Do not require blobs from checkpoint servers from Fulu epochs. #8413
Conversation
jimmygchen
force-pushed
the
fix-checkpoint-sync
branch
from
f78e400 to
0341fd6
Compare
jimmygchen
added
ready-for-review
jimmygchen
force-pushed
the
fix-checkpoint-sync
branch
from
5bfa044 to
0341fd6
Compare
|
Some required checks have failed. Could you please take a look @jimmygchen? 🙏 |
mergify
bot
added
waiting-on-author
chong-he
left a comment
chong-he
left a comment
There was a problem hiding this comment.
Looks good.
From my understanding the ClientGenesis::WeakSubjSszBytes variant is when then data (block, state and blob) is provided, i.e., during manual checkpoint sync. I will put this small update in the PR about updating Lighthouse book
michaelsproul
left a comment
michaelsproul
left a comment
There was a problem hiding this comment.
Tested manually on Hoodi and it works
I have some hesitation about introducing a gap in the data column DB, but I think maybe it can be worked around using manual custody backfill sync anyway (the POST /lighthouse/custody/backfill API). It's just quite inefficient -- we have to wait for block backfill to complete, and then start custody backfill, just to download one slot's worth of missing data columns. I think this will also re-download all the existing columns too, but not sure.
I think the current column backfill logic may need a bit of work to support this, it currently just backfill all the way to the start of DA window |
jimmygchen
added
ready-for-merge
This hot fix release includes the following fixes: * #8388 * #8406 * #8391 * #8413 Co-Authored-By: Jimmy Chen <[email protected]>
3 participants
Issue Addressed
Addressed this comment here: #6837 (comment)
Lighthouse can only checkpoint sync from a server that can serve blob sidecars, which means they need to be at least custdoying 50% of columns (semi-supernodes)
This PR lifts this constraint, as blob sidecar endpoint is getting deprecated in Fulu, and we plan to fetch the checkpoint data columns from peers (#6837)