Testnet Mempool Depth

Possible causes:

• Stuck mempool — a sender re-sends the same nonce with insufficient gas bump— Fix: Identify the spam sender via raw mempool sampling; if it's an internal service (faucet, indexer), fix its nonce management. Last resort: `systemctl restart intgd` on the gateway to flush local mempool (peers re-gossip — also restart serially if needed).
• External spam attack on the public RPC— Fix: Add a Caddy-level rate-limit on /evm POST and /rpc/broadcast_tx_*; consider blocking the sender address at the application layer.
• Block-level rejection — txs admit to mempool but proposer can't include them— Fix: Check intgd logs for `failed to deliver` or ABCI errors; often a gas-cap or consensus-param mismatch after an upgrade.