Locate a Solana catch-up delay between shred intake and replay

Two lagging nodes can have different bottlenecks. One may be receiving little new ledger data while another keeps inserting shreds faster than it replays them.

Sample both counters on one backend

Read the highest slot with shreds inserted into blockstore using the request below. Next read getSlot with processed commitment on the same backend. Record the time and both slot numbers. Polling a load-balanced address can silently mix two machines.

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "getMaxShredInsertSlot",
  "params": []
}

Compare movement, then inspect the local node

If the insert slot moves from 50000 to 50400 while the processed slot moves from 49990 to 50010, newer data is arriving and reached slots are diverging. Check replay logs, disk latency and CPU load. A few isolated shreds for a later slot can raise the maximum even when earlier data is incomplete.

If both counters stand still while an independent reference advances, inspect network intake and the node process first. Repeat measurements across several intervals before assigning the bottleneck to one subsystem.

Documentation