I was just about to check some unlock data, and then the subgraph got stuck again... It took three minutes to spit out the result, and I almost thought my node had crashed.

The pressure on these indexers is really huge now. With the wave of traffic from staking unlocks, the free tier is basically a lottery experience. Run my own local node? The electricity bill costs more than the query fees, so forget it.

Simply put, "getting stuck" is most likely RPC rate limiting or subgraph sync lag. The transaction is actually confirmed on-chain, but the data layer is queued up. You'll get used to it. When in a hurry, prepare two more endpoints for polling—it's a hassle but more stable.

Oh, by the way, someone asked me which RPC provider is more stable—

I haven't found a perfect one either. That's it for now.