A month ago I wrote about the presence of route objects for undelegated IPv4 address space within the RIPE database (strictly I should say RIPE NCC — the body who looks after this database).
The folks at RIPE NCC removed a number of these dubious route objects which had been entered by AS204224.
And they were put straight back again!
This continues to this day — it looks to me as if once the RIPE NCC staff go home for the evening the route objects are resurrected.
So for AS204224 (CJSC Mashzavod-Marketing-Servis) you can (at the moment of writing) find route objects for four /19s and two /21s which have a creation times between 17:53 and 17:55 this evening (2 November). This afternoon (in RIPE NCC working hours) there were no such route objects.
As an aside: as well as AS204224 I see route objects for undelegated space (these are all more recent than my original blog article) from:
AS200439 LLC Stadis, Ekaterinburg, Russia AS204135 LLC Transmir, Blagoveshensk, Russia AS204211 LLC Aspect, Novgorod, Russia
I’d like to give a detailed account of the creation and deletion of the AS204224 route objects, but I don’t believe that there’s a public archive of RIPE database snapshots (you can find the latest snapshot taken at about 03:45 each morning at ftp://ftp.ripe.net/ripe/dbase, but if you don’t download it that day then it’s gone!).
However, I have been collecting copies of the database for the past few days and the creation times for the route objects are:
Thu 2015-10-29 18:03 Fri 2015-10-30 15:01 Sat 2015-10-31 17:54 Sun 2015-11-01 18:31 Mon 2015-11-02 17:53
There are two conclusions to draw from this: perhaps the AS204224 people only come out at night and dutifully delete their route objects when the sun rises before repeating the activity the following night (sounds like one of Grimm’s fairy tales doesn’t it?).
The alternative, less magical explanation, is that the staff at RIPE NCC are playing “whack-a-mole” INSIDE THEIR OWN DATABASE! (and although they work weekends, they go home early on Friday afternoons!)
1 thought on “Ongoing badness in the RIPE database”
Can they not simply set up the database to refuse to accept a route object for an unallocated block (or at the very least flag it for manual approval)?