spetz commented on code in PR #4615: URL: https://github.com/apache/iggy/pull/4615#discussion_r4338668501
########## core/server/src/http/submit.rs: ########## @@ -167,13 +167,14 @@ pub(in crate::http) async fn submit_committed( /// [`submit_committed`] then forgets the entry so the caller's retry /// re-registers cleanly. /// -/// Both shard-0 request rewrites run here before consensus, mirroring the TCP +/// The [`http_chain`] stages run here before consensus, mirroring the TCP /// dispatch path: the PAT rewrite mints a raw token and replicates only its hash -/// (`CreatePersonalAccessToken`), and the user-password rewrite hashes the new +/// (`CreatePersonalAccessToken`), the user-password rewrite hashes the new /// password and, for `ChangePassword`, strips the current one and verifies it -/// against the stored hash (`CreateUser` / `ChangePassword`). Both are no-ops -/// for every other operation, so plaintext secrets never enter consensus on any -/// write. A third resolution handles `DeleteSegments`, which is not itself a +/// against the stored hash (`CreateUser` / `ChangePassword`), the static bounds +/// admit every options block through the catalog, and the secret options +/// rewrite replaces a stream key with its wrap set (`CreateStream`). So +/// plaintext secrets never enter consensus on any write. A further resolution handles `DeleteSegments`, which is not itself a Review Comment: Fixed. Metadata requests use a random 128-bit id at `core/binary_protocol/src/consensus/request.rs:80`, retained across retries. Partition operations stay at 0. `given_register_retry_when_encoding_again_should_keep_request_identifier` and the Node replay test at `foreign/node/src/client/client.operation.test.ts:34` pin both cases. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
