Open jun0tpyrc opened 3 days ago
Thank you for the issue @jun0tpyrc
I just tested Sidecar (v19.3.1) with some public RPC endpoints connected to Kusama and I either get almost instant results from /blocks/head
or with some delay. However, I cannot reproduce the timeout you are experiencing.
On Sidecar's side, there are no significant changes in the code of the last releases that could result in delays in endpoints like /blocks/head
.
Have you changed something in your nodes ? I am checking now to see if there is something related to the Kusama network in general that could justify a delay.
We are experiencing the same, sidecar using more than 4gbs of ram. Try with blocks around 25875378.
I'm currently looking into this issue. I have seen some degradation for certain blocks in Kusama due to the spammening reported here. Specifically certain blocks in the range: 25875509-25875808
Those blocks have about 4/5k extrinsics per block (compared to 40/50 average) every 6s. So the endpoints in sidecar /blocks/:number or /blocks/head (when those blocks were the head) would struggle to get all extrinsics in order to respond in a timely manner. We are currently looking into a solution to resolve the extrinsics decoding overhead once all the blocks will be averaging that level of traffic. It would be quite helpful to understand your use cases for sidecar specifically to /blocks/head and /blocks/:number so we can work on a more performant solution, thanks!!
Description Performance issue with kusama private nodes running behind sidecar hang & stuck => rpc timeout etc , also maybe the cause of explorer slowdown https://kusama.subscan.io/
Steps to Reproduce
run sidecar (v19.2.0/v19.3.1) on your rpc nodes ,tested on polkadot 1.16.2-dba2dd59101, run some calls like /blocks/head etc