June 2017 - db-wg - mailman.ripe.net

<auto-dbm@ripe.net> is slow this afternoon.
by RIPE Database Administration 16 May '18

16 May '18

Dear Colleagues, Because of a very high level of use this afternoon, the response time of <auto-dbm(a)ripe.net> is longer than normal. I have made contact with those who are responsible and I expect the situation to improve as the machine works through the queue. Please be patient. Obviously we at the RIPE NCC Database Group are continuously working to make the performance of the database as good as possible. However, we cannot make improvements in zero time. We thank you for your co-operation up to now and we ask for your continued co-operation so that everyone receives the best possible service. If you have anymore questions, please contact <ripe-dbm(a)ripe.net>. Regards, Ambrose Magee ____________________________ RIPE Database Administration.

2 2

Re: [db-wg] NWI-7: abuse-c implementation request
by Tim Bruijnzeels 28 Jun '17

28 Jun '17

Dear working group, Here is the RIPE NCC's impact analysis and proposed implementation plan for the changes to “abuse-c:” that the working group seeks. We expect that this will have a MEDIUM impact on the RIPE NCC. The RIPE Database Group Software team will have to implement changes, but because a clean-up of “abuse-mailbox:” is also in-scope of this proposal we also expect impact on our Customer Service department due to questions from users of the RIPE Database. We propose an implementation using the following phases: 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. 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)." 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. 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. 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 Kind regards Tim Bruijnzeels Assistant Manager Software Engineering RIPE NCC Database Group > On 22 May 2017, at 20:56, 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 request > Date: 22 May 2017 at 19:59:44 GMT+2 > To: DB-WG <db-wg(a)ripe.net> > > > DB-WG Members, > > Support was shown for the proposal NWI-7 and no objections were raised after this round of discussion. At this time, the chairs request that the RIPE NCC schedule implementation of NWI-7 as described. > > This request is to add the "abuse-c:" attribute directly to the INET(6)NUM and AUT-NUM resource objects. The db-wg therefore ask the RIPE NCC to prepare an impact analysis, followed by an implementation plan and timeline for this request and the other issues raised in the problem statement of NWI-7 as follows: > > 1) Add the "abuse-c:" attribute to the INETNUM, INET6NUM and AUT-NUM objects > 2) Adjust the query logic to take account of 'local' "abuse-c:" attributes in preference to those referenced in ORGANISATION objects > 3) Allow both local and indirect references (via ORGANISATION objects) to "abuse-c:" attributes to operate hierarchically > 4) Deprecate "abuse-mailbox:" attribute from PERSON, ORGANISATION, MNTNER and IRT objects. > 5) Clean up data relating to the deprecated attributes. > 6) Update the 'person' keyword to take account of "abuse-c:" > > Thank you all for your work on this proposal! > > Kind regards, > > William > on behalf of DB-WG co-chairs > > >

1 0

Re: [db-wg] 2017 and RIPEdb only handles 700 updates per *MINUTE*
by Tim Bruijnzeels 21 Jun '17

21 Jun '17

Dear Working Group, The average update rate to the RIPE Database is 12 updates per minute. As shown the system is capable of handling a sustained rate of over 700 updates per minute for inetnums through mail updates without issues. That is 60 times the average rate. Cases where people actually need to do updates exceeding this rate are very few and far in between. This does not warrant allocating resources to improve this further - our efforts are better spent in other areas such as improving the usability of the user interface to help people manage their content better. Kind regards, Tim Bruijnzeels Assistant Manager Software Engineering RIPE NCC Database Group

1 0

Re: [db-wg] 2017 and RIPEdb only handles 700 updates per *MINUTE*
by Horváth Ágoston János 20 Jun '17

20 Jun '17

Hi Jeroen, I'm probably the most relevant person to asnwer your question right now, having been a core developer during the ripe database reimplementation project in 2013. We did that by running side by side with the old, C version of the ripe database software. As such, we were in no position to change the underlying database or its format. And guess what, the underlying database is coming from the previous server's reimplementation project, all the way from 2000. Reminds me of the story of the NASA booster rocket and the roman horse. There have been minor updates done over the years (like switching to innodb and sql transactions), the basics are still the same. Its data model is 100% optimized for query performance (since queries outnumber updates 100 to 1 or more). And most importantly, since updates (can and often do) have effect on one another, they are strictly serialized (as in, single-threaded). On top of that, all updates go through a rigorous rule checks before being accepted and taking effect. All this translates to that update performance. Note that the update performance is much higher for simple object types (say, person; my last knowledge was over 2000 objects/min), and very slow for complex ones (namely aut-num, with all the extra DNS checks). In practice, it was never an issue that a longer update ran for hours. This performance could be improved greatly given enough priority though. It was always a simple case of this requiring quite some dirty hack or major refactors, while this issue was always considered a minor nuisance, not enough to warrant either. BTW, if you look at the update graph you included, you see that updates are normally almost negligible compared to your batch. Thus most users of ripe database updates see very quick response times. I'd maintain that unless this really stops you from achieving your goals, it's probably best to leave the engineers work on deeper / more pressing issues. Like the fact that the RIPE DB actually needs a massive documentation and a day long training course even for basic usage. And even then, people are struggling with the weird object hierarchy and rules. While could be considered an acceptable approach in '99, this is way outdated by today's standards and user expectations. Cheers, Agoston On Tue, Jun 20, 2017 at 2:23 PM, Jeroen Massar via db-wg <db-wg(a)ripe.net> wrote: > > > ---------- Forwarded message ---------- > From: Jeroen Massar <jeroen(a)massar.ch> > To: db-wg(a)ripe.net > Cc: > Bcc: > Date: Tue, 20 Jun 2017 14:22:49 +0200 > Subject: 2017 and RIPEdb only handles 700 updates per *MINUTE* > https://www.ripe.net/analyse/statistics/ripe-database/ripedb-updates > > (see attached png for the current version as over time it will disappear) > > shows a nice peak with a max of 723.47 updates. > > Thus either the software is really bad, the updates are rate limited, or > the queries are so high that no further performance is to be expected. > > > As alluded to in the other thread, if there was less junk (those 400k > inet6nums for instance) then likely this would all scale better.... > > Are the RIPE DB-WG chairs still actively involved in the maintenance of > the RIPE Database!? > > As two of the co-chairs are working for RIPE NCC, are they able to tell > why such a process is slow and why these problems that are being raised > over and over on the various lists are not being resolved? > > Greets, > Jeroen >

1 0

Re: [db-wg] 2017 and RIPEdb only handles 700 updates per *MINUTE*
by denis walker 20 Jun '17

20 Jun '17

HI Jeroen The three DB-WG co-chairs (2 of whom are former RIPE NCC staff) are still actively involved in overseeing the maintenance of the RIPE Database. We don't have any 'hands on' involvement. Things may have changed a little since I was working on the RIPE Database so I am sure Tim will correct me if I am wrong here. Updates sent by email are queued and processed sequentially. It is a batch process, not real time as with Syncupdates or Webupdates. If you submit a single email with a huge number of objects, each of those objects are processed one at a time sequentially. If you submit several emails in quick succession with that huge list of updates spread over them, the emails will be queued and processed sequentially. Not sure now if there is only one queue or several for email updates. AFAIK the database can only update a single object at a time, no matter how many processors are running to feed updates to the database. The bottom line is, if you submit a huge number of objects to be updated it will take some time to complete. The number of daily queries massively outweighs the number of updates. So performance is naturally optimised to serve the query load. The average number of updates is easily handled, but when you get spikes it does slow down the update process for a while. The number of objects in the database makes no difference to performance on either queries or updates. So a large amount of junk will not affect performance. To address a couple of your points from your previous email. No one has ever defined what the "country:" attribute in an INET(6)NUM object means. It could be the location of the registries head office, their data centre, their service desk, their user base... It is not uncommon to see this value 'apparently' conflict with netname or address. If you believe 40% of the INET6NUM objects in the RIPE Database are junk, that is something the RIPE NCC Registration Services needs to comment on. cheersdenisco-chair DB-WG From: Jeroen Massar via db-wg <db-wg(a)ripe.net> To: db-wg(a)ripe.net Sent: Tuesday, 20 June 2017, 14:23 Subject: [db-wg] 2017 and RIPEdb only handles 700 updates per *MINUTE* https://www.ripe.net/analyse/statistics/ripe-database/ripedb-updates (see attached png for the current version as over time it will disappear) shows a nice peak with a max of 723.47 updates. Thus either the software is really bad, the updates are rate limited, or the queries are so high that no further performance is to be expected. As alluded to in the other thread, if there was less junk (those 400k inet6nums for instance) then likely this would all scale better.... Are the RIPE DB-WG chairs still actively involved in the maintenance of the RIPE Database!? As two of the co-chairs are working for RIPE NCC, are they able to tell why such a process is slow and why these problems that are being raised over and over on the various lists are not being resolved? Greets, Jeroen

1 0

2017 and RIPEdb only handles 700 updates per *MINUTE*
by Jeroen Massar 20 Jun '17

20 Jun '17

https://www.ripe.net/analyse/statistics/ripe-database/ripedb-updates (see attached png for the current version as over time it will disappear) shows a nice peak with a max of 723.47 updates. Thus either the software is really bad, the updates are rate limited, or the queries are so high that no further performance is to be expected. As alluded to in the other thread, if there was less junk (those 400k inet6nums for instance) then likely this would all scale better.... Are the RIPE DB-WG chairs still actively involved in the maintenance of the RIPE Database!? As two of the co-chairs are working for RIPE NCC, are they able to tell why such a process is slow and why these problems that are being raised over and over on the various lists are not being resolved? Greets, Jeroen

2 1

Re: [db-wg] Cleaning up SixXS inet6nums & 400.000++ junk inet6nums from 2a07:7ec0::/29 in the RIPE database
by Ronald F. Guilmette 19 Jun '17

19 Jun '17

In message <mailman.1675.1497881039.4885.db-wg(a)ripe.net>, Jeroen Massar <jeroen(a)massar.ch> wrote: >Really, a Dutch address, claiming a British country, while the prefix >clearly show "NL" as prefix. > >Website at http://www.onlineserviceproviderbv.nl/ is empty... You may have a bit more luck if you try this: http://onlineserviceproviderbv.nl/ But anyway, it still does seem quite odd that the IPv6 space of this "Dutch" company is apparently being routed by a company located in India whose main stock in trade is the manufacturing of portable searchlights. http://bgp.he.net/AS58905#_prefixes6 --> adwiser.co --> "Point Blank Range Pvt Ltd" --> https://www.indiamart.com/pointblank-rangeprivate-limited/ But it is all rather humorous, don't you think? (I personally prefer my Internet criminals to be ones with a sense of humor. It makes the process of hunting them down at least somewhat entertaining.) And if you would like to see even -more- humor, try this google search: "+31 20 218 2999" It would seem that your "Dutch Internet Service Provider" has another business which is even more notable (and even more linked to, going by the google results ordering), i.e. they also are apparently selling some sort of service that allegedly will help us all to MAKE MONEY FAST while gambling on sports games. https://oddstreak.com/terms-of-service https://oddstreak.com/ Terrific! I just can't wait to make my fortune betting on futbol and/or cricket matches, full time, and with every last penny I have. These people vitrually guarrantee that if I just put in the time and effort, and if I just give them a little bit of money every month, I'll soon be wealthy beyond my wildest dreams. I confess that it does seem just slightly odd that these people who are behind this commercial enterprise, having invested all of the time and energy needed to build a sophisticated mathematical system that will allow any ordinary man on the street (like me) to beat the odds and to come out a winner, would not have applied this system -themselves-, you know, to help their own betting. I mean gee! They have this sophisticated betting system which will allow you to win all the time, so why aren't they themselves presently lounging on a beach somewhere without a care in the world? I guess they are just too generous, and they felt obliged to share their magical mathematical betting strategies with even poor little sods like me. God bless them! covfefe

1 0

Re: [db-wg] Cleaning up SixXS inet6nums & 400.000++ junk inet6nums from 2a07:7ec0::/29 in the RIPE database
by Ronald F. Guilmette 19 Jun '17

19 Jun '17

In message <mailman.1675.1497881039.4885.db-wg(a)ripe.net>, Jeroen Massar <jeroen(a)massar.ch> wrote: >Thus the big question: why does the RIPE NCC allow that junk to exist >there? (Noting that it was shown last year that that "organisation" >inserted bogus stuff too... and it has not been resolved) That is indeed the Big Question. What is the answer? I am seriously asking. I also am aware of at least one very large clump of IPv4-related "pollution" within the RIPE data base... about which I will say more soon (although as far as I can tell, nobody on -any- of the RIPE mailing lists gives a damn about such things)... and I also wonder why no one seems to care about all of the blatantly bogus pollution in the RIPE data base. I can only guess that the answer to that probably has a lot to do with money, one way or another. covfefe

1 0

Cleaning up SixXS inet6nums & 400.000++ junk inet6nums from 2a07:7ec0::/29 in the RIPE database
by Jeroen Massar 19 Jun '17

19 Jun '17

TLDR: - SixXS inet6nums will be cleansed out from RIPE DB - Somebody has 400k junk inet6nums (40% of total) in there... Hola, In regards to https://www.sixxs.net/sunset/, a few hours ago I let our little robot do one last chore: remove all SixXS inet6num's from the RIPEdb as those prefixes are now unused and if any abuse claims where outstanding (we didn't receive any over the last multiple years actually) then those folks had enough time to whois them... Normally, that robot was responsible for ensuring that all tunnels & subnets where properly registered in the database and also removed when they where not used (marked deleted) by that user for two weeks. I now instructed it to consider all prefixes (tunnels and subnets) as deleted and then let it batch out the changes (the RIPE mailserver does not like very large mails it seems...). It is already taking a few hours, but it seems that the mails that where sent in a period of 9 minutes, are being processed slowly, as responses of the updates are coming in. I am looking forward to the graph updating: https://www.ripe.net/analyse/statistics/ripe-database/ripedb-updates to see how much impact it had, as it seems that it is very slow... # Some stats Some stats, the robot takes: ftp://ftp.ripe.net/ripe/dbase/split/ripe.db.inet6num.gz $ cat ripe.db.inet6num |grep "^inet6num:" | wc -l 1000583 (that number is way too high, see below for why) 13 years ago in 2004 I wrote: https://www.ripe.net/ripe/mail/archives/db-wg/2004-May/002728.html "Thus SixXS takes care of 53.98% of the inet6num's" when we had 2898 prefixes ;) This time the robot found 70581 prefixes to clean (quite a few have been wiped already in the last months btw, thus that was not the max number, just the ones still left ;) Tomorrow all these SixXS prefixes should be gone. According to that number, SixXS would be 7% of the prefixes, that is rather quite a different number, thus I wondered, why is that. # Why are there now 1 million inet6nums? Junk! Thus lets look at the prefixes with more than 10k inet6nums registered: $ cat ripe.db.inet6num.clean | grep "^inet6num:" | awk '{print $2}' | cut -f1,2 -d: | sort -n |uniq -c | sort -n ... 10575 2001:6f8 <---- Easynet, mostly SixXS prefixes 13675 2001:4dd0 <---- Netcologne, mostly SixXS 14038 2001:41f0 <---- Ziggo, wow! now to actually enable users... 33982 2a02:1630 <---- AS30848 "TWT Spa" in .it nice! 37372 2a00:2381 <---- British Telecom nice! 61500 2a07:7ec4 <---- vv junk registrations 61501 2a07:7ec2 64575 2a07:7ec1 64575 2a07:7ec5 64577 2a07:7ec0 65600 2a07:7ec3 65600 2a07:7ec6 65600 2a07:7ec7 <---- ^^ junk registrations 210219 2a01:4f8 <---- Hetzner (makes sense to me...) So apparently, somebody who is not a LIR and who was last year caught with inserting bogus entries: https://www.ripe.net/ripe/mail/archives/db-wg//2016-May/005264.html currently has !!!!~400k!!!! inet6nums in there, that is ~40% of the total!!!! I picked a random few, most of them do not appear to actually be routed. If you look at the contents, you also notice that the details are bogus. Thus the big question: why does the RIPE NCC allow that junk to exist there? (Noting that it was shown last year that that "organisation" inserted bogus stuff too... and it has not been resolved) On top of that those 400k++ inet6nums all show bogus geo location (and are generally useless as they do not contain info at all), lets look at what registers that: inet6num: 2a07:7ec0::/29 netname: NL-OSP-20160517 country: GB ... address: Spuistraat 249-II address: 1012VP Amsterdam ... org-type: OTHER ... Really, a Dutch address, claiming a British country, while the prefix clearly show "NL" as prefix. Website at http://www.onlineserviceproviderbv.nl/ is empty, hosted on throw-away Digital Ocean, not on "own" infrastructure (the prefixes they have..). Checking RIS: https://stat.ripe.net/2a07%3A7ec0%3A%3A%2F29#tabId=routing Next to storing junk in RIPEDB, apparently also wasting a lot of routing space, as they all end up in a land of randomness. As that office address is close by RIPE NCC, can somebody maybe visit them and try to educate these folks (if anybody is there), to become quite a bit better netizen.... though, I bet you'll find people more home at: aut-num: AS202792 as-name: CLOUDSRV ... org-name: CloudSRV Limited org-type: LIR address: Unit 24 Oakleigh Court, Church Hill Road address: EN4 8UX address: East Barnet address: UNITED KINGDOM who are announcing a /31 of that space... I guess that after all these years it is still a very good idea to apply: https://www.space.net/~gert/RIPE/ipv6-filters.html Greets, Jeroen

1 0

Launching My Resources in the RIPE Database
by Alex Band 08 Jun '17

08 Jun '17

Dear colleagues, We are pleased to announce My Resources, a new service in the RIPE Database that allows resource holders and members who sponsor resources to get a comprehensive overview of all IPv4, IPv6 and ASN resources they hold or sponsor at a glance. Resource holders can also manage their resources with an easy-to-use inline editor. For more information, please see the news announcement: https://www.ripe.net/manage-ips-and-asns/db/announcements/launching-my-reso… Kind regards, Alex Band Product Manager RIPE NCC

1 0