[{"author": "Chris Morrow", "text": "Jared is taking notes, thanks!
", "time": "2022-03-23T12:05:20Z"}, {"author": "Chris Morrow", "text": "notes: https://notes.ietf.org/notes-ietf-113-grow
", "time": "2022-03-23T12:05:31Z"}, {"author": "Chris Morrow", "text": "warren is 'jabber scribing' here as well :)
", "time": "2022-03-23T12:05:41Z"}, {"author": "Warren Kumari", "text": "[ Morning AM DJ Tone ]
Good Morning, Afternoon or Evening all.
I will be your friendly Jabber scribe for today! It's a wonderful day here in Vienna, and I just *know* we are going to have a great meeting!
[ /Morning AM DJ Tone ]
", "time": "2022-03-23T12:08:56Z"}, {"author": "Warren Kumari", "text": "Square brackets, because MeetEcho eats stuff in angle brackets, like <this> ...
", "time": "2022-03-23T12:09:45Z"}, {"author": "George Michaelson", "text": "12 minutes FLEW PAST. Time has no meaning when you're having fun
", "time": "2022-03-23T12:10:39Z"}, {"author": "Jeffrey Haas", "text": "You mean like <this>?
", "time": "2022-03-23T12:10:49Z"}, {"author": "George Michaelson", "text": "no like >this<
", "time": "2022-03-23T12:11:05Z"}, {"author": "Chris Morrow", "text": "clearly you two have a different angle on this problem than warren does.
", "time": "2022-03-23T12:12:01Z"}, {"author": "George Michaelson", "text": "its the sanity claws
", "time": "2022-03-23T12:14:08Z"}, {"author": "Chris Morrow", "text": "rawr
", "time": "2022-03-23T12:19:24Z"}, {"author": "Jeffrey Haas", "text": "https://datatracker.ietf.org/meeting/97/materials/slides-97-idr-code-point-management-02
", "time": "2022-03-23T12:20:28Z"}, {"author": "Beno\u00eet Claise", "text": "I won't go to the mic. to stress (again) what Paolo mentioned: We did not see Ben's concern in IPFIX IEs history, hopefully.
", "time": "2022-03-23T12:22:58Z"}, {"author": "Warren Kumari", "text": "Did y'all *really* do html entities just to make me look <stupid>?
", "time": "2022-03-23T12:27:30Z"}, {"author": "Bill Fenner", "text": "Make you look ><?
", "time": "2022-03-23T12:27:48Z"}, {"author": "Warren Kumari", "text": "Erk. It deleted those too! <
", "time": "2022-03-23T12:27:48Z"}, {"author": "Randy Bush", "text": "fwiw, you can let the speaker control the slides. speaker clicks second icon from left.
", "time": "2022-03-23T12:27:49Z"}, {"author": "Jeffrey Haas", "text": "It's been almost two decades since I touched NRTM for irrd. Didn't it do a detached md5 signature?
", "time": "2022-03-23T12:29:17Z"}, {"author": "George Michaelson", "text": "RRDP in turn was inspired by ... NRTMv3!
", "time": "2022-03-23T12:30:29Z"}, {"author": "George Michaelson", "text": "strange loop
", "time": "2022-03-23T12:30:36Z"}, {"author": "Jeffrey Haas", "text": "\"Virtuous circle\"
", "time": "2022-03-23T12:30:47Z"}, {"author": "George Michaelson", "text": "@Warren its you. I typed \"<\" not \"<\" or some unicode/UTF-8 wierdness.
", "time": "2022-03-23T12:33:13Z"}, {"author": "George Michaelson", "text": "oh god. & l t translated on me
", "time": "2022-03-23T12:33:30Z"}, {"author": "George Michaelson", "text": "they're ordered? so ADD 1.1.1.0/24 MOD 1.1.1.0/24 DEL 1.1.1.0/24 is the partially ordered set equivalent to <NUL> ?
", "time": "2022-03-23T12:37:09Z"}, {"author": "George Michaelson", "text": "(ie, is it deterministic, because ordered)
", "time": "2022-03-23T12:37:39Z"}, {"author": "Job Snijders", "text": "@ggm i will relay this question after the talk
", "time": "2022-03-23T12:37:58Z"}, {"author": "George Michaelson", "text": "Nah, the spec is explicit:
", "time": "2022-03-23T12:40:10Z"}, {"author": "George Michaelson", "text": "* The Delta File MUST include all changes that happened during the
time frame, in the order in which they occurred.
", "time": "2022-03-23T12:40:11Z"}, {"author": "George Michaelson", "text": "they're ordered, its an array.
", "time": "2022-03-23T12:40:20Z"}, {"author": "Job Snijders", "text": "ok
", "time": "2022-03-23T12:40:23Z"}, {"author": "Jeffrey Haas", "text": "my comments are for after presentation is complete.
", "time": "2022-03-23T12:41:08Z"}, {"author": "George Michaelson", "text": "* The client MUST process all changes in the Delta Files in order:
lowest Delta File version number first, and in the order of the
changes list in the Delta File.
", "time": "2022-03-23T12:41:21Z"}, {"author": "Ties de Kock", "text": "In RRDP some repos do see spurious fallback to snapshot & multiplication of traffic because of that.
", "time": "2022-03-23T12:42:38Z"}, {"author": "Ties de Kock", "text": "In the end it will depend on how often the clients check in..
", "time": "2022-03-23T12:43:02Z"}, {"author": "Jeffrey Haas", "text": "Ties, so you're saying the re-fetch is due to not enough delta state available?
", "time": "2022-03-23T12:43:24Z"}, {"author": "Ties de Kock", "text": "re-fetch is due to not enough diffs available _or_ clients not wanting to download that many diffs.
", "time": "2022-03-23T12:48:04Z"}, {"author": "Ties de Kock", "text": "Limiting it to 1/minute will help a lot there (for RRDP there is repos with > 1/minute)
", "time": "2022-03-23T12:49:02Z"}, {"author": "Ties de Kock", "text": "*are repos
", "time": "2022-03-23T12:49:14Z"}, {"author": "Job Snijders", "text": "@Ben - new draft to establish procedures to add/remove/modify sources from the to-be-established registry is probably easier to grok, and combining the two efforts. The protocol draft can reference your to-be-written draft :-)
", "time": "2022-03-23T12:49:17Z"}, {"author": "Chris Morrow", "text": "250k files over http is ... about an hour? :) depending on your bw and parallelism... :)
", "time": "2022-03-23T12:49:32Z"}, {"author": "Chris Morrow", "text": "note that 1yr of data is 512k-ish files, not 250k.
", "time": "2022-03-23T12:52:21Z"}, {"author": "Chris Morrow", "text": "I suppose, my point here is: \"the number or files required for a client to catchup is not important\"
", "time": "2022-03-23T12:53:38Z"}, {"author": "Jeffrey Haas", "text": "The number of files on your filesystem of choice is usually more irritating than the size in many current circumstances.
", "time": "2022-03-23T12:54:28Z"}, {"author": "Jeffrey Haas", "text": "Sometimes fopen/fclose takes longer than the fread to pull in the whole file.
", "time": "2022-03-23T12:55:00Z"}, {"author": "Ties de Kock", "text": "deltas may not even need to touch the disk?
", "time": "2022-03-23T12:56:31Z"}, {"author": "Jeffrey Haas", "text": "My knowledge of current IRR operational practices is very out of date, clearly. :-)
", "time": "2022-03-23T12:57:59Z"}, {"author": "Chris Morrow", "text": "I think Jeff \"yes, of course if you choose a poor FS or poor directory structure... ouch\"
and yea, you COULD store them in RAM, you COULD propose that 'if you are out of date more than X <time> then just get the current snapshot and any deltas since then\"
", "time": "2022-03-23T12:58:48Z"}, {"author": "Jeffrey Haas", "text": "I would speculate you'd want to keep a local cache of state so that if you restart for some reason that a querier of your database doesn't get answers from the \"past\" if it is working from snapshot without deltas.
", "time": "2022-03-23T12:58:49Z"}, {"author": "Chris Morrow", "text": "'from the \"past\" <time>
", "time": "2022-03-23T12:59:12Z"}, {"author": "Chris Morrow", "text": "it's also possible that the delta/snapshot aren't even actual files! (but generated on demand at the https endpoint)
", "time": "2022-03-23T13:00:18Z"}, {"author": "Jeffrey Haas", "text": "Feel free to s/files/state/
", "time": "2022-03-23T13:00:42Z"}, {"author": "Jeffrey Haas", "text": "Usual integration is via leafref or similar
", "time": "2022-03-23T13:02:43Z"}]