Closed morelazers closed 2 years ago
Permalink to the substrate fn that is throwing this: https://github.com/paritytech/substrate/blob/20a9bbb1fe47fcd62fcd64b2fa32456b4f434aaf/frame/executive/src/lib.rs#L452-L478
Just to clarify - this is technically not a runtime panic (otherwise all nodes would panic). It seems to be a localised issue with those nodes, although given that it has happened twice now we should of course get to the bottom of it. It also seems to be panicking on an assertion, ie. this is not some unexpected panic, it's a deliberate failsafe in the substrate node.
Just a thought, could this have something to do with runtime upgrades and some nodes not having received it and then trying to connect? or something of that nature?
Given we've not really seen this before, and we've only really done a runtime upgrade now, could be related (or I could be way off)
Saw this on Robbie's Validator on the morning of Dec 17, before we pushed any runtime upgrades.
Got a response from parity:
I think this may be related with this issue https://github.com/paritytech/substrate/issues/9697, in such case running your node with --state-cache-size 0 should avoid it for now
That issue was logged by Alain from Moonbeam and seems to be fairly widespread.
It's due to a caching bug and can be avoided by adding the params as above, at the expense of slightly slower block imports, apparently.
I'll chuck it in the docs and then close the issue. Cheers Dan.
Description
Logging quickly as I'm mostly AFK today. Reported by a Discord member and I've seen something similar on Robbie's Validator.