After #2202 was merged, flooding of recovery attempts is still occurring under a different set of conditions.
The current implementation sets the publisher in the context when downloading content; so that content recovery process is triggered only when the publisher was explicitly set in the download query/request. But it seems this is not enough to fix all of the flooding.
When a made-up hash is requested and the publisher is set, the flooding also occurs.
Something similar occurs when setting a made-up publisher.
Verify that the test scenarios behave as specified, without flooding in any case:
This issue was originally raised in #2205.
After #2202 was merged, flooding of recovery attempts is still occurring under a different set of conditions.
The current implementation sets the publisher in the context when downloading content; so that content recovery process is triggered only when the publisher was explicitly set in the download query/request. But it seems this is not enough to fix all of the flooding.
When a made-up hash is requested and the publisher is set, the flooding also occurs.
Something similar occurs when setting a made-up publisher.
Verify that the test scenarios behave as specified, without flooding in any case:
(valid/invalid means existing/non-existing)