuration:  Configuration th= at is provided by the system<= /p>

 

      itself.  System confi= guration is present in the system

 

      configuration datastore (r= egardless of whether it is applied or

 

      referenced) and appears in= <intended> unless explicitly<= /span>

 

      overridden.  System c= onfiguration that is considered active

 

      appears in <operational= > with origin=3D"system".  It is a different

 

      and separate concept from = factory default configuration defined in

 

      RFC 8808 (which represents= a preset initial configuration that is

 

      used to initialize the con= figuration of a server).

 

RFC 8808 should turn into a proper reference, it looks li= ke it is just text here.

[Qiufang] Agree, will fix this.

 

(11) p 5, sec 1.4.  Updates to RFC 6241 and RFC 8526=

 

   This document defines a NETCONF protocol cap= ability to indicate

 

   support for this parameter.  NETCONF se= rver that supports "resolve-

 

   system" parameter MUST advertise the fo= llowing capability identifier:

 

Are we ambiguous as to whether this must be supported, or= is optional to implement?  Ah, I see that this is specified later in = the document (which is arguably the right place).  Is the capability really an update to RFC 6241 and 8526?  I wonder wh= ether this last paragraph (i.e., the capability definition) would be better= under section 5.3.

[Qiufang] You’re right, I’ll move the capab= ility definition to sec.5.3, but note that this is only specific to NETCONF= protocol and sec.5.3 generally is intends to be protocol-independent.=

 

(12) p 5, sec 1.5.  Updates to RFC 8040

 

   This document extends Sections 4.8 and 9.1.1= of [RFC8040] to add a

 

   new query parameter "resolve-system&quo= t; and corresponding query

 

   parameter capability URI.

 

Again, I think that possibly sections 1.5.1 and 1.5.2 wou= ld be better outside of the introduction, perhaps as subsections of 5.3.&nb= sp; Then section 1.5, could then forward reference to those sections.

[Qiufang]Sure, and this is the part that is only specif= ic to RESTCONF protocol.

 

(13) p 6, sec 2.  Kinds of System Configuration

 

   Active system configuration refers to system= configuration that is

 

   currently in use.  As per definition of= the operational state

 

   datastore in [RFC8342], if system configurat= ion is inactive, it does

 

   not appear in <operational>.  How= ever, system configuration is=

 

   present in <system> once it is generat= ed, regardless of whether it is

 

   active or not.

 

I'm not sure that calling this "active configuration= " is a great choice, because it seems to be a slightly different conce= pt to inactive configuration defined in RFC 8342.  Specifically, I thought that the inactive configuration in RFC 8342 contro= lled whether or not it would appear in <intended>, but in this case, = presumably it always turns up in <intended> if it is in <system>= ; and instead doesn't appear in <operational>?

[Qiufang] You are right, and maybe it also includes the= naming of immediately-active vs. conditionally-active? I agree it would be= good to make a distinction here, maybe use “applied” instead of “active”?

 

(15) p 7, sec 3.  The System Configuration Datastore= (<system>)

 

   *  Management operations: The content o= f the datastore is set by the=

 

      server in an implementatio= n dependent manner.  The content can not

 

      be changed by management o= perations via protocols such as NETCONF,

 

      RESTCONF, but may change i= tself by license change, device upgrade

 

      and/or system-controlled r= esources change.  The datastore can be

 

      read using the standard ne= twork management protocols such as

 

      NETCONF and RESCTCONF.

 

Rather than saying that the contents can change itself, I= think that it would be better to say that the server may change the conten= ts under various conditions, such as ...

[Qiufang]Yes, it should be fixed. Thanks for pointing t= his out.

 

(16) p 7, sec 3.  The System Configuration Datastore= (<system>)

 

   *  Origin: This document does not defin= e any new origin identity when

 

      it interacts with <inte= nded> and flows into <operational>.  The

 

      "system" origin = Metadata Annotation [RFC7952] is used to indicate

 

      the origin of a data item = is system, which is achieved by updating

 

      the definition of "in= tended" origin metadata annotation in=

 

      [RFC8342].

 

If a different value is configured in <running> tha= t overrides a value in <system> then it is clear that the origin shou= ld be <intended>.  Do we specify what the origin should be if the same value exists in both <running> and <system> (wh= ich could be a very common occurrence if resolve-system is used)?

[Qiufang] Yes, it is still “intended” for s= ystem config copied into <running>, being specified in sec.5.1. Do yo= u expect it to be specified here?

 

(17) p 8, sec 3.  The System Configuration Datastore= (<system>)

 

   The system datastore is defined as a convent= ional configuration

 

   datastore and shares a common datastore sche= ma with other

 

   conventional datastores.

 

This paragraph should probably move up to "YANG modu= les".

[Qiufang] Sure, will move to sec.6.1.=

 

(18) p 8, sec 4.2.  May Change via Software Upgrades= or Resource Changes

 

   *  Servers rejects the operation to cha= nge system configuration

 

      (e.g., device upgrade fail= s) and needs the client to correct the

 

      configuration in <runni= ng> as a prerequisite to ensure validity

 

Should we add a recommendation for servers to document ho= w they handle these issues?

[Qiufang] Okay, so how about adding the following text:=

Servers are recommended to include some hints in error = responses to help clients understand how <running> should be updated.=

 

(19) p 10, sec 5.1.  Conceptual Model of Datastores<= /span>

 

    ct =3D config true; cf =3D config fals= e

 

    rw =3D read-write; ro =3D read-only

 

    boxes denote named datastores

 

In this diagram, (1) please move the system box 1 line to= the left to keep is more cleanly separate from the arrow into running.

[Qiufang] That is easy to fix.

(2) I think that we should discuss whether the running an= d system arrows should merge at a common point rather than running flowing = into the side.

[Qiufang] By “running flowing into the side”= ;, are you referring to running directly flowing into intended?

In the previous version (https://datatracker.ietf.org/doc/html/draft-ietf-netm= od-system-config-04#section-5.1), we do have both running and system (equally) flowing into intended. But I = feel that way we cannot emphasize running takes precedence over system, and= now it is updated like, to have system as the “overlay”, and r= unning is merged into system to create intended. But I am unsure if that makes sense, e.g., to have one flow arrow points t= o another one. Thoughts?

 

(20) p 10, sec 5.1.  Conceptual Model of Datastores<= /span>

 

    ct =3D config true; cf =3D config fals= e

 

    rw =3D read-write; ro =3D read-only

 

    boxes denote named datastores

 

I know that it isn't directly related to this work, but I= wonder whether the "default configuration" arrow is really in th= e right place, and whether that shouldn't also be feeding this arrow into <intended>, since validation would surely ta= ke default values into account.  But this is perhaps a question for an= other day ...

[Qiufang] I agree that validation should take default c= onfiguration into account.  Should we allow the default configuration = to be present in <system>?

Jan raised a comment which is about the interplay betwe= en system config and default configuration, which might not tightly relate = to your comment, I am not sure that discussion might end up with, but maybe it is worth documenting some outcome and also= pointing this out?

 

(21) p 11, sec 5.1.  Conceptual Model of Datastores<= /span>

 

   Any deletable system-provided configuration = that is populated as part

 

   of <running> by the system at boot up,= without being part of the

 

   contents of a <startup> datastore, mus= t be defined in <factory-<= /p>

 

   default> [RFC8808], which is used to init= ialize <running> when the

 

   device is first-time powered on or reset to = its factory default

 

   condition.<= /o:p>

 

I agree with the sentiment of what is written here, but I= 'm not sure that it is wise to restate it in this document, or whether it w= ould be better to delete this paragraph and just reference RFC 8808.=

 

E.g., maybe something like ..=

 

<factory-default> [RFC8808] defines a mechanism for= populating <running> at system boot up with regular configuration da= ta nodes, that hence can be deleted.=

[Qiufang] I think there might be some slight difference= , the intent here is not to highlight contents in <factory-default> i= s deletable, but deletable system config must be defined in <factory-default> and outside the scope of this draft.= Would it be okay for you if we remove the part starting from “which = is used to initialize <running> …”?

 

(22) p 12, sec 5.3.  Servers Auto-configuring Refere= nced System Configuration

      ("resolve-system"= ; parameter)

 

   The "resolve-system" parameter is = optional and has no value.  If it

 

   is present, and the server supports this cap= ability, the server MUST

 

   copy referenced system nodes into the target= datastore (e.g.,

 

   <running>) without the client doing th= e copy/paste explicitly, to

 

   resolve any references not resolved by the c= lient.  The server acting

 

   as a "system client" like any othe= r remote clients copies the

 

   referenced system-defined nodes when trigger= ed by the "resolve-

 

   system" parameter.  Legacy clients= interacting with servers that

 

   support this parameter don't see any changes= in <edit- config>/<edit-

 

   data> and <copy-config> behaviors.<= /span>

 

How does resolve-system interplay with the candidate conf= iguration datastore?  E.g., should it also be listed in the examples o= f datastores.  What about the <validate> or <commit> operations?  Is there any impact of private-candidate datastores, and= if so, where should that be documented?

[Qiufang] I assume the resolve-system parameter can be = used for any r-w config datastore, including private-candidate (will list t= hem all in the examples). Given the auto-copy happens during validation time, for candidate/private-candidate, the <e= dit-config> does not necessarily cause the server to perform the auto-co= py, e.g., if the test-option value in <edit-config> is “setR= 21;. And in another mail response to the comments from Jan, some question in mind is whether we should also augment the  <comm= it> and <validate> RPCs to support this parameter.

Regarding the impact of introducing this parameter to p= rivate-candidate datastores, generally, once the config is copied into priv= ate-candidate, it’s no different from the configuration explicitly configured by the client, and thus IMO needs no s= pecific handling. For example:

  1. If <system> contains an interface named “l= oopback” with “enabled” value “true” which is= not in <running> at t0 time;
  2. Client1 issues an <edit-config> to create priv-c= andidate #1 and leafrefs the loopback interface without specifying the R= 20;resolve-system” at t1 time;
  3. Client2 issues an <edit-config> to set the loopb= ack interface “enable” value to “false” towards <= ;running> at t2;
  4. Client1 issues the <commit> (assume “resol= ve-system” is carried) which will automatically issue the <update&= gt; as per priv-candidate draft, and thus what is set in <running> (s= tep3) will be synced into priv-candidate #1, no extra copy for the server needs to be done; and if we don’t have step 3, the server jus= t copies config of loopback interface from <system>;

Make sense?

 

(23) p 12, sec 5.3.  Servers Auto-configuring Refere= nced System Configuration

      ("resolve-system"= ; parameter)

 

   The server's copy referenced nodes from <= system> to the target

 

   datastore MUST be enforced at the end of the= <edit-config>/<edit-

 

   data> or <copy-config> operations d= uring the validation processing,

 

   regardless of which target datastore it is.<= /span>

 

This probably means that it isn't a separate "system= client" because I would expect that to turn in the commit history as = a separate commit, but instead, the update to running via resolve-system is exactly the same as if the client had made the modif= ication directly as part of an edit-data (or similar) operation.

[Qiufang] The idea is that the server writing some conf= iguration into <running> when triggered by the “resolve-system&= #8221; behaves just like a client writing into it, because the server won’t make a distinction in <running> then, origin = value for both type is reported as “intended”. But you interpre= tation makes sense to me, I guess the “system client” expressio= n should be removed in the next revision to avoid any potential confusion.<= o:p>

 

(24) p 13, sec 5.3.  Servers Auto-configuring Refere= nced System Configuration

      ("resolve-system"= ; parameter)

 

   If the "resolve-system" parameter = is not given by the client, the

 

   server should not modify <running> in = any way otherwise not specified

 

   by the client.  Not using capitalized &= quot;SHOULD NOT" in the previous

 

   sentence is intentional.  The intention= is to bring awareness to the=

 

   general need to not surprise clients with un= expected changes.  It is=

 

   desirable for clients to always opt into usi= ng mechanisms having

 

   server-side changes.  This document ena= bles a client to opt into this

 

   behavior using the "resolve-system"= ; parameter.  An example of this

 

   type of opt-in behavior can also be found in= RFC 7317, which enables

 

   a client to opt into its behavior using a &q= uot;$0$" prefix (see

 

   ianach:crypt-hash type defined in [RFC7317])= .

 

Arguably, I don't think that above paragraph is needed at= all and can just be removed.  Otherwise, you could argue that it perh= aps conflicts with the text in 4.2?<= /span>

[Qiufang]As the co-author, I want to keep this paragrap= h as it reflects what the WG has been expecting for <running>: to hav= e control over it. The initial proposal for this draft is to allow the server to populate system config into <running>= ; when the device is powered on, but it has evolved into what it is now, I = think this general principle of client-control is important. Or would it wo= rk for you if we only keep the first sentence, i.e., “If the &quo= t;resolve-system" parameter is not given by the client, the server sho= uld not modify <running> in any way otherwise not specified by the cl= ient. “?

But you are right that this seems to contradict the tex= t in 4.2. For sec.4.2, the case is (or should be, will make it clear in sec= .4.2) locked down to software upgrades. And above text is applied to the scenarios other than software upgrades.

 

(25) p 13, sec 5.3.  Servers Auto-configuring Refere= nced System Configuration

      ("resolve-system"= ; parameter)

 

   Implementation specifics are beyond the scop= e of this document,

 

   however, due to the extra complexity brought= by the "resolve-system"

 

   parameter, clients should be aware that it w= ould cost a reasonable

 

   amount of time for the server to resolve ref= erence, retrieve and copy

 

   the referenced system configuration from <= ;system>, which could take=

 

   multiple rounds since some errors may depend= on the resolution of

 

   previous ones.

 

Suggest changing "it would cost" to "it ma= y take".  But I'm also not really sure that this paragraph should= be in the document (e.g., what is a reasonable amount of time?  Is it 1 second, or a minute, or a few minutes).

[Qiufang] This is about some implementation considerati= on for “resolve-system” parameter, the authors added this due t= o one comment received from the WG that  this parameter might be expensive to implement properly because of the reason illustrated= above. So do you think the extra complexity introduced by this parameter, = and the delayed response time of operation deserved to be mentioned here? O= r any suggestion to improve the text? I am really open to this.

 

(26) p 24, sec 6.2.  Example Usage

 

   The local port and remote port are used when= the BGP peer connection

 

   is established.  Since both are not sup= plied explicitly in <running><= /span>

 

   and <intended>, the default value for = "bgp/peer/remote-port" is used,<= /o:p>

 

   and there is no default statement for "= bgp/peer/local-port", the

 

   system will select a value for it.  So = the contents of <system> are

 

   shown as follows:

 

There is some level of interplay here between YANG defaul= t values and the information present in system.  E.g., depending on ho= w the YANG data model is written (i.e., sometimes complex default values are specifying in description statements rather tha= n as formal YANG defaults), then the choice as to whether to report a value= in system vs a default value in the configuration may be a bit ambiguous.&= nbsp;

[Qiufang] Agree, note that the “default” or= igin value defined in RFC8342 is specified as follows:

default: represents configuration = using a default value specified

      in = the data model, using either values in the "default" statement

      or = any values described in the "description" statement.  The

      def= ault origin is only used when the configuration has not been

      pro= vided by any other source.

Currently <system> doesn’t contain any defa= ult configuration, and mainly to seek consistence with the origin definitio= n in 8342, i.e., only the configuration with origin value “system” is treated as system config and thus defined in= <system>.

Is the consistence necessary and worth maintaining?

E.g., this example of using the local-port as an example = of system configuration potentially feels like the weakest of the alternati= ve justifications that have been provided.=

[Qiufang] Are you suggesting to document this explicitl= y? I am wondering whether we need a brand-new subsection to talk about the = interplay between default vs. system, once agreement is reached, maybe this is worth documenting.

 

(27) p 29, sec 7.3.  YANG Module

 

         descript= ion

 

         &nb= sp; "When present, the server is allowed to automatically

 

         &nb= sp;  configure referenced system configuration into the

 

         &nb= sp;  target configuration datastore.";

 

Should this be "is allowed to automatically configur= e", or should it be "must automatically configure"?

[Qiufang] Yes, you are right, the description needs to = be updated to be consistent with the related description in sec.5.3. Thanks= for pointing it out. I will also check other description statements in the yang module.

 

(28) p 31, sec 9.2.  Regarding the "ietf-netcon= f-resolve-system" YANG Module

 

   The security considerations for the base NET= CONF protocol operations

 

   (see Section 9 of [RFC6241] apply to the new= extended RPC operations

 

   defined in this document.

 

Possibly, this section should say a bit more about the se= curity impacts of supporting the resolve-system option, i.e., that there ar= en't any beyond the potential performance impacts of implementing resolve-system, which may mean that employing some= form of rate limiting of requests specifying this option might be a good i= dea to avoid DoS attacks.

[Qiufang]Having to say I don’t have much security= knowledge, but as far as I know, some systems uses rate limiting already, = but given the extra performance impacts introduced by resolve-system parameter, we might ask implementations to adjust the li= mited rate threshold for better protection. Would the following tweaking wo= rk for you:

There is not any beyond the potential performance impac= ts of implementing the “resolve-system” parameter, which may me= an employing some form of rate limiting or adapting the rate threshold might be a good idea to avoid DoS attacks.

 

(29) p 35, sec Appendix A.  Key Use Cases

 

A.1.  Device Powers On

 

Please provide a short prose description of what the exam= ple illustrates.

[Qiufang]Yes, I agree this needs to be provided.

 

(30) p 35, sec Appendix A.  Key Use Cases

 

   <running>:=

 

Please expand these, e.g.  The <running> datas= tore contains:

[Qiufang] Will expand these in the new revision, thanks= .

 

(31) p 35, sec Appendix A.  Key Use Cases

 

   <system>:<= o:p>

 

Please expand these, e.g.  The <system> datast= ore contains:

[Qiufang] Will expand these in the new revision, thanks= .

 

(32) p 35, sec Appendix A.  Key Use Cases

 

   <intended>:

 

Please expand these, e.g.  After merging, the <in= tended> datastore contains:

[Qiufang] Will expand these in the new revision, thanks= .

 

(33) p 35, sec Appendix A.  Key Use Cases

 

   <operational>:

 

Please expand these, e.g.  Once the configuration is= applied, the <operational> datastore contains:

[Qiufang] Will expand these in the new revision, thanks= .

 

(34) p 36, sec Appendix A.  Key Use Cases

 

   <running>:=

 

Please expand these simiarly to above for the other examp= les, A.2 and A.3.

[Qiufang] Sure, will fix them all.

 

(35) p 36, sec Appendix A.  Key Use Cases

 

   <interfaces xmlns:or=3D"urn:ietf:par= ams:xml:ns:yang:ietf-origin"

 

         &nb= sp;     or:origin=3D"or:intended"><= span lang=3D"EN-US">

 

As per a previous comment, I wonder whether the origin of= the 'interfaces' container itself should be 'intended' or 'system' (given = than loopback always exists and hence it can never be removed).

[Qiufang] Yes, config in <system> always exists a= nd can never be removed.

I failed to mention in response to your above comments = that this is also the very first issue that we discussed at the interim we = had in January (what’s the origin value for config copied from <system> into <running>), and the conse= nsus of the participants was that system configuration copied from <syst= em> into <running> should have origin value “intended”= when flowing into <operational>.

So configuration once written into <running>, it = reflects the intent of operators and thus always takes precedence over what= ’s in <system>, this is independent of whether it’s an overridden or copy of system config. Does this make sense?

 

(36) p 37, sec Appendix A.  Key Use Cases

 

     <interface or:origin=3D"= or:system">

 

       <name>lo0</= name>

 

       <ip-address>12= 7.0.0.1</ip-address>

 

       <ip-address>::= 1</ip-address>

 

     </interface>

 

   </interfaces>

 

A.3.  Operator Installs Card into a Chassis

 

Please provide a short prose description of what the exam= ple illustrates.

[Qiufang]Sure, will do.

 

 

Nit level comments:

 

(37) p 5, sec 1.3.  Updates to RFC 8342

 

   Configuration in <running> is merged i= nto <system> to create the

 

   contents of <intended> after the confi= guration transformations to

 

   <running> (e.g., template expansion, r= emoval of inactive

 

   configuration defined in [RFC8342]) have bee= n performed.  This

 

   document updates the definition of "int= ended" origin metadata

 

   annotation identity to allow a subset of con= figuration provided by

 

   <intended> to use "system" a= s origin value as it flows into

 

   <operational>.  Applied system co= nfiguration appears in <operational>=

 

   with origin value being reported as "sy= stem" (Section 5.1).

 

I think that "<running> is merged into <sys= tem>" is confusing.  I would say that <running> is merge= d with the contents of <system> and how that merge is performed must = be specified.

[Qiufang] Will update it to “<running> is m= erged with the contents of <system>”, and let’s discuss t= he merge behavior in your comments #3?

 

(38) p 8, sec 3.  The System Configuration Datastore= (<system>)

 

   *  Defining YANG module: "ietf-sys= tem-datastore".

 

   The datastore's content is defined by the se= rver and read-only to

 

   clients.  Upon the content is created o= r changed, it will be merged<= /p>

 

   into <intended>.  Unlike <fact= ory-default> [RFC8808], it MAY change

 

   dynamically, e.g., depending on factors like= license change, device

 

   upgrade or system-controlled resources chang= e (e.g., HW available).

 

   The system configuration datastore doesn't p= ersist across reboots;

 

   <factory-reset> RPC operation defined = in [RFC8808] can reset it to<= /p>

 

   its factory default configuration without in= cluding configuration

 

   generated due to the system update or client= -enabled functionality.

 

 

Upon the content =3D> When the content.  Some of = the content here seems to repeat the text in "Management operations&qu= ot;, I think the examples would be better in only a single place.

[Qiufang]You’re right, will remove the example de= scription. Thanks.

 

 

(39) p 8, sec 4.2.  May Change via Software Upgrades= or Resource Changes

 

   If system configuration changes (e.g., due t= o device upgrade),

 

   <running> MAY become invalid.  Th= e server behaviors of migrating

 

   updated system data into <running> is = beyond the scope of this

 

   document.  That said, the following giv= es a list of examples of

 

   server implementations that might be possibl= e:

 

 

Suggest rewording to: "That said, here are some exam= ples of how a server might handle this scenario:"

[Qiufang]Sure, will update this in the next version.

 

 

(40) p 9, sec 4.3.  No Impact to <operational>=

 

   This work intends to have no impact to <o= perational>.  System<= /p>

 

   configuration appears in <operational>= with origin value being

 

   reported as "system" if not config= ured or overridden explicitly in

 

   <running>.  This document enables= a subset of those system generated<= /span>

 

   nodes to be defined like configuration, i.e.= , made visible to clients

 

   in order for being referenced or configurabl= e prior to present in

 

   <operational>.  "Config fals= e" nodes are out of scope, hence existing<= o:p>

 

   "config false" nodes are not impac= ted by this work.

 

As per above, does "Overridden explicitly" mean= "has a different value" in running?<= o:p>

[Qiufang]Yes,  having a different value can be int= erpreted as “overridden”, and there are other cases, e.g., clie= nts expanding a list entry to add new descendant nodes which are not in <system> is overriding the system list entry.<= /span>

maybe also refer to sec.5.4. Generally, configuration i= n <system> with its origin being reported as <system> unless th= e system configuration is copied or overridden in <running>. Is the existing text clear enough? Or please feel free to propose text.

 

Note sec.5.1 al= so specifies that “Configuration copied from <system> into <running> has its origin value repor= ted as "intended" when it flows into <operational>.<= span lang=3D"EN-US" style=3D"font-family:"Calibri",sans-serif;col= or:#1F4E79">” Maybe things related to the origin sho= uld be specified in one place, e.g., adding a new subsection to talk about = origin value?

 

(41) p 12, sec 5.3.  Servers Auto-configuring Refere= nced System Configuration

 

      ("resolve-system"= ; parameter)

 

   Note that even an auto-configured node is al= lowed to be deleted from

 

   the target datastore by the client, the syst= em may automatically

 

   configure the deleted node again to make con= figuration valid, when a

 

   "resolve-system" parameter is carr= ied.  It is also possible that the

 

   operation request (e.g., <edit-config>= ) may not succeed due to

 

   incomplete referential integrity.

 

 

Perhaps "recreate the deleted node" rather than= "configure the deleted node".

[Qiufang] Yes, your proposal is better.

 

(42) p 12, sec 5.3.  Servers Auto-configuring Refere= nced System Configuration

      ("resolve-system"= ; parameter)

 

 

   Support for the "resolve-system" p= arameter is OPTIONAL.  Servers not

 

   supporting NMDA [RFC8342] MAY also implement= this parameter without

 

   implementing the system configuration datast= ore, which would only

 

   eliminate the ability to expose the system c= onfiguration via protocol

 

   operations.  If a server implements <= ;system>, referenced system

 

   configuration is copied from <system> = into the target datastore

 

   (e.g., <running>) when the "resol= ve-system" parameter is used;

 

   otherwise it is an implementation decision w= here to copy referenced

 

   system configuration into the target datasto= re (e.g., <running>).

 

 

Perhaps 'examine' rather than 'expose'.

[Qiufang] I am not sure if word “examine” i= s proper here. By using “expose”, it means that to enable the c= lient to read the system config via standard network management protocols (e.g., netconf and restconf).

Could you please explain the intent to use “exami= ne” here? Maybe “retrieve”?

 

(43) p 21, sec 5.5.3.  Modifying a System-instantiat= ed Leaf's Value

 

   <interfaces xmlns=3D"urn:example:int= erface">

     <interface>

       <name>lo0</= name>

       <mtu>65536<= /mtu>

       <ip-address>12= 7.0.0.1</ip-address>

       <ip-address>::= 1</ip-address>

     </interface>

   </interfaces>

 

   A client modifies the value of MTU to 65535 = and adds the following

 

   configuration into <running>:

 

I initially hadn't spotted the subtle change, perhaps use= an MTU value that is more obviosuly different from the value in system.&nb= sp; E.g., perhaps 9216.

[Qiu= fang] Sure, will fix this in the new revision!

 

Thanks a lot taking the time for reviewing this documen= t! All excellent comments!

 

Regards,

 

Rob

 

Best= Regards,

Qiuf= ang

--_000_SN6PR08MB4847F2005D9219409B6CDCC29BE22SN6PR08MB4847namp_--