Re: [netmod] [TICTOC] Using instance-number or instance-name issue - RE: WG Last Call resolutions incorporated in draft-ietf-tictoc-1588v2-yang-06

[Jiangyuanlong <[e-mail]>]

Juergen & Bob, 

Can I assume you are in support of using instance-number in this case? ^&^
But seriously, maybe it makes sense to restrict the valid characters of a string, especially when it is used as a key.
Otherwise, we need to note this risk in "Security considerations" of a RFC (actually it can become a kind of DoS attack). 

Thanks,
Yuanlong

-----Original Message-----
From: TICTOC [mailto:[e-mail]] On Behalf Of Juergen Schoenwaelder
Sent: Wednesday, November 22, 2017 10:51 PM
To: Bob kb8tq
Cc: Xian Liu; Xujinchun; Karen O'Donoghue; [e-mail]; [e-mail]
Subject: Re: [TICTOC] [netmod] Using instance-number or instance-name issue - RE: WG Last Call resolutions incorporated in draft-ietf-tictoc-1588v2-yang-06

According to RFC 7950, section 9.4, YANG strings are Unicode and ISO/IEC 10646 characters, including tab, carriage return, and line feed but excluding the other C0 control characters, the surrogate blocks, and the noncharacters. Of course, handling Unicode correctly can still be a lot of fun and systems that do not get this right might still crash or lockup.

/js

On Wed, Nov 22, 2017 at 08:16:22AM -0500, Bob kb8tq wrote:
> Hi
> 
> If you change to “instance name” be very clear on the character set(s) 
> allowed. I have seen some really bad side effects when unexpected 
> character sets show up in ID fields. Software tries to parse it for presentation on a screen or into a log. The result is a crash or lockup.
> 
> Bob
> 
> > On Nov 21, 2017, at 10:18 PM, Xujinchun <[e-mail]> wrote:
> > 
> > Hi Rodney,
> > 
> > In my opinion, instance-name or instance-number does not matter if the number of instances are small. But if the instances may grow into hundreds or more in scale, then string should not be a choice.
> > 
> > We know how awkward it is to store and sort out a key of string compared with an integer.
> > 
> > Thanks
> > 
> > Jinchun XU
> > 
> > -----邮件原件-----
> > 发件人: Rodney Cummings [mailto:[e-mail]]
> > 发送时间: 2017年11月22日 1:56
> > 收件人: Jiangyuanlong; [e-mail]; Alex Campbell; Karen O'Donoghue
> > 抄送: Xian Liu; Xujinchun; [e-mail]
> > 主题: RE: Using instance-number or instance-name issue - RE: WG Last 
> > Call resolutions incorporated in draft-ietf-tictoc-1588v2-yang-06
> > 
> > Hi Yuanlong,
> > 
> > I have no concerns with instance-number, as that is what the upcoming 1588 revision outlines for management.
> > 
> > I also have no strong objections against changing instance-number to instance-name. If we do that, I think it would be best to make the same change in the upcoming 1588 revision. I asked the 1588 working group for opinion, but I haven't heard back on that.
> > 
> > All things being equal, my preference is to go with instance-number.
> > 
> > Rodney
> > 
> > -----Original Message-----
> > From: Jiangyuanlong [mailto:[e-mail]]
> > Sent: Monday, November 20, 2017 7:34 AM
> > To: [e-mail]; Alex Campbell ; Rodney Cummings ; Karen 
> > O'Donoghue
> > Cc: Xian Liu ; Xujinchun ; [e-mail]
> > Subject: Using instance-number or instance-name issue - RE: WG Last 
> > Call resolutions incorporated in draft-ietf-tictoc-1588v2-yang-06
> > 
> > Hi all,
> > 
> > Item #5 below is the last open issue we discussed both in emails and in IEEE 1588 mailing list on draft-ietf-tictoc-1588v2-yang.  
> > In a summary: in draft-ietf-tictoc-1588v2-yang, list instance-list has a key of "instance-number", but there were discussions whether to use instance-name (a string) instead.
> > 
> > Currently, "instance-number" in draft-ietf-tictoc-1588v2-yang-06 aligns well with the texts in the new revision of IEEE 1588 (D1.2/2017): 
> >   "The instanceList is indexed using a number that is unique per PTP Instance within the PTP Node, applicable to the 
> >   management context only (i.e. not used in PTP messages). The domainNumber of the PTP Instance must not be used as the index 
> >   to instanceList, since it is possible for a PTP Node to contain multiple PTP Instances using the same domainNumber."
> > 
> > The main requirement of instanceList in IEEE 1588 is the uniqueness of its index, and the "key" statement of YANG serves this purpose very well.
> > 
> > That is, when instance-number is used as a key, a PTP Node with multiple PTP Instances cannot use the same instance-number value for these PTP Instances (just according to YANG semantics).
> > 
> > Using instance-name (string) can also guarantee the uniqueness of the index of a list, but compared with an integer, a string is usually more complex to process and store. If instance-name is modeled as an arbitrary length of string, there is even a risk of buffer-overflow attack.
> > 
> > Furthermore, it should be noted that draft-ietf-tictoc-1588v2-yang is targeted at IEEE 1588-2008, for which most products today only have a single PTP instance, and not have a name for this instance, it seems quite weird to introduce a name for this instance.
> > 
> > Therefore, I would suggest we keep on using instance-number as a key. But as 65536 limit is a concern, I further suggest to change its type to uint32.
> > 
> > Any comments or concerns on this suggestion to move forward?
> > 
> > Thanks,
> > Yuanlong
> > 
> > ----- Original Message -----
> > From: "Jiangyuanlong" <[e-mail]>
> > To: "Alex Campbell" <[e-mail]>; <[e-mail]>
> > Cc: "Xian Liu" <[e-mail]>; "Xujinchun"
> > <[e-mail]>; <[e-mail]>
> > Sent: Tuesday, November 07, 2017 7:53 AM
> > Subject: Re: [netmod] WG Last Call resolutions incorporated in
> > draft-ietf-tictoc-1588v2-yang-06
> > 
> > 
> >> Hi Alex,
> >> 
> >> Sorry for a late reply as I spent the last week for an urgent 
> >> business
> > trip.
> >> Please see my comments in line with [YJ]
> >> 
> >> Thanks,
> >> Yuanlong
> >> 
> >> -----Original Message-----
> >> From: Alex Campbell [mailto:[e-mail]]
> >> Sent: Monday, October 30, 2017 10:15 AM
> >> To: Jiangyuanlong; [e-mail]
> >> Cc: Xian Liu; Xujinchun; [e-mail]
> >> Subject: Re: WG Last Call resolutions incorporated in
> > draft-ietf-tictoc-1588v2-yang-06
> >> 
> >> Hi,
> >> I've reviewed this latest draft and have some more comments.
> >> 
> >> 1. I find the introduction to be unnecessarily wordy; it feels like 
> >> it
> > was written with a view of not missing any information out, rather than trying to keep it concise.
> >>   For example, there is no need to elaborate on YANG data types here.
> > It is also not here to sell YANG.
> >> 
> >> [YJ] Yes, we are trying to give some introductory information for 
> >> an
> > outsider who may not be familiar with PTP or YANG, and explain why a YANG for PTP is needed. The juicy part of this document is its YANG module, and people can skip all the other texts if they are familiar with PTP and YANG.
> >> Besides, these texts have been contributed by multiple sources and
> > undergone several rounds of reviews, thus I will wait for a clear message from the TICTOC chairs to introduce any big changes at this last call stage.
> >> 
> >> 
> >> OLD:
> >> 
> >>   As a synchronization protocol, IEEE 1588-2008 [IEEE1588] is widely
> >>   supported in the carrier networks, industrial networks, automotive
> >>   networks, and many other applications. It can provide high
> >>   precision time synchronization as fine as nano-seconds. The
> >>   protocol depends on a Precision Time Protocol (PTP) engine to
> >>   decide its own state automatically, and a PTP transportation layer
> >>   to carry the PTP timing and various quality messages. The
> >>   configuration parameters and state data sets of IEEE 1588-2008 are
> >>   numerous.
> >> 
> >>   According to the concepts described in [RFC3444], IEEE 1588-2008
> >>   itself provides an information model in its normative
> >>   specifications for the data sets (in IEEE 1588-2008 clause 8). Some
> >>   standardization organizations including the IETF have specified
> >>   data models in MIBs (Management Information Bases) for IEEE 1588-
> >>   2008 data sets (e.g. [RFC8173], [IEEE8021AS]). These MIBs are
> >>   typically focused on retrieval of state data using the Simple
> >>   Network Management Protocol (SNMP), furthermore, configuration of
> >>   PTP data sets is not considered in [RFC8173].
> >> 
> >>   Some service providers and applications require that the management
> >>   of the IEEE 1588-2008 synchronization network be flexible and more
> >>   Internet-based (typically overlaid on their transport networks).
> >>   Software Defined Network (SDN) is another driving factor, which
> >>   demands an improved configuration capability of synchronization
> >>   networks.
> >> 
> >>   YANG [RFC6020] is a data modeling language used to model
> >>   configuration and state data manipulated by network management
> >>   protocols like the Network Configuration Protocol (NETCONF)
> >>   [RFC6241]. A small set of built-in data types are defined in
> >>   [RFC6020], and a collection of common data types are further
> >>   defined in [RFC6991]. Advantages of YANG include Internet based
> >>   configuration capability, validation, rollback and so on. All of
> >>   these characteristics make it attractive to become another
> >>   candidate modeling language for IEEE 1588-2008.
> >> 
> >> NEW:
> >> 
> >>   IEEE 1588-2008 is a time protocol that provides high precision time
> >>   synchronization as fine as nano-seconds.
> >> 
> >>   IEEE 1588-2008 itself provides an information model in its
> > normative
> >>   specifications for the data sets (IEEE 1588-2008 clause 8).
> >>   Standard information models (e.g. [RFC8173], [IEEE8021AS]) have
> > been
> >>   previously defined as MIBs focused on the retrieval of state data
> > using
> >>   SNMP [RFC1157].
> >> 
> >>   YANG [RFC6020] is a data modeling language used to model
> > configuration
> >>   and state data manipulated by network management protocols like
> > NETCONF
> >>   [RFC6241].
> >> 
> >> 2. Can we refer to the system as simply PTP rather than IEEE
> > 1588(-2008)?
> >> [YJ] Advice from IEEE 1588 is, we need to use "1588-2008" as much 
> >> as
> > possible to help clarify that the scope of this YANG is limited to the published 1588 standard.
> >> 
> >> 
> >> 3. There is insufficient spacing here to separate the terms from 
> >> their
> > definitions:
> >> OLD
> >> 
> >>   PTP dataset  Structured attributes of clocks (an OC, BC or TC) used
> >>   for PTP protocol decisions and for providing values for PTP message
> >>   fields, see Section 8 of [IEEE1588].
> >> 
> >>   PTP instance A PTP implementation in the device (i.e., an OC or BC)
> >>   represented by a specific PTP dataset.
> >> 
> >> NEW
> >> 
> >>   PTP dataset
> >>     Structured attributes of clocks (an OC, BC or TC) used
> >>     for PTP protocol decisions and for providing values for PTP
> > message
> >>     fields, see Section 8 of [IEEE1588].
> >> 
> >>   PTP instance
> >>     A PTP implementation in the device (i.e., an OC or BC)
> >>     represented by a specific PTP dataset.
> >> [YJ] OK.
> >> 
> >> 4. There's a singular/plural mismatch here:
> >> 
> >>   module. Query and configuration of device wide or port specific
> >>   configuration information and clock data set is described for this
> >>   version.
> >> [YJ] Good, we will change 'is' to 'are'.
> >> 
> >> and here:
> >> 
> >>   Query and configuration of clock information include:
> >> 
> >> 
> >> 5. The choice of uint16 as instance-number limits implementations 
> >> to
> > 65536 distinct instances.
> >>   While I have a hard time imagining a system with more than 65536
> > PTP instances, I would prefer to avoid imposing arbitrary limits.
> >>   I would recommend changing instance-number to a string (and
> > renaming it to instance-name or just name).
> >> [YJ] The 1588-2008 supports multiple instances of PTP, but it is
> > ambiguous in its organization of those PTP instances, especially with regard to management.
> >> In the 1588 new revision, there is an explicit list of PTP 
> >> instances,
> > and that list is indexed using a number (not name). Thus to align with the new revision, we need to keep it instance-number.
> >> If 65536 limit is a concern, how about change it to uint32, any
> > concerns?
> >> 
> >> 
> >> 6. I still recommend removing -ds from the YANG element names that
> > still include it. It doesn't appear to add any value.
> >> [YJ] Rodney's opinion: the value of using 'ds' is that the 1588
> > document on which this YANG model is based uses "DefaultDS" as a term.
> > PTP experts even say "default dee ess" verbally when referring to this data. If we changed this to just "default", PTP experts might assume that we are referring to something entirely new to YANG. Thus, to align with 1588-2008, the same set of terminologies are used.
> >> 
> >> 7. What;s the relevance of injection attacks relevant to this YANG
> > module?
> >> [YJ] This is a general statement which is applicable to this YANG
> > module and other YANG modules as well.
> >> Thanks again,
> >> Yuanlong
> >> 
> >> Alex
> >> 
> >> 
> >> ________________________________________
> >> From: netmod <[e-mail]> on behalf of Jiangyuanlong
> > <[e-mail]>
> >> Sent: Friday, 27 October 2017 3:21 p.m.
> >> To: [e-mail]
> >> Cc: Xian Liu; Xujinchun; [e-mail]
> >> Subject: [netmod] WG Last Call resolutions incorporated in
> > draft-ietf-tictoc-1588v2-yang-06
> >> 
> >> Dear all,
> >> 
> >> Based on all the comments we received during the WG Last Call 
> >> process,
> > we've updated the document to version 6.
> >> We believe all the LC comments are resolved and the consensus is
> > reflected in this new revision.
> >> Many thanks to Martin, Tal, Opher, Alex, John and many others who 
> >> had
> > reviewed and commented on this draft.
> >> 
> >> Cheers,
> >> Yuanlong on behalf of all coauthors
> >> 
> >> -----Original Message-----
> >> From: [e-mail] [mailto:[e-mail]]
> >> Sent: Friday, October 27, 2017 9:48 AM
> >> To: Xian Liu; Rodney Cummings; [e-mail]; 
> >> Jiangyuanlong;
> > Xujinchun
> >> Subject: New Version Notification for
> > draft-ietf-tictoc-1588v2-yang-06.txt
> >> 
> >> 
> >> A new version of I-D, draft-ietf-tictoc-1588v2-yang-06.txt
> >> has been successfully submitted by Yuanlong Jiang and posted to the
> > IETF repository.
> >> 
> >> Name:           draft-ietf-tictoc-1588v2-yang
> >> Revision:       06
> >> Title:          YANG Data Model for IEEE 1588-2008
> >> Document date:  2017-10-26
> >> Group:          tictoc
> >> Pages:          30
> >> URL:
> > https://urldefense.proofpoint.com/v2/url?u=https-3A__www.ietf.org_in
> > ternet-2Ddrafts_draft-2Dietf-2Dtictoc-2D1588v2-2Dyang-2D06.tx&d=DwIF
> > gQ&c=I_0YwoKy7z5LMTVdyO6YCiE2uzI1jjZZuIPelcSjixA&r=WA71sf2o7Dw7CbYhF
> > t24DPjt3lJuupswWYdnboKbZ8k&m=oQTuhx0E_rtyk18e89Fir82kLxurUC4yXciXZMI
> > qStw&s=N0N5kBPcGMivXWwspHEOc-bP0mbYpkKu2IvM2Asyf_8&e=
> > t
> >> Status:
> > https://urldefense.proofpoint.com/v2/url?u=https-3A__datatracker.iet
> > f.org_doc_draft-2Dietf-2Dtictoc-2D1588v2-2Dyang_&d=DwIFgQ&c=I_0YwoKy
> > 7z5LMTVdyO6YCiE2uzI1jjZZuIPelcSjixA&r=WA71sf2o7Dw7CbYhFt24DPjt3lJuup
> > swWYdnboKbZ8k&m=oQTuhx0E_rtyk18e89Fir82kLxurUC4yXciXZMIqStw&s=aYlovx
> > _kTQtAiJAUMTJn8NCRZQIi_jEVNa-tC_5HFlk&e=
> >> Htmlized:
> > https://urldefense.proofpoint.com/v2/url?u=https-3A__tools.ietf.org_
> > html_draft-2Dietf-2Dtictoc-2D1588v2-2Dyang-2D06&d=DwIFgQ&c=I_0YwoKy7
> > z5LMTVdyO6YCiE2uzI1jjZZuIPelcSjixA&r=WA71sf2o7Dw7CbYhFt24DPjt3lJuups
> > wWYdnboKbZ8k&m=oQTuhx0E_rtyk18e89Fir82kLxurUC4yXciXZMIqStw&s=j1aDjiU
> > 7AeuEV-p20PNwNpvZPrGSbBiHLHta7q05pak&e=
> >> Htmlized:
> > https://urldefense.proofpoint.com/v2/url?u=https-3A__datatracker.iet
> > f.org_doc_html_draft-2Dietf-2Dtictoc-2D1588v2-2Dyang-2D06&d=DwIFgQ&c
> > =I_0YwoKy7z5LMTVdyO6YCiE2uzI1jjZZuIPelcSjixA&r=WA71sf2o7Dw7CbYhFt24D
> > Pjt3lJuupswWYdnboKbZ8k&m=oQTuhx0E_rtyk18e89Fir82kLxurUC4yXciXZMIqStw
> > &s=7tYOv1M_EYHCPG1MiOq3BVl7vpB0w-LSiDYcQHM4ayM&e=
> >> Diff:
> > https://urldefense.proofpoint.com/v2/url?u=https-3A__www.ietf.org_rf
> > cdiff-3Furl2-3Ddraft-2Dietf-2Dtictoc-2D1588v2-2Dyang-2D06&d=DwIFgQ&c
> > =I_0YwoKy7z5LMTVdyO6YCiE2uzI1jjZZuIPelcSjixA&r=WA71sf2o7Dw7CbYhFt24D
> > Pjt3lJuupswWYdnboKbZ8k&m=oQTuhx0E_rtyk18e89Fir82kLxurUC4yXciXZMIqStw
> > &s=Z12Xm_2k7cEAlV7lFQb_zw7A-D-HL77C-Kuy2BgyCHA&e=
> >> 
> >> Abstract:
> >>   This document defines a YANG data model for the configuration of
> >>   IEEE 1588-2008 devices and clocks, and also retrieval of the
> >>   configuration information, data set and running states of IEEE
> >>   1588-2008 clocks.
> >> 
> >> 
> >> 
> >> 
> >> Please note that it may take a couple of minutes from the time of
> > submission until the htmlized version and diff are available at tools.ietf.org.
> >> 
> >> The IETF Secretariat
> >> 
> >> _______________________________________________
> >> netmod mailing list
> >> [e-mail]
> >> https://urldefense.proofpoint.com/v2/url?u=https-3A__www.ietf.org_m
> >> ailman_listinfo_netmod&d=DwIFgQ&c=I_0YwoKy7z5LMTVdyO6YCiE2uzI1jjZZu
> >> IPelcSjixA&r=WA71sf2o7Dw7CbYhFt24DPjt3lJuupswWYdnboKbZ8k&m=oQTuhx0E
> >> _rtyk18e89Fir82kLxurUC4yXciXZMIqStw&s=cUl0d0wDX9fIwjIEHlhcrfg17x7ph
> >> pI9QiF5PuXsYd4&e=
> >> 
> >> _______________________________________________
> >> netmod mailing list
> >> [e-mail]
> >> https://urldefense.proofpoint.com/v2/url?u=https-3A__www.ietf.org_m
> >> ailman_listinfo_netmod&d=DwIFgQ&c=I_0YwoKy7z5LMTVdyO6YCiE2uzI1jjZZu
> >> IPelcSjixA&r=WA71sf2o7Dw7CbYhFt24DPjt3lJuupswWYdnboKbZ8k&m=oQTuhx0E
> >> _rtyk18e89Fir82kLxurUC4yXciXZMIqStw&s=cUl0d0wDX9fIwjIEHlhcrfg17x7ph
> >> pI9QiF5PuXsYd4&e=
> > 
> > _______________________________________________
> > TICTOC mailing list
> > [e-mail]
> > https://www.ietf.org/mailman/listinfo/tictoc
> 
> _______________________________________________
> netmod mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/netmod

-- 
Juergen Schoenwaelder           Jacobs University Bremen gGmbH
Phone: [phone]         Campus Ring 1 | 28759 Bremen | Germany
Fax:   [phone]         <http://www.jacobs-university.de/>

_______________________________________________
TICTOC mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/tictoc