2026-08-27

Portal client: streams ride out brief outages

A configured retryAttempts — including the Infinity the quickstart recommends — was
being overridden with 6 on every portal request, so a stream gave up after about
33 seconds regardless of what you set. It now applies, and the default budget rises to
20 retries, roughly five minutes of backoff, so a processor waits out a short portal
outage instead of dying. The flip side: a request that keeps failing now takes about
five minutes to surface its error rather than half a minute.

@subsquid/squid-sdk 0.6.0

Minor changes

• add Solana RPC and fallback data sources: SolanaRpcDataSourceBuilder (solana/rpc) and SolanaFallbackDataSourceBuilder (solana/fallback), mirroring the EVM counterparts

@subsquid/solana-stream 1.2.0

Minor changes

• export the data request types (DataRequest, item request/relations interfaces) from the package barrel

@subsquid/http-client 1.9.0

Minor changes

• expose the configured retryAttempts and retrySchedule as readonly properties, so wrappers can defer to a caller's retry configuration instead of overriding it, and can work out how long a retry budget spans

@subsquid/portal-client 0.7.0

Minor changes

• a configured retryAttempts (including Infinity) is no longer overridden by the client's own default, and that default goes from 6 retries to 20, so portal requests now keep retrying for about five minutes (was about 33 seconds) before failing

@subsquid/rpc-client 4.16.1

Patches

• redact RPC URLs (which routinely carry API keys) from transport error messages, stack traces and connection-failure logs

---

Contributors

@abernatskiy

Full Changelog: 2026-07-30...2026-08-27