Improve snap sync

[nazar-pc]

This simple change ensures that the first imported block is finalized and allows to finalize all other blocks afterwards successfully.

Code contributor checklist:

• I have read, understood and followed contributing guide

[shamil-gadelshin]

When I set pub(crate) const FINALIZATION_DEPTH_IN_SEGMENTS: SegmentIndex = SegmentIndex::new(0); and manually set snap_sync target block to some value (e.g.:1000000) should I get blocks of the first couple of segments finalized?

[nazar-pc]

You will not get any finalization during sync because the whole archiving process is skipped for already archived blocks and finalization is only called from there.

To re-enable archiving of already archived history you'll have to do something like this:

diff --git a/crates/sc-consensus-subspace/src/archiver.rs b/crates/sc-consensus-subspace/src/archiver.rs
--- a/crates/sc-consensus-subspace/src/archiver.rs	(revision 51c4da575186f81c98b550ad92bcc6a7ba8fffe4)
+++ b/crates/sc-consensus-subspace/src/archiver.rs	(date 1723552439720)
@@ -845,7 +845,7 @@
                 "Checking if block needs to be skipped"
             );
             if best_archived_block_number >= block_number_to_archive
-                || last_archived_block_number > block_number_to_archive
+            // || last_archived_block_number > block_number_to_archive
             {
                 // This block was already archived, skip
                 debug!(
@@ -958,6 +958,13 @@
         block_number_to_archive, block_hash_to_archive
     );
 

But then you don't need to modify sync target or archiving depth at all if you want to just verify finalization, something like this is sufficient:

diff --git a/crates/sc-consensus-subspace/src/archiver.rs b/crates/sc-consensus-subspace/src/archiver.rs
--- a/crates/sc-consensus-subspace/src/archiver.rs	(revision 51c4da575186f81c98b550ad92bcc6a7ba8fffe4)
+++ b/crates/sc-consensus-subspace/src/archiver.rs	(date 1723552439720)
@@ -845,7 +845,7 @@
                 "Checking if block needs to be skipped"
             );
             if best_archived_block_number >= block_number_to_archive
-                || last_archived_block_number > block_number_to_archive
+            // || last_archived_block_number > block_number_to_archive
             {
                 // This block was already archived, skip
                 debug!(
@@ -958,6 +958,13 @@
         block_number_to_archive, block_hash_to_archive
     );
 
+    finalize_block(
+        client,
+        None,
+        *block.block.header().parent_hash(),
+        *block.block.header().number() - One::one(),
+    );
+
     if parent_block_hash != best_archived_block_hash {
         let error = format!(
             "Attempt to switch to a different fork beyond archiving depth, \

[shamil-gadelshin]

Works well!

[nazar-pc]

I'll submit a patch that will warn and suggest users to wipe the database separately (if we can't transparently fix it before then)