Exexes crashes due to failing canonical chain check when syncing host node from scratch

[Evalir]

Describe the bug

When an exex which has “persistence” (can, on boot, set the exexhead to a particular block) is used on a host node that is syncing from scratch, the exex will fail the canonical chain check due to the provider returning 0 when syncing. This then leads to the exex crashing, as it can't find the committed notification for the block hash requested from the WAL.

Steps to reproduce

This issue is not straightforward to reproduce quickly, but the steps are:

• Start with a reth node that needs to sync fully from scratch.
• Make an exex that sets its head to an arbitrary block.
• Boot up the reth node with the exex enabled, let it sync.
• After a while, the exex will boot up, set the head, and perform the canonical chain check. It will fail, and crash after.

Node logs

Platform(s)

Linux (x86), Linux (ARM)

Container Type

Kubernetes

What version/commit are you on?

Reth 1.1.1.

What database version are you on?

2

Which chain / network are you on?

Holesky

What type of node are you running?

Archive (default)

What prune config do you use, if any?

N/A

If you've built Reth from source, provide the full command you used

No response

Code of Conduct

• I agree to follow the Code of Conduct

[emhane]

could this have to do with #13940 @klkvr @shekhirin ?

[mattsse]

@Evalir just to clarify, this happens if the exexhead is ahead of the host's head at launch.

e.g. resync host: 0, exex_with_head(10)

[Evalir]

yep @mattsse that is correct!