"Unable to find beacon header" on fresh sync of 1.25.3 and experimental.

[kamilchodola]

Description @matilote on his nodes which were using experimental and 1.25.3 stucked on very early stage of sync with error: Unable to find beacon header at height 19070150. This is unexpected, forcing a new beacon sync. It happens on 4 nodes so very severe for him which i don't have at all and didn't seen any report around that.

Steps to Reproduce To be found - not clear why it works once and second time not.

Actual behavior Node stuck at early stage of sync.

Expected behavior Node should progress just fine.

LOGS execution (1).txt execution (2).txt

[asdacap]

Try turning on struct mode --Sync.StrictMode true. The problem occured about 5000 block before the logs. can't do much here.

[asdacap]

24 Jan 17:06:20 | New pivot block has been set based on ForkChoiceUpdate from CL. Pivot block number: 19077693, hash: 0x866f4f8f1ce9c71d6b3ca2da792953f93ff5182720b81a3ffdd9931dcbf31959
24 Jan 17:06:22 | Unable to find beacon header at height 19077629. This is unexpected, forcing a new beacon sync.

Forward sync trying to download from before the new pivot.

[asdacap]

New sync pivot is set after the new payload block.

[kamilchodola]

@asdacap @flcl42 Is it in progress? I can confirm it still happens on master. Weird thing is also that Beacon Headers are started and afeter a moment it gets reset back to 0.

Easy to reproduce on Holesky Archive, I can see that it restarts beacon header sync a few times before filling it well.

[flcl42]

@kamilchodola I'll take a look in next couple of weeks if no one picks