State sync restart: Potential long-range attack

[turuslan]

Description
Parachain state sync completes successfully,
but then it repeatedy restarts after "Potential long-range attack" message.

Steps to Reproduce

1. Download release and unpack astar-collator
2. Run validator, wait until some blocks are finalized

./astar-collator \
  --collator \
  --no-hardware-benchmarks --no-mdns --no-telemetry \
  --state-pruning archive --blocks-pruning archive \
  --base-path data1-1 --chain dev \
  --node-key f8dfdb0f1103d9fb2905204ac32529d5f148761c4321b2865b0a40e15be75f57 \
  --listen-addr /ip4/127.0.0.1/tcp/10000 --rpc-port 10002 --prometheus-port 10003 \
  --force-authoring --alice \
  -lverbose \
  -- \
  --state-pruning archive --blocks-pruning archive \
  --base-path data1-2 --chain dev \
  --node-key 96c891b8726cb18c781aefc082dbafcb827e16c8f18f22d461e83eabd618e780 \
  --listen-addr /ip4/127.0.0.1/tcp/10010 --rpc-port 10012 --prometheus-port 10013 \
  --force-authoring --alice \
  -lverbose \
  2>&1 | tee -a data1.log

3. Run sync node

RUST_LOG="verbose,sync=trace" ./astar-collator \
  --collator \
  --bootnodes /ip4/127.0.0.1/tcp/10000/p2p/12D3KooWT3CsfuervgduMG6RdjVq66z9JnN1GcsRBCWZPxPSBZSV \
  --no-hardware-benchmarks --no-mdns --no-telemetry \
  --base-path data2-1 --chain dev \
  --node-key 619d5e68139f714ee8e7892ce5afd8fbe7a4172a675fea5c5a06fb94fe3d797d \
  --listen-addr /ip4/127.0.0.1/tcp/10020 --rpc-port 10022 --prometheus-port 10023 \
  --sync fast \
  -- \
  --bootnodes /ip4/127.0.0.1/tcp/10010/p2p/12D3KooWNgqk3v24c2Qvi3ZrkJVLayPeUmGbqnNuyt4hHmtxMghk \
  --base-path data2-2 --chain dev \
  --node-key 8d0c5f498a763eaa8c04861cac06289784140b4bbfa814fef898f1f4095de4a3 \
  --listen-addr /ip4/127.0.0.1/tcp/10030 --rpc-port 10032 --prometheus-port 10033 \
  2>&1 | tee -a data2.log

Expected vs. Actual Behavior
Expected: state sync doesn't restart after completion.
Actual: state sync restarts.

Environment

• Operating system: macos, ubuntu
• Project version/tag: v5.46.0
• Rust version (run rustup show): -

Logs, Errors or Screenshots

2025-09-11 11:45:36.027 DEBUG tokio-runtime-worker sync: Starting state sync for #19 (0x434e…ea1c)
2025-09-11 11:45:36.050 DEBUG tokio-runtime-worker sync: State download is complete. Import is queued
2025-09-11 11:45:36.058  WARN tokio-runtime-worker sc_client_db: Potential safety failure: reverting finalized block (19, 0x434e9dca592d07ac335eceacca227af046275cc879e29ab58d950886e80aea1c)
2025-09-11 11:45:36.059  WARN tokio-runtime-worker sc_service::client::client: Block import error: Potential long-range attack: block not in finalized chain.
2025-09-11 11:45:37.080 DEBUG tokio-runtime-worker sync: Starting state sync for #20 (0xb8ca…1662)
2025-09-11 11:45:37.106 DEBUG tokio-runtime-worker sync: State download is complete. Import is queued
2025-09-11 11:45:37.116  WARN tokio-runtime-worker sc_client_db: Potential safety failure: reverting finalized block (20, 0xb8ca82d2368b1298da82d436052151e53f7a1f5c655341339f998541874a1662)
2025-09-11 11:45:37.116  WARN tokio-runtime-worker sc_service::client::client: Block import error: Potential long-range attack: block not in finalized chain.

Additional Information
This behaviour was observed on both dev and astar (*) chain.
(*) Currenlty state sync may crash with OOM, but we were testing our branch with OOM fix

[Dinonard]

Hi @turuslan, can you re-try with clean DB directory?

[turuslan]

Hi @turuslan, can you re-try with clean DB directory?

I remove DB directory before running two nodes from #1540 (comment)

[Dinonard]

I see, thanks.

Please consult our official docs on how to run the node.

The arguments you're using are incorrect.
E.g. you're specifying things like --alice and --chain dev which cannot be used for production node.