- db-wg - mailman.ripe.net

Re: [db-wg] RIPE Policy Proposal 2017-02 Validates Database Attributes
by Sascha Luck [ml] 10 Sep '17

10 Sep '17

All, I will discuss this here as I do not accept the Anti-Abuse WG as a forum for this proposal. For one thing, this proposal affects every ripedb user - in fact, as this entails changes to how the NCC provides services, the services-wg would be an even better venue. For another, given the "population" and culture of "debate" on the AAWG, any "consensus" reached there would be so worthless as to be farcical. (If anyone wants amplification on this, https://www.ripe.net/ripe/mail/archives/anti-abuse-wg/ provides ample evidence. To the meat of the proposal: Firstly, this proposal conflicts somewhat with NWI-7 in this very WG. (Another reason why it should have gone here in the first place) With the upcoming possibility of delegating abuse-c: to (customers') resource objects, who bears any consequences of these customers not replying to this proposed email? >Rationale >a. Arguments supporting the proposal >Accurate and validated information in the RIPE Database is >essential to establish a trusted and transparent environment in >which all network operators can operate safely. abuse-c: is a contact object, just as admin-c: or tech-c: and (correct me if I'm wrong) audited at the same time as the rest of the contact information. What makes it so unique that this verification is not enough? >Accurate and >validated information helps Internet troubleshooting at all >levels, >but it also helps to attribute malicious online >activities that undermine this trusted environment. See above. >The lack of reliable accurate and validated information in the >database negatively impacts legitimate uses of the RIPE Database, >including: An *email adress* that doesn't reply once a year does NOT equate to a "lack of reliable accurate and validated information". I find this statement somewhat insulting to the NCC team who do make the effort to keep the ripedb data accurate and do audit resource holders. There is an issue with the reliability of out-of-region and legacy resource data but as the NCC has no "enforcement" powers over these resource holders, in these cases this proposal snatches at thin air. >Assuring the security and reliability of the network by >identifying points of contact for IP addresses for network >operators, ISPs, and certified computer incident response teams; "org:", "admin-c:", "tech-c:", "mnt-by:" and, yes, "abuse-c:" exist. >Ensuring that IP address holders are accountable, so individuals, >consumers and the public are empowered to resolve abusive >practices that impact safety and security; >Assisting businesses, consumer groups, healthcare organisations >and other organisations that are combating fraud (some of which >have mandates to electronically save records) to comply with >relevant legal and public safety safeguards; The contact object that does (or *should*) stand for the person(s) who can speak for a LIR, legally, is "admin-c:". "abuse-c:" is some role account in a NOC or even a ticket system unlikely to have any decision-making power. An attempt to make these roles (perhaps even personally) responsible for the behaviour of a LIR and its customers is counter-productive. I for one would flatly refuse to do any abuse report handling under these circumstances. >Complying with national, civil and criminal due process laws in >support of investigations and providing justice for victims. Would the proposers please amplify exactly which law or due process is violated by the NCC not sending an email once a year to an abuse-c:? Myself, and I'm sure the NCC legal team, would be interested to know. On a technical note, email is neither a secure nor a reliable transport for such verification. In the day of blocklists and large email providers imposing arbitrary restrictions on email senders it is not guaranteed that a verification email reaches the intended address or that a reply reaches the sender. The NCC ARC procedure, however employs both email and personal contact via telephone to verify the accuracy of the ripedb information. In toto, this proposal would impose unneccessary work on LIRs and the NCC, using unsuitable means to rectify a non-existing issue, and I therefore oppose it.

4 3

Re: [db-wg] RIPE Policy Proposal 2017-02 Validates Database Attributes
by William Sylvester 08 Sep '17

08 Sep '17

DB WG Members, We understand there are many opinions on this proposal. We encourage everyone to discuss this request on the anti-abuse mailing list. We have been coordinated with the AA-WG chairs on this and would appreciate expressing your support or objections in the AA-WG as that is the forum this proposal was raised. If you have any questions or concerns please contact the db-wg chairs. Thanks, William, Denis, and David Database WG Chairs On Sep 7, 2017, at 1:05 PM, Marco Schmidt via db-wg <db-wg(a)ripe.net<mailto:db-wg@ripe.net>> wrote: <mime-attachment>

1 0

RIPE Policy Proposal 2017-02 Validates Database Attributes
by Marco Schmidt 08 Sep '17

08 Sep '17

Dear colleagues, We just published RIPE Policy proposal, 2017-02, "Regular abuse-c Validation", to the Anti-Abuse Working Group mailing list. The goal of this proposal is to give the RIPE NCC a mandate to regularly validate "abuse-c:" information, and to follow up in cases where contact information is found to be invalid. As this proposal relates to contact information put into the RIPE Database, we would like to notify the Database Working Group specifically. You can find the full proposal at: https://www.ripe.net/participate/policies/proposals/2017-02 We encourage you to review this proposal and send your comments to <anti-abuse-wg(a)ripe.net> before 6 October 2017. Regards, Marco Schmidt Policy Development Officer RIPE NCC Sent via RIPE Forum -- https://www.ripe.net/participate/mail/forum

2 2

Re: Last Call for Agenda Items DB-WG Meeting - RIPE75 - Dubai, UAE
by William Sylvester 01 Sep '17

01 Sep '17

Dear DB-WG Members, This is the last call for agenda items for the DB-WG Meeting at RIPE75 at the Conrad Dubai in Dubai, UAE. Due to restrictions by UAE law we will not be able to add last minute presentations. Please submit your proposals to the working group chairs as soon as possible not later than September 6, 2017. Please submit agenda items or propose additional topics to be discussed both on the mailing list or direct to the WG Chairs at db-wg-chairs(a)ripe.net<mailto:db-wg-chairs@ripe.net> The DB-WG Meeting is scheduled for Wednesday, October 25th, starting at 14:00 local time, after lunch and before the coffee break. The length of our timeslot is 90 minutes. For the overall RIPE 74 meeting plan please refer to its most up-to-date version at https://ripe75.ripe.net/programme/meeting-plan/ Note: due to UAE law, confirmed speakers will need to submit a copy of their passport and contact information in advance of the meeting. More details are available here: https://ripe75.ripe.net/attend/dctm-requirements We are looking forward to another successful meeting in Dubai. Kind regards, William, Denis, & David DB WG Chairs

1 0

Whois Master Database Patch
by Edward Shryane 31 Aug '17

31 Aug '17

Dear working group, Tomorrow afternoon, Friday 1 September, we will apply a patch to the Whois Master database server. This will cause a short downtime for Whois updates. Queries will not be affected. Regards, Ed Shryane RIPE NCC

1 0

Re: [db-wg] NWI-7: abuse-c implementation plan
by Tim Bruijnzeels 30 Aug '17

30 Aug '17

Dear working group, We have a proposed timeline for the implementation. Let me quote the proposal sent out on 28 June here, and comment in-line: > 1) Allow “abuse-c:” on INET(6)NUM and AUT-NUM objects > > We will deploy a new version of whois where the “abuse-c:” attribute will be allowed to exist on INET(6)NUM and AUT-NUM objects. The syntax will be the same as for the ORGANISATION object: > abuse-c: [optional] [single] [inverse key] > > The Abuse Finder logic will be updated to use the first "abuse-c:" found in this order of priority: > -the local "abuse-c:" in the resource object > -the "abuse-c:" in an ORGANISATION object referenced by the "org:" attribute in the resource object > -search up the resource object hierarchy (less specific objects) > --the "abuse-c:" in the less specific resource object > --the "abuse-c:" in an ORGANISATION object referenced by the "org:" attribute in the less specific resource object > > The search will stop when the top level resource object is reached. > > In addition to this we plan to introduce a business rule that will disallow “abuse-mailbox:” to be modified, or (re-)added to any object, other than ROLE objects. This will enable us to do a phased clean-up of this data without invalidating objects. We plan to have a whois release 1.90 for this (and the updated use of the ‘pn’ keyword mentioned under item #5 below) in the Release Candidate environment no later than Monday 2 October. We plan to deploy this version to production on Monday 16 October. > 2) Clean up of “abuse-mailbox:” on ORGANISATION objects > > Three cases exist: > > A) “abuse-mailbox:” is present and it is the same as the “abuse-mailbox:” in the also present “abuse-c:” reference (8919 cases) > > In this case we propose to simply delete the “abuse-mailbox:” attribute and inform the holder. We will work on a clear text for this email, but the main message would be: > "We removed “abuse-mailbox:” from your ORGANISATION object because this attribute is deprecated as the final step in the deployment of the "abuse-c:" attribute. However, we see that you were already using the preferred way to document abuse contact details for your organisation and you are referencing an “abuse-c:” ROLE object with the same email address in its its “abuse-mailbox:” attribute. No action is required.” > > > B) “abuse-mailbox:” is present and it is different from the “abuse-mailbox:” in the also present “abuse-c:” reference (1848 cases) > > Note that some of these cases have been caused by the fact that since the introduction of “abuse-c:” LIRs can no longer modify the value of the “abuse-mailbox:” attribute. The intent was that “abuse-mailbox:” would have been cleaned up earlier, but unfortunately that dit not happen. That being said “abuse-c:” has been kept up to date by the LIR, and out of the two is also the one that is being shown on the Abuse Finder. > > Therefore we propose to delete the “abuse-mailbox:” attribute in these cases, and send a different notification to the holder with the following main message: > > "We removed “abuse-mailbox:” with value $email from your ORGANISATION object because this attribute is deprecated as the final step in the deployment of the "abuse-c:" attribute. However, we see that you were already using the preferred way to document abuse contact details for your organisation and you are referencing an “abuse-c:” ROLE object with that uses $email_2 in its “abuse-mailbox:” attribute. If the latter is correct no action is required. However if you want to modify the email address in your “abuse-c:” ROLE object, you can do so here (link to webupdates).” > > > C) “abuse-mailbox:” is present, and there is no “abuse-c:” reference (28312 cases) > > We *could* create “abuse-c:” ROLE objects for these cases, using the same maintainers as the ORGANISATION object. However, the “abuse-mailbox:” attribute has not been shown in the Abuse Finder, and this change could lead to wrong or out-of-date addresses suddenly receiving abuse reports. Therefore we suggest that we delete the attribute instead, and send the following main message to the holder of the ORGANISATION object: > > "We removed “abuse-mailbox:” with value $email from your ORGANISATION object because this attribute is deprecated as the final step in the deployment of the "abuse-c:" attribute. However, if you wish to add an email address for abuse reports you can do so by editing your object here (link to web updates for the ORGANISATION object)." > > We plan to perform these clean-ups on Monday 30 October. > 3) Clean up of “abuse-mailbox:” on PERSON (89k cases), MNTNER (2500 cases) and IRT (238 cases) > > In these cases we propose to send a warning email to the holders of these objects with the following main message: > > “The “abuse-mailbox:” is deprecated on these object and will be removed in 4 weeks, if you want to set up abuse contact information on your resource objects you can do so by having a default “abuse-c:” on your ORGANISATION referenced by your resource objects, or possibly overriding that default with a specific “abuse-c:” reference on applicable resource objects." > > And then 4 weeks later we will preform the clean-up and refer back to this communication. On further reflection we would prefer to alter the plan, and just a single clean-up of these objects on Monday 30 October as well. This way the clean-up process will be consistent across all object types. Furthermore, there is no action to be taken on these objects and “abuse-mailbox:” on these objects is ignored in the Abuse Finder. In short there does not seem to be a strong motivation for bothering the holders of these objects twice. > 4) Clean up whois code > > Now that “abuse-mailbox:” only appears where it should, we will be able to update the whois code itself so that “abuse-mailbox:” is no longer syntactically allowed on objects other than ROLE objects. This will allow us to remove the business rule added in phase 1 and simplify the whois code and maintenance. We will also update the RIPE Database Documentation to reflect this. We plan to do this as part of a future whois 1.91 release, as yet unplanned. This clean-up will not change any publicly visible functionality. > > > 5) Inverse queries > > As requested we can change the behaviour of the “pn” short cut to also search for “abuse-c:” in inverse queries. I.e. one can use: > whois -r -i pn uk41-ripe > > Instead of: > whois -r -i abuse-c,admin-c,tech-c,zone-c,author uk41-ripe As mentioned above, we plan to include this in the whois 1.90 release. Kind regards Tim Bruijnzeels Assistant Manager Software Engineering RIPE NCC Database Group > On 4 Aug 2017, at 14:39, William Sylvester via db-wg <db-wg(a)ripe.net> wrote: > > > From: William Sylvester <william.sylvester(a)addrex.net> > Subject: NWI-7: abuse-c implementation plan > Date: 3 August 2017 at 21:12:06 GMT+2 > To: "db-wg(a)ripe.net" <db-wg(a)ripe.net> > > > Working group members, > > Based on the responses we have received for NWI-7 : Abuse-c implementation, there is a clear consensus for the proposed implementation that has been presented by Tim from RIPE NCC as written for improving "abuse-c:". > > At this time, the DB-WG co-chairs would like to ask the RIPE NCC to proceed with this implementation. > > Kind Regards, > > DB-WG Chairs > > > > >

1 0

Re: [db-wg] out of region routing in the RIPE Database
by Clement Cavadore 14 Aug '17

14 Aug '17

On Tue, 2017-07-18 at 15:26 +0200, Nick Hilliard via db-wg wrote: > > I am not in favour of having the RIPE database as an open-access > database on the basis that this mixes up two sets of data, > authoritative > and non-authoritative, and it it is impossible for someone casually > querying the database to determine which is which. > > Some people are inserting random route: objects into the database, and > those route: objects are being picked up by provisioning systems and > ending up configured on routers and IXP route servers. This enables > prefix hijacking, which is a pressing operational issue. I agree with Nick's position. It legitimates what seems to be rogue announcements, like for example 196.16.0.0/14, as mentionned recently on the NANOG mailing list (*). We should, IMHO not be able to insert out of region route(6) object without having a prior authentication mechanism, or making it be specially flagged, so the auto ACL system from upstreams wouldnt match it. (*) https://mailman.nanog.org/pipermail/nanog/2017-August/091954.html -- Clément Cavadore

1 0

Re: [db-wg] out of region routing in the RIPE Database
by Daniel Shaw 11 Aug '17

11 Aug '17

On 11/08/2017, 05:14, Randy Bush via db-wg typed: > >> >> databases. It's that they never have, and it's just the "way things >> have been done". It's always ended up that they *can* support multiple >> databases if asked to. > > when negotiating with a prospective peer, i would rather not add "you > will have to change operational practice and code for this to work." That is a good point. I would also rather not, generally. And that is the basis of my original fears around this something like a year ago. What I found in practise (so far) has been that when stated not as "you will *have* to change operations", but as "would it be possible for you to change operations for me?", the answer has usually been "Ok". I have to be honest, that was a surprise. - Daniel

1 0

Re: [db-wg] out of region routing in the RIPE Database
by Randy Bush 11 Aug '17

11 Aug '17

> That has proven to never really be an issue. When a network upstream > of another network says they don't support additional databases for > IRR, so far that has always proven to actually mean they've never > needed to think about supporting multiple databases, and never had the > demand to do the work to change that. > > In other words, it's not that they *can't* support other > databases. It's that they never have, and it's just the "way things > have been done". It's always ended up that they *can* support multiple > databases if asked to. when negotiating with a prospective peer, i would rather not add "you will have to change operational practice and code for this to work." randy

1 0

Re: [db-wg] out of region routing in the RIPE Database
by Daniel Shaw 10 Aug '17

10 Aug '17

Hi there, On 09/08/2017, 19:34, Stefan Ideler via db-wg typed: > > > As a result our route objects regardless of origin are noted in the RIPE database. Our upstreams use it to generate their prefix filters (and not all of them support multiple databases) I won't comment here on your specific use case or the overall reasoning or motivation for wanting to use the RIPE DB exclusively, even for resources that may come from various other regions. I have an opinion, but it's just that. An opinion. What I'd like to comment on is the excerpt above. Specifically about upstream providers not supporting multiple databases. That has proven to never really be an issue. When a network upstream of another network says they don't support additional databases for IRR, so far that has always proven to actually mean they've never needed to think about supporting multiple databases, and never had the demand to do the work to change that. In other words, it's not that they *can't* support other databases. It's that they never have, and it's just the "way things have been done". It's always ended up that they *can* support multiple databases if asked to. I speak from experience here. If you go back into the archives of this list, when this issue was first brought to the list, I responded with this exact fear: But what about the upstream support? Since then I've had the opportunity to speak to folks from some of the larger networks that are often someone's upstream on behalf of AFRINIC members from time to time, and also in my own capacity in the context of an RIR meeting event network. We've also encouraged member operators to just ask their upstreams to change. And this has proved to never have remained an issue to the best of my knowledge. Summary: While you may (or may not) have other good reasons for the RIPE DB to be open (or not), upstream support of alternates is not likely one of the reasons. Best regards, Daniel

1 0