Fix flaky reputation change test

[AndreiEres]

Fixes #7407

That job requires more than 1ms to complete on cold cache, so reputation change triggers twice. I changed the interval significantly, same as in other test uses reputation change.

Could you please check if this works?

[ggwpez]

Suggested change

	let reputation_interval = Duration::from_millis(100);
	let reputation_interval = Duration::from_secs(1);

Can we bump it even more?

[AndreiEres]

I don't see a point. It will increase the test time by a whole second. 100ms are already too much. Actual execution time, I think, is about 2ms.

[koute]

I don't suppose we could make it so that it doesn't require any timeout at all? By, for example, waiting for something that we know must happen instead of just waiting for a predefined amount of time.

[AndreiEres]

No, it's not just a timeout for tests. Here we determine a value to use in the subsystem.

polkadot/node/network/bitfield-distribution/src/lib.rs

Lines 206 to 213 in 1346281

	loop {
	select! {
	_ = reputation_delay => {
	state.reputation.send(ctx.sender()).await;
	reputation_delay = new_reputation_delay();
	},
	message = ctx.recv().fuse() => {
	let message = match message {

[ggwpez]

Please mention the issue in your MR description so its possible to track it #7407

[ggwpez]

Test fetch_one_collation_at_a_time just failed in the CI. I assume this is unrelated?

[AndreiEres]

bot rebase

[paritytech-processbot]

Rebased

[AndreiEres]

Test fetch_one_collation_at_a_time just failed in the CI. I assume this is unrelated?

I think so, I rebased, and the failed test is gone.

[AndreiEres]

bot merge