Increase in error rate after upgrade to v0.9.36

[BulatSaif]

On 2022-12-20 16:50:21 UTC we updated Polkadot RPC nodes v0.9.33 -> V0.9.36, and after that, we see more errors in logs than usual.

Error:

ERROR tokio-runtime-worker jsonrpsee_server::transport::ws: WS transport error: i/o error: Transport endpoint is not connected (os error 107); terminate connection:

As far as I understand, the logs are due to broken WS connections.

Is the new release breaking the WS or did Polkadot enable more logs?

[bkchr]

@niklasad1 do we really need to log these errors? Probably using debug logging should be enough?

[niklasad1]

This is really a non-clean shutdown or if the socket if the websocket send failed but sure I think we can downgrade these to debug especially this is just noise for polkadot.

[bLd75]

We also noticed another error on HTTP endpoints since upgrade, but in a much bigger proportion than on WS endpoints.
The message seems to be a warning but should we be concerned about requests dropped or is it just node internal warning?
https://substrate.stackexchange.com/questions/7485/parachain-rpc-errors-since-polkadot-0-9-36-client-upgrade

[niklasad1]

@bLd75 That's something completely different and I tried to reply to your stackexchange question.

[BulatSaif]

related: paritytech/jsonrpsee#871

[niklasad1]

@BulatSaif This will be fixed by upgrading jsonrpsee and the PR is up for review now.