Good afternoon. Recently I used subgraph to check the TVL of a protocol. When I switched the time window, it suddenly froze—I waited more than ten seconds before it finally loaded. At first, I thought my internet had crashed, but later I realized it was the indexer syncing slowly. The data in Subgraph was lagging behind the actual on-chain state by a dozen or so blocks. What really annoyed me was that when I went through RPC, it rate-limited several free endpoints and returned 503 directly. In situations like this, going back to etherscan is actually faster.

To put it plainly, rate limiting happens because these public RPCs are usually freely scraped by a bunch of bots. The moment there’s activity pressure, they blow up. The indexer also has to replay data block by block—if something is missed in the middle, it has to catch up. That’s just how it is.

My current approach: I’d rather spend a bit of Gas each month to run a lightweight node myself and connect to a full RPC. At least when the data comes in, you can be sure it’s clean, and you don’t have to keep wondering whether something went wrong. It costs a little more, but it saves you from anxiety.

When the funding rate is at extremes, that delay of a few blocks means while others are rushing into the gap, you’re stuck here… sigh. Having the data hang for a moment like that is really annoying at times like this.