anton-vinogradov commented on code in PR #13095:
URL: https://github.com/apache/ignite/pull/13095#discussion_r3589129887
##########
modules/core/src/main/java/org/apache/ignite/internal/managers/communication/GridIoManager.java:
##########
@@ -2011,25 +2042,102 @@ else if (async)
ackC.apply(null);
}
else {
+ MessageMarshaller.marshal(ctx.messageFactory(), ioMsg, ctx,
null);
+
+ sendMarshalled(node, ioMsg, ackC);
+ }
+ }
+ }
+
+ /**
+ * Wraps {@code msg} into a marshalled {@link GridIoMessage} without
sending it. Message marshalling is not
+ * idempotent (see {@code MessageMarshalOnceTest}), so a caller that
retries transmission must prepare the
+ * message once and re-send it via {@link #sendPrepared} on each attempt.
+ */
+ public GridIoMessage prepare(Object topic, Message msg, byte plc, boolean
ordered, long timeout,
Review Comment:
The split is deliberate: the retry loop lives in `GridCacheIoManager`, which
owns the (cache-specific) send-retry policy — `GridIoManager` can't own it.
`prepare` marshals once and returns the `GridIoMessage`; the caller then
retries only `sendPrepared` on each attempt, which doesn't mutate the message.
Folding them into one `GridIoManager` method would pull marshalling back inside
the retry loop — exactly the non-idempotent re-marshal
(`MessageMarshalOnceTest`) that this split fixes. Happy to reconsider the API
shape if you have a specific encapsulation in mind that keeps marshalling
outside the retry.
--
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]
