Probability of BABE stalling?

[c-geek]

I've found in the docs + checked code that BABE has only 2 fallbacks in case no primary blocks: SecondaryPlain + SecondaryVRF.

While I'm happy to have discovered this extra SecondaryVRF I was not aware of, yet two secondary authors seems a bit low to me because it's quite easy to have a faulty node in a epoch, even more with sleeping attackers.

Is there another fallback I'm not aware of? Don't you think there is a security issue here? Am I missing something?

[bkchr]

You only need one block per epoch/session to prevent BABE from stalling.

[c-geek]

I don't get it. The problem is the same to me whether the block is the first of an epoch, the second or the last.

Don't we have a stalling if no next block is proposed?

[bkchr]

What you mean? When no block is produced anymore? Why should there be no block be produced anymore?

[c-geek]

Because of VRF leading to no primary block (no validator matching), same for secondary VRF, and same for secondary plain because, as I understand it, only one validator is selected per slot and he might be faulty.

[bkchr]

There is no problem if one slot is skipped.

[c-geek]

That was my blind spot. Thanks.