Ton-20 index

Public API: api.tonano.io

Index process

1. Index all messages from TON from all shards over tonpy.blockscanner.blockscanner.BlockScanner over liteservers
2. Save all ton-20 transactions to DB
   1. All ton-20 logic related is in ton20logic.py
   2. State serialization/deserialization ton20state.py
3. Load all transactions from latest state (ton20state.tlb) and index them
   1. Wallets & ticks & transactions saved to DB separately from state
4. State commited to DB each COMMIT_STATE_EACH txs

---

Warning: if your reboot index it will roll-back to latest state in DB including wallets&ticks and index all TXs from latest state again

Dumps

1. LiteDump: latest actual state
   1. This is just BOC of latest state if costs several mb and it's fast to download
   2. To do this just run manage.py get_latest_state https://api.tonano.io/api/v1/latest-state-boc/
2. PostgresDump: full transactions history dump
   1. It's costs several gb (~160) and it's slow to download, but you will have full history of transactions
   2. Download postgres dump from TBD
   3. [OPTIONAL] You can reindex all ton20 index based on RAW data from liteserver:
      1. Delete all serialized states and transactions status (indexes also if you want faster insert)
      2. Run manage.py start_ton20 it'll load all transactions and index data from start, it'll take ~8h depends on your server

---

Server requirements

With liteserver on same node

• 16 cores CPU
• 128 GB RAM
• 1TB NVME
• 1 Gbit/s network connectivity

Without liteserver on same node

• 64gb RAM
• 16 CPU
• 500gb NVME

Liteserver

Liteserver stability must be excellent, consider host own node or use @liteserver_bot instances

Add your liteserver config to .env (check out .env_example)

Hints if you want to index from start

Since there's a lot of TXs to load (~90m) it needed to be setup correctly

1. Remove all indexes on tables for faster push
2. Change chunk_size in start_index.py. >chunk = faster process, higher RAM usage.