urbit / urbit

An operating function
https://urbit.org
MIT License
3.43k stars 359 forks source link

L2 planet boot: %verify-sig-and-nonce fails for all known L2 ships #5501

Closed jamesacklin closed 2 years ago

jamesacklin commented 2 years ago

Long printf from new L2 planet booted via command line, complaining about 12-sig-failed. Running the 1.8 binary and the latest version of Port. This blocks the L2 planet from booting with Port, but lets me in via command-line.

To reproduce: Steps to reproduce the behavior:

  1. Claim a new L2 ship and download its passport.
  2. Extract the passport and either (a) boot it via command line or (b) boot it via Port.
  3. Wait a minute, then observe https://pastebin.com/raw/ttHSRaKF

Ethereum network:

Desktop (please complete the following information):

Login method (please complete the following information):

drbeefsupreme commented 2 years ago

Trying to reproduce:

https://zerobin.net/?be2ab45301d7e663#0sfnNj4EqsVbZthXmqpnLhnRnKDjXiv9wQ1wIXjM87E=

joemfb commented 2 years ago

I saw similar printfs booting a comet.

yosoyubik commented 2 years ago

I've seen the same on moons that didn't have the full azimuth state, but if this happens after booting (at least it started processing the logs.. loading snapshot with 291.850 events). It'd be interesting to know what's in the azimuth state of those ships:

> =n -build-file %/lib/naive/hoon
> %~  wyt  by  points:.^(^state:n %gx /=azimuth=/nas/noun)
> .^((unit point:n) %gx /=azimuth=/point/~rilfun-lidlen/noun)
yosoyubik commented 2 years ago

Confirmed on a new comet that it only has 1.589 points in state. I saw this behavior and we addressed it here but looking at what the comet (and presumably the L2 planets) has in /app/azimuth shows version %1 on the state, which would point to an old pill? I believe that this would get fix as soon as they get the new OTA, but the pill should still be updated I think

PS: for the L2 planes, since their sponsor is down, I'm guessing that they won't get the OTA unless they switch to some other star

PPS: it looked like ~binzod was too busy (or maybe something else was amiss) so the comet was not getting the OTA. I then switched to ~marzod and it came down fine—/app/azimuth reprocessed all the logs and got the state up to date

jamesacklin commented 2 years ago

PS: for the L2 planes, since their sponsor is down, I'm guessing that they won't get the OTA unless they switch to some other star

Does this imply that every L2 point booted without a running sponsor will squawk in everyone's logs until they OTA? As evidenced by @joemfb's repro

yosoyubik commented 2 years ago

everyone's logs

it should only be for newly booted ships that use the old pill i think, or ships that are behind, unable to get the OTA, or stuck there for some reason (i have moons that get those logs because their azimuth state has only 1.589 points)

yosoyubik commented 2 years ago

Also, doing :azimuth %rerun should reprocess all the logs and bring the azimuth state up to date